Department of Environmental Engineering Centre for Water Research, The University of Western Australia EVALUATION OF REMEDIATION OPTIONS FOR LAKE MARACAIBO, VENEZUELA Robert Medley This thesis is submitted in partial fufillment for the degree of Bachelor of Engineering (Hons.) from the Department of Environmental Engineering, at the University of Western Australia. November 2001 Robert Medley 6 Glover Street Dianella, WA 6156 5th November 2001 The Dean Faculty of Engineering The University of Western Australia Nedlands WA 6009 Dear Sir, This thesis entitled "Evaluation of Remediation Options for Lake Maracaibo, Venezuela" is submitted as partial fulfillment of the requirements for the combined degree of Bachelor of Engineering (Environmental)/ Bachelor of Commerce (Gen. Mngt. & Hum.Res. Mngt.) with honours. Yours Sincerely, Robert Medley Abstract Historically Lake Maracaibo has been a focal point for the people of Venezuela, providing aesthetic, economic and recreational benefits to the residents. Increasing salinisation and anthropogenic nutrient inputs has culminated in Lake Maracaibo showing signs of accelerated eutrophication, indicated by intense phytoplankton blooms. Salinity, along with the methods required to reduce it, has been a central issue in many past discussions and reports on the future of Lake Maracaibo. Presently there is discussion to close the navigation channel and re-locate the port facilities to the adjacent Gulf of Venezuela. An in-depth review and interpretation of past data and studies was undertaken, in combination with hydrodynamic-ecological modelling in order to advise on the effectiveness of environmental remediation strategies aimed primarily at reducing salinity and nutrient inputs. The abundance and composition of the phytoplankton standing crop is one of the most important indicators of water quality and eutrophication in lakes and reservoirs. Thus, a critical issue that must be considered when assessing the merit of any proposed remediation strategy is the impact on primary production and algal species composition. This study was the first to model both dominant algal species in Lake Maracaibo and accurately represent the observed phytoplankton dynamics, allowing for the conclusion that closing the navigation channel would lead to a slight improvement in water quality. However, the Lake would still suffer from intense blooms of toxic blue-green algae; the main problem being the nutrients coming from the Catatumbo River which has its catchment in Columbia. Acknowledgements Acknowledgements I would like to acknowledge the following people for their contribution to this thesis – whether it has been through inspiration, support, encouragement, or friendship. My Supervisor Prof. Jörg Imberger for creating this project for me, motivating me to work to my full potential, imparting his knowledge of all things hydrodynamic, and for the stimulating conversations throughout the year Dr. David Hamilton for taking an interest in my project, finding time to see me, even though he was busy with students of his own, and for his knowledge on all things ecological, willingly shared. My fellow Maracaibo man, Bernard Laval, who has helped me throughout the year, as we both grappled with the beast that is Lake Maracaibo. Thanks Bernard, your help has been invaluable! Peter Yeates for his help with modelling and providing me with background on the project. David Horn for taking an interest in my work and providing me with documentation on previous studies. Andre_ Gomez-Giraldo for his help with coding the new mixing algorithm. The final year class – special thanks to Tom Ewing, Alia Abid and Rachel Murphy. You guys have made a tough year an enjoyable one! To my mates – thanks for taking me out on the weekends and keeping me sane. Tash, thanks for everything, your support throughout the year has been enormous. Mum and Dad – I dedicate this thesis to you…without your help, assistance, guidance and support throughout the years, none of this would have been possible! Cheers, Rob. Table of Contents 1 MOTIVATION ..........................................................................................................................................................1 2 COMPONENTS OF THE LAKE MARACAIBO SYSTEM ..............................................................................4 3 4 2.1 LAKE MARACAIBO ..................................................................................................................................................4 2.2 STRAIT OF MARACAIBO ..........................................................................................................................................5 2.3 TABLAZO BAY .........................................................................................................................................................5 2.4 THE GULF OF VENEZUELA ......................................................................................................................................6 2.5 THE NAVIGATION CHANNEL...................................................................................................................................6 HYDROLOGY...........................................................................................................................................................7 3.1 THE LAKE MARACAIBO CATCHMENT ....................................................................................................................7 3.2 PRECIPITATION ........................................................................................................................................................8 3.3 HYDROLOGICAL BALANCE .....................................................................................................................................8 HYDRODYNAMICS OF LAKE MARACAIBO................................................................................................10 4.1 4.1.1 Tides ............................................................................................................................................................10 4.1.2 Winds...........................................................................................................................................................11 4.1.3 Currents ......................................................................................................................................................11 4.2 5 6 MIXING PROCESSES...............................................................................................................................................12 4.2.1 Internal Dynamics ......................................................................................................................................13 4.2.2 The influence of the earth’s rotation..........................................................................................................14 PREVIOUS STUDIES ............................................................................................................................................15 5.1 THE WOODS HOLE AND MIT STUDIES ...............................................................................................................15 5.2 THE BATTELLE STUDY ..........................................................................................................................................17 5.3 THE PARRA PARDI STUDIES ..................................................................................................................................18 5.4 THE NEDECO STUDY ..........................................................................................................................................18 5.5 THE ICLAM STUDIES ...........................................................................................................................................20 5.6 THE INTEVEP STUDIES .......................................................................................................................................20 5.7 STUDIES AT THE UNIVERSITY OF ZULIA ...............................................................................................................21 5.8 OTHER STUDIES OF INTEREST ...............................................................................................................................21 5.9 THE BECHTEL INTEGRATED ENVIRONMENTAL REMEDIATION STUDY ...............................................................22 HISTORICAL SALINITY TREND......................................................................................................................24 6.1 7 HYDRODYNAMIC FORCINGS .................................................................................................................................10 BEHAVIOUR OF THE UNDERFLOW.........................................................................................................................25 PHYTOPLANKTON COMPOSITION & SUCCESSION ...............................................................................27 7.1 HISTORICAL PHYTOPLANKTON COMPOSITION & SUCCESSION ...........................................................................28 iv Table of Contents 8 9 PREDICTIVE WATER QUALITY MODELLING...........................................................................................32 8.1 INPUT-OUTPUT MODELS .......................................................................................................................................33 8.2 PROCESS BASED MODELLING/ECOLOGICAL MODELS .........................................................................................35 MODEL SELECTION............................................................................................................................................36 9.1 DYRESM-CAEDYM...........................................................................................................................................37 9.2 CAEDYM .............................................................................................................................................................37 9.2.1 REPRESENTATION OF PHYTOPLANKTON GROWTH WITHIN CAEDYM .......................................39 Light Limitation.................................................................................................................................................................. 40 Nitrogen Limitation ............................................................................................................................................................ 41 9.3 10 DYRESM: REVIEWING THE 1-D ASSUMPTION.........................................................................................42 APPLYING DYRESM-CAEDYM TO LAKE MARACAIBO .........................................................................48 10.1 DYRESM SETUP ...........................................................................................................................................49 10.1.1 PHYSICAL AND MORPHOMETRIC DATA.............................................................................................49 10.1.2 INITIAL PROFILE .....................................................................................................................................50 10.2 INPUT DATA .................................................................................................................................................50 10.2.1 Meteorological Forcings............................................................................................................................51 Data..................................................................................................................................................................................... 51 Meteorological Stations ..................................................................................................................................................... 51 Location .............................................................................................................................................................................. 51 10.2.2 Inflows.........................................................................................................................................................53 10.2.3 Water Quality Parameters .........................................................................................................................56 10.3 ASSUMPTIONS .............................................................................................................................................58 10.3.1 Forcing Data...............................................................................................................................................58 10.3.2 Combining Surface Exchange with River Flow.........................................................................................58 10.3.3 Outflow Condition ......................................................................................................................................59 10.4 CONFIGURING DYRESM TO MODEL LAKE MARACAIBO ...............................................................................60 10.4.1 Benthic Boundary Layer (BBL)..................................................................................................................61 10.4.2 Internal Mixing ...........................................................................................................................................62 10.4.3 Developing a New Internal Mixing Algorithm ..........................................................................................64 Background......................................................................................................................................................................... 64 The Algorithm .................................................................................................................................................................... 65 11 CALIBRATION.......................................................................................................................................................69 11.1 FIELD DATA..................................................................................................................................................69 11.2 CALIBRATION METHOD............................................................................................................................70 11.2.1 Phytoplankton Composition & Succession for the Calibration period ....................................................82 v Table of Contents 12 13 PREDICTIVE SIMULATIONS OF MANAGEMENT OPTIONS..................................................................86 12.1 ASSUMPTIONS AND DATA USED IN PREDICTIVE SIMULATIONS .......................................................................86 12.2 TOOLS USED IN ASSESSING THE MANAGEMENT OPTIONS ................................................................................87 12.3 REFERENCE CASE (CURRENT CONDITION WITH REPEATED FORCING) ............................................................88 12.4 EXISTING BATHYMETRY: REDUCED NUTRIENTS (10%).................................................................................95 12.5 PRE-DREDGING BATHYMETRY: REDUCED NUTRIENTS (10%) .......................................................................97 DISCUSSION.........................................................................................................................................................102 13.1 EXISTING CONDITIONS ...................................................................................................................................104 Vertical Transport............................................................................................................................................................. 104 Oxygen Dynamics ............................................................................................................................................................ 104 Nutrient Dynamics............................................................................................................................................................ 105 Phytoplankton Composition and Succession .................................................................................................................. 105 13.2 EFFECT OF A 10% REDUCTION IN NUTRIENT LOADING .................................................................................106 13.3 EFFECT OF CLOSING THE SHIPPING CHANNEL ................................................................................................106 Phytoplankton Composition and Abundance .................................................................................................................. 108 14 CONCLUSIONS....................................................................................................................................................112 15 RECOMMENDATIONS FOR FUTURE WORK............................................................................................113 16 REFERENCES ......................................................................................................................................................114 17 APPENDICES........................................................................................................................................................123 vi Motivation 1 Motivation Historically, Lake Maracaibo has been a focal point for the people of Venezuela, providing aesthetic, economic and recreational benefits to the residents. The catchment is approximately 83,000 km2, encompassing a wide range of land uses, including both urban and agricultural areas. Lake Maracaibo itself provides fresh water for agriculture, supports a local fishery, and is the location of a petroleum industry, which accounts for 80% of Venezuelan oil production. Increasing salinisation due to the deepening of the shipping channel and increasing anthropogenic nutrient inputs, has culminated in Lake Maracaibo showing signs of accelerated eutrophication, indicated by intense phytoplankton blooms. The problems associated with eutrophication are not isolated to Lake Maracaibo. In the early 1960’s it became apparent that a large number of lakes and reservoirs, particularly those located in industrialised countries, were rapidly changing in character and becoming eutrophic, due to the addition of nutrients originating primarily from anthropogenic sources (OECD 1982). The eutrophication of inland water bodies has become synonymous with deteriorating water quality, and has frequently led to considerable costs being incurred by those responsible for managing the water bodies (OECD 1982). Accelerated anthropogenic eutrophication can be reversed by the elimination or reduction of the nutrient supply (OECD 1982). Thus, it is important to understand the complex relationship between nutrient supply and the ecology of the lake, the ecological response. A thorough understanding of the complex interaction between lake physics, chemistry, and biology provides a solid foundation on which a sound lake management strategy can be developed in order to obtain the desired water quality outcomes. Water quality issues in the Lake Maracaibo system are complex and include salinity changes, eutrophication, coliform bacteria, hydrocarbons, toxic contaminants, and anoxic conditions in the hypolimnion. Some of these issues can be addressed individually, through source control and management strategies. For example, pollution from oil spills and discharges of toxic chemicals, including pesticides and heavy metals, requires a strategy that focuses on the reduction of these releases and continued monitoring, in parallel with specific management efforts. Other issues, particularly those related to salinity, eutrophication, and oxygen levels are inherently more 1 Motivation complex because of their interrelationships. Salinity, along with the methods required to reduce it, has been a central issue in many past discussions and reports on the future of Lake Maracaibo. Presently, there is discussion on the merits of closing the shipping channel and moving the port facilities located within Lake Maracaibo to the Gulf of Venezuela in an attempt to restore the condition of the Lake. In general, salinity changes can influence the Lake’s ecology and the ability to use the water for other purposes. In addition, they can also influence other water quality parameters, e.g., dissolved oxygen in the hypolimnion and the vertical distribution of nutrients, both of which affect primary production. The “Integrated Study for the Environmental Remediation of Lake Maracaibo” (Bechtel 2000a) contributed significantly to the understanding of the above interactions in the context of the present state of the Lake Maracaibo system. The study used both a one-dimensional (DYRESMWQ), and a three-dimensional model (MIKE-3), to evaluate the overall water quality impacts of a range of environmental remediation options primarily aimed at reducing salinity and/or nutrient inputs. The abundance and composition of the phytoplankton standing crop is one of the most important indicators of water quality and eutrophication in lakes and reservoirs. Thus, a critical issue that must be considered when assessing the merit of any proposed remediation strategy, is the impact on primary production and algal species composition. Neither DYRESM-WQ nor MIKE-3 modelled both of the dominant algal species present in Lake Maracaibo (cyanobacteria and chlorophytes). DYRESM-WQ modelled cyanobacteria; conversely MIKE-3 modelled chlorophytes. Due to the differing assumptions as to the dominant phytoplankton group, the impact of various remediation strategies on primary production and algal species composition could not be assessed with any confidence. The dynamics of phytoplankton succession within Lake Maracaibo is a leading focus of this project, which seeks to provide advice on the effect of management options proposed by Bechtel (1999b), based on a detailed scientific study of the dynamics and ecology of Lake Maracaibo. The use of a coupled hydrodynamic-ecological model, DYnamic REservoir Simulation ModelComputational Aquatic Ecosystem Dynamics Model (DYRESM-CAEDYM), is an integral part 2 Motivation of this process. Numerical models of ecological processes are increasingly being used as lake management tools (Walters, 1986; Schladow and Hamilton, 1995; Harris, 1997). In a complex ecosystem, such as Lake Maracaibo, models allow for exploration of possible outcomes of various management options (e.g. closing the shipping channel and/or reducing anthropogenic nutrient inputs) without inadvertently causing an undesirable change in the ecosystem. However, Walters (1986) suggests that models may not represent the most efficient method available to predict the complex relationships and interactions between each ecosystem component, because standard scientific investigations fail to resolve many of the uncertainties implicit in models. To resolve many of the uncertainties associated with models, the best available data set was used in calibration, while numerous past field studies and data sets were used to verify the numerical model. The specific aims of this study are to: • Accurately model the complex hydrodynamics of Lake Maracaibo using the one-dimensional model DYRESM. • Couple DYRESM to the ecological model CAEDYM, and for the first time, model the observed dynamics of the two dominant algal species within Lake Maracaibo. • Use the fully calibrated DYRESM-CAEDYM model to determine the effect of proposed management options on long-term changes in water quality in Lake Maracaibo. 3 Background and Literature Review 2 Components of the Lake Maracaibo System 2.1 Lake Maracaibo Lake Maracaibo is approximately 160km long in the N-S direction, and 110km wide in the E-W direction, with an average depth of 25.9m and a maximum depth of 30m (Battelle 1974). The Lake is connected with the Gulf of Venezuela through the Strait of Maracaibo and Tablazo Bay, which together have a surface area of 1,080km2; this compares with the Lake surface area of approximately 12,000km2 (Jha et al. 1999). The lake basin is bounded on the east, south and west by high mountain ranges with a maximum altitude of 5,000 m (Battelle 1974). As one of the world’s largest lakes its protection and management presents a difficult and exceedingly important problem to which limnologists at all latitudes must make a contribution (Lewis 2000). GULF OF VENEZUELA Malecon WL Punta Diablo WL Tablazo WL TABLAZO BAY Malecon WL Cap. Chico MARACAIBO MARACAIBO STRAIT Palmas del Sur Punta Hicotea LAKE MARACAIBO Catatumbo Figure 2.1.1: The Lake Maracaibo System 4 Background and Literature Review It is believed that until relatively recent geologic times Lake Maracaibo was a body of fresh water with its surface above the sea level (Redfield, 1956). A river flowing along the western side of the valley, where the Strait of Maracaibo is today, drained the lake. A rise in sea level of about 20 m at the end of the last ice age created a link between the lake and the sea, resulting in the periodic influx of saline water into the lake. 2.2 Strait of Maracaibo The Strait of Maracaibo is approximately 40 km long, with an average depth of 10 m. Its width varies from 6 km at the narrowest point in the north to 17 km at the southern end, with an average width of 7.7 km. Tidal currents in the Strait are strong. The water in the Strait is fully mixed for the majority of the year, however, it becomes stratified during the dry season, as discharge through the Strait declines and saline water from the Gulf forms an underflow into the Strait. 2.3 Tablazo Bay Tablazo Bay connects the Gulf of Venezuela in the north with the Strait of Maracaibo in the south. Tablazo Bay is very shallow, with depths varying between less than 1 m and 5 m (Bautista, 1997), and with an average depth of 2.5 m. Approximately 25% of Tablazo Bay is less than 1.5 m in depth (Parra-Pardi 1980). The following openings connect the Bay with the Gulf: - Boca San Carlos-Zapara, which is 1.5 km wide, and through which a navigation channel is continuously dredged to a depth of 13.7 m; - Boca Ca onera, which is 0.9 km wide; and - Boca Ca onerita, which is 0.5 km in width. Although natural channels in Boca Ca onera and Boca Ca onerita can reach depths of up to 5-6 m, there is limited navigational access from Tablazo Bay to the Gulf of Venezuela through the “Bocas”, as shifting sands have narrowed and in some instances blocked these relatively unstable inlets (Bechtel 1999a). Most of the mixing in the Lake Maracaibo system takes place in Tablazo Bay, where the waters from the lake, Rio Limón, and the Gulf, interact (NEDECO 1993). The Rio Limón watershed in the north drains into Tablazo Bay in the vicinity of the navigation channel inlet and directly 5 Background and Literature Review influences the water exchange between the Gulf and the Bay. The large amount of sediment these rivers unload into the Bay plays a major role in shaping local coastal morphology. 2.4 The Gulf of Venezuela The Gulf of Venezuela is approximately 180 km long and 75 km wide, with an average depth of 30 m. It is delimited on its northern boundary by the 80-m isobath, which runs between the Gaojira Peninsula and the Paraguana Peninsula. A smaller gulf, 45 km in length and 6 m deep, is located on the eastern boundary of the Gulf of Venezuela (Bechtel 1999a). The Bay is separated from the Gulf of Venezuela by a series of shifting sand islands and sandbars. Outside these islands, the coast changes continuously due to littoral drift, predominantly in a south-westerly direction. Changes along the southern coastline of the Gulf have been reported since the end of the 18th century (Jha et al. 1999). The Gulf of Venezuela is connected with Tablazo Bay by the three openings (Boca San Carlos-Zapara, Boca Ca onera, and Boca Ca onerita) described above. 2.5 The Navigation Channel Before 1938, navigation between the Gulf and the Lake relied on the natural channels, particularly the Larrazabal channel passing between San Carlos and Zapara Islands. Construction of a stable channel was initiated in 1939 and a 6-m deep channel was maintained until 1947. In 1952, the Instituto Nacional de Canalizaciones (INC) commenced dredging operations for the creation of a new navigation channel. The new channel, with a depth of 12 m and a 3-km breakwater at the western end of Zapara Island, was completed in 1956. By 1963, the depth of the Canal was increased to 13.5 m. Through dredging this depth has been maintained to the present day (Bechtel 1999a). 6 Background and Literature Review 3 Hydrology 3.1 The Lake Maracaibo Catchment River lakes such as Maracaibo, receive water from large areas (river drainages) relative to their size. As such they are especially vulnerable to changes in the hydrology or water quality of connected rivers (Lewis 2000). Lake Maracaibo has a total watershed catchment of 71,000 km2, excluding its own water surface area of 12,000 km2 . Figure 3.1.1, below, depicts the total watershed and rivers draining into Lake Maracaibo. The watershed can be sub-divided into 30 catchments. The total mean annual runoff rate from these catchments is 1623 m3/s (CGR 1994). The distribution of this total runoff is uneven, both spatially and temporally. Figure 3.1.1: The watershed and rivers draining to Lake Maracaibo. 7 Background and Literature Review The largest tributary to Lake Maracaibo is the Catatumbo river, situated in the south-west corner of the lake with a total catchment area of 32,000 km2 (45% of the total watershed), contributing 60% of the total runoff (975 m3/s) to the lake. The catchment of this river extends into the Republic of Colombia. Santa Ana (184 m3/s) and Rio Escalante (103 m3/s) are the other two large rivers draining into the south-west corner of the lake. Rio Chama, situated at the southern end of the lake, is the fourth largest river (50 m3 /s) in terms of runoff. The runoffs from Catatumbo, Santa Ana, Rio Escalante and Rio Chama account for 81 % of the total runoff from the watershed of Lake Maracaibo. The total catchment area of these four rivers is equal to 65% (46,000 km2) of the total watershed area. The western side of Lake Maracaibo (catchment area 60%) yields 68% of total runoff, whereas the eastern side of Lake Maracaibo catchment, with 40% of catchment area, yields 32% of the total runoff. The catchment yield reduces towards the northern end of the lake (CGR 1994). 3.2 Precipitation Of the elements that effect the hydrological balance of Lake Maracaibo the most variable is precipitation. The basin-wide average is 1260mm/year, with values that range from 300mm on the west coast and 400mm in the northern portion of the basin, to 3500mm in the Colombian portion of the Rio Catatumbo sub-basin (Bautista 1997). The annual regime of precipitation follows a bimodal distribution, with two peaks, one between April and May, and the other in October. In general the dry season occurs between JanuaryMarch and the wet season between May-November, with April and December in transition. 3.3 Hydrological Balance The hydrological balance of the lake includes five elements: (1) land drainage; (2) rainfall onto the open waters; (3) evaporation losses; (4) ground water discharge and recharge, and (5) net flow through the Strait. The contribution of ground water to the water balance can be neglected, as it is relatively small. Several studies on the hydrologic balance of the Lake Maracaibo system have shown a close relationship between the annual cycle of rainfall on the basin, and the water exchange between the lake and the sea. In his 1964 study, Luis Corona analysed hydrologic data from 1940 to 1960. He estimated that the nett fresh water entering the lake in 1960 was 1,565 m3/s (Corona, 1964). Battelle (1974) estimated that the fresh water entering Lake Maracaibo was 2,095 m3/s. Using precipitation, evaporation, and stream gauge data, a MARNR study 8 Background and Literature Review concluded that between 1958 and 1977 fresh water inflow was 1484 m3/s (Parra Pardi, 1977, 1979). Further analysis using precipitation, evaporation, and current measurements taken between 1973 and 1983 by Rincón, estimated that the Catatumbo River discharges 1,147 m3/s, or 60% of the total drainage (1909 m3/s) into the Lake. A 1994 hydrologic study conducted for ICLAM, CGR (1994), estimated that the average annual fresh water flowing into the lake is in the order of 1,600 m3/s. The results of the different hydrologic balance studies are summarised in Table 3.3.1. Table 3.3.1: Nett fresh water inflow into Lake Maracaibo Study Fresh water inflow, m3/s 1565 2095 1484 1909 1600 Corona, 1964 Battelle, 1974 Parra Pardi, 1977, 1979 Rincón, 1993 CGR, 1994 9 Period 1960 1975 1958-1977 1973-1983 1973-1990 Background and Literature Review 4 Hydrodynamics of Lake Maracaibo In this section the hydrodynamic regimes operating in Lake Maracaibo are discussed. An understanding of lake hydrodynamics is important because the mixing and transport processes operating in a lake determine, to a large extent, the ecological response of the lake to the meteorological forcing, inflows and outflows (Imberger 2001) 4.1 Hydrodynamic Forcings 4.1.1 Tides A summary of earlier studies on tidal hydrodynamics can be found in Lynch et al. (1990), where studies by Redfield (1961), Kjerve (1981), Molines and Fornerino (1985), Werner and Lynch (1986), Lynch and Werner (1986, 1987) and Molines et al. (1989) are reviewed. These papers describe both data analysis and modelling efforts to explain the tidal dynamics of the Lake Maracaibo system. The tide in the Caribbean Sea is mixed, but predominantly diurnal (Redfield 1961), with a micro tidal range of 10 to 20 cm. In Aruba the mean range at spring tide is 21 cm. In the Gulf of Venezuela, the semi-diurnal components of the tides are greatly augmented by resonance. The tidal range initially increases to 1.1 m at Malecon but subsequently decreases to 0.6 m at Punta de Palmas, inside Tablazo Bay. At the northern end of the lake spring tides appear to be diurnal, while semi-diurnal components are evident during neap tides. The tidal range declines from 0.2 m at the south end of the strait to a few centimeters within the lake (Redfield 1961, NEDECO 1993). There are considerable phase lags from north to south. As tides propagate from Malecon in the north, towards the south, they experience a delay of 50 minutes at Maracaibo, over a distance of 40 km (INTEVEP 1998b). Along the Strait of Maracaibo, high water is nearly synchronous. Tides in the lake are nearly 180° out of phase with respect to those found in the Gulf of Venezuela, thus when the tide at Malecon is rising, the currents flow towards the lake at the inlet to the Bay of Tablazo, but flow seaward in the strait (Lynch et al. 1990). 10 Background and Literature Review 4.1.2 Winds The trade winds affecting the lake system blow predominately in the northeasterly direction from November to April, and diminish considerably for the rest of the year (Bautista 1997). The monthly wind speed in the bay peaks at 5-10 m/s in March, dropping to 3 m/s from May to December. This windy period coincides with the dry season. The local wind field over the Lake develops from the differential heating of land and water masses, and exhibits a counterclockwise pattern. The wind field over the lake differs from that found in the bay and the gulf (INTEVEP 1998b). The wind speed is of a smaller magnitude and never persists for more than 12 hours from any direction. Gusts of wind might have a magnitude four or more times the mean wind speed. The maximum wind waves range from 1.5 m in the bay to 1.2 m in the lake. The mean wave period ranges 3.5 seconds to 4.3 seconds (INTEVEP 1998b). 4.1.3 Currents A counter-clockwise current dominates the circulation in the lake. This current is driven primarily by wind and is supported by the density stratification and the Coriolis force. As the current spins the epilimnion water, the saline water in the hypolimnion is forced to rise at the centre of the lake to within 15 m of the surface, forming a cone-shaped hypolimnion (Redfield, 1956). Measurements conducted in March 1974 show the same counter-clockwise circulation pattern in the surface and the bottom current. The velocity at the centre of the lake was low near the surface (7 cm/s) but increased to 12 cm/s near the bottom. On the other hand, the velocity in the shallow areas of the lake had a large magnitude (on the order of 20 cm/s), which decreased with the depth (Battelle 1974). Currents play an important role in mixing and the formation of a conical hypolimnion at the centre of the lake (Redfield 1961), the top of which, has been observed within 5 to 15 m from the water surface (Bautista 1997). Parra Pardi (1996) In the lake, wind-driven seiche can, on some occasions, be so severe as to upset water column stability, allowing nutrient-rich, hypolimnetic waters to mix into the epilimnion (Parra Pardi 1996). 11 Background and Literature Review 4.2 Mixing Processes Water quality in Lake Maracaibo is determined predominantly by an influx of phosphorus and nitrogen particulates carried by rivers, with additional anthropogenic sources on the eastern shores of the lake, and from population centres along the strait. These nutrients spread out horizontally over the lake surface and feed a substantial primary production. The nutrients attached to particles, as well as the particles formed due to primary production, then settle to the lake bottom. At the lake bottom, whenever anoxic conditions prevail (due to salt stratification), nutrients are released in forms which are accessible to further primary production. Trapped within the saline hypolimnion, these nutrients are recycled by continuous entrainment, driven by the circulation in the lake and by mixing events that vent the deep nutrient pool to the surface, where they sustain further growth. The above-described processes, particularly nutrient loading to the surface layer, sedimentation, and stratification that supports anoxic conditions and controls entrainment across the density interface, are all extremely important for predicting primary production in Lake Maracaibo. Lake Maracaibo is subject to a spatially varying wind-field, which in association with the Coriolis force creates a basin-scale cyclonic gyre (Redfield 1960, Bautista et al. 1997). The interplay between the strong density stratification and the cyclonic gyre induces vertical shear. The subsequent result is a very interesting doming the dense more saline hypolimnion. Rhines (1995) observed a similar phenomenon in a small, rotating, stratified basin, in which vertical shear was accommodated by thermal wind-tilting of isopycnals. The doming of the hypolimnion raises nutrient rich, highly saline water nearer to the Lake surface, where it can be entrained into the surface mixed layer. Tidal action induces a periodic saltwater influx which enters the lake at is northern end and flows down along the bottom to form a saline hypolimnion. From there it is mixed upwards into the epilimnion. The anti-clockwise cyclonic circulation of the surface waters causes an upconing of the hypolimnion in the centre of the Lake. As a result the hypolimnion is withdrawn from the bottom around the margins, with the epilimnion occupying the entire water column, even to a depth of 30m (Redfield 1960). The resultant dome-shaped hypolimnion has an apex at a depth of 5-15m, and salinity ranging between 5-15 ‰ (Jha et al. 1999). The epilimnion and hypolimnion 12 Background and Literature Review mix most intensely at the apex of this dome. Consequently this area is the principal source of the salt in the epilimnion and the chlorides in the Lake’s surface are highest in the centre (Redfield 1960). Tidal and wind induced mixing generate vertical and longitudinal salinity gradients, typical of a partially mixed estuary (Jha et al. 1999). The counter-clockwise current which circulates around Lake Maracaibo at an average speed of half a knot, mixes the surface waters thoroughly, down to the upper limit of the pool of saline water over the bottom, and produces almost uniform conditions across the entire lake (Redfield 1958). 4.2.1 Internal Dynamics Acoustic Doppler Current Profiler (ADCP) field data taken from the two northernmost stations in Lake Maracaibo, LM10 and LM20 (see Figure 4.2.1), confirms the existence of a counterclockwise circulation of 20-30 cm s-1 (Horn et al. 2001). The velocity of the current at the southern end of the lake is slower than that in the north. However, the flow is still in a direction consistent with a general counter-clockwise circulation (Horn et al. 2001). Figure 4.2.1: 1998-99 (wet-dry) Campaign Field Stations 13 Background and Literature Review The downward curvature of isopycnals in the main basin of Lake Maracaibo, and consequent higher surface salinity near the lake centre, first documented by Redfield (1958), is evident in the 1998 salinity data (Laval et al. 2001b). Based on the stratification measured in the field, the internal Rossby radius of deformation was approximately 22 km (Laval et al. 2001b). This indicates that rotation is important in the internal dynamics, further supporting the hypothesis that the doming of the hypolimnion is due to a geostrophic balance with the cyclonic circulation (Redfield 1958; Parra-Pardi 1983; Bautista et al. 1997). 4.2.2 The influence of the earth’s rotation In large lakes such as Maracaibo, the earth’s rotation is often important in that it may influence the motion in the lake. When a particle moves in a straight line on a rotating coordinate system it appears to be deflected. The object does not actually deviate from its path, but it appears to do so because of the motion of the coordinate system. The apparent deflection of particles moving over the Earth is explained in terms of the Coriolis force. The non-dimensional Burger number is often used to determine the importance of the earth’s rotation on a lakes internal dynamics. The Burger number (Equation 4.2.1) is defined as the ratio of time of travel of an internal wave across the lake (L/ci), to the time it takes for the lake to rotate about its axis (f –1) (Imberger 2001). (Equation 4.2.1) c Si = i Lf where ci is the phase speed of the wave, L is the typical “radius” or width of the lake at the depth of the thermocline, and f is the inertial frequency at the latitude of the lake. When Si < 1 the earth’s rotation dominates the internal dynamics and the waves have the character of an internal oscillation with most of the energy in the wave being kinetic energy. As Si increases above one the internal oscillations progressively takes on the characteristics of simple gravitational seiches (Imberger 2001) In Lake Maracaibo the Burger number ranges seasonally between 0.12 and 0.39 (Roza-Butler 2001) illustrating the importance of the earth’s rotation on the internal dynamics of Lake Maracaibo. 14 Background and Literature Review 5 Previous Studies This Section provides a brief overview of past studies emphasising major findings and conclusions that are still valid and relevant. The major studies fall into one nine groups: 1. The studies by Woods Hole in the 50’s and MIT in the late 50’s and 60’s; 2. The Battelle study in 1972; 3. The Studies by Parra Pardi in the late 70’s and early 80’s; 4. The NEDECO study in the early 90’s; 5. The ICLAM studies and data collection program in the 90’s; 6. The INTEVEP studies in the 90,s; 7. The Studies at the University of Zulia; 8. Other studies of interest; and 9. The Bechtel Integrated Environmental Remediation Study 5.1 The Woods Hole and MIT Studies In a series of papers, Alfred Redfield analysed organic matter in the sediments of Lake Maracaibo (1956), water quality processes in the Gulf of Venezuela (1960), and tidal characteristics of the Lake system (1961). The importance of his work lies not only in the identification of various processes which have since been confirmed by other researchers, but also in providing a set of historical data against which the changes in water quality parameters can be measured. Redfield was the first to describe the existence of a counter-clockwise current in Lake Maracaibo (on the order of half a knot), attributing the cause to the prevailing winds (1956). He observed that the salt water intrusion into the lake occurred only during the three months of the dry season. In the late 50’s, the lake’s surface water contained about 33‰ seawater, with a total phosphorous content of 40 _g/l. The concentrations of both chloride and nutrients were considerably higher in the hypolimnion, with chloride >2ppt and total phosphorus >6 _g/l (1960). A combination of hydrographic conditions and biological activity resulted in 15 Background and Literature Review depleted hypolimnetic oxygen content, with no oxygen present in the lake centre. Redfield (1956) determined that phosphorous and presumably other nutrients were derived almost exclusively from land drainage. He found that in its particulate forms, from river inflows and as organic detritus from primary production, phosphorous enters the saline hypolimnion through sedimentation, accumulating in the “deep waters”. The available oxygen was noted to be insufficient to oxidize this accumulation, resulting in conditions unusually favourable for the entrapment of organic matter in sediments. One third of the phosphorous was present in inorganic form, available for plant nutrition. Consequently, growth is not limited by the availability of phosphorous in the photic zone (Redfield 1956). Quoting a 1933 sediment analysis by Trask, Redfield (1956) argued that the concentrations at which organic matter is entrapped in sediments depend upon the relative rates of sedimentation, and of the accumulation of organic matter at the lake bottom. The principal source of sediment input into Lake Maracaibo is the Catatumbo River. The distributions of organic carbon, nitrogen, and organic extract in bottom sediments mimic the distribution of oxygen deficiency in the water over the bottom, with the exception of the pattern being skewed toward the eastern side of the lake due to that side’s greater rate of sedimentation. Redfield also analysed tidal data (1961) and suggested that the dimensions of the Gulf of Venezuela are such that they encourage the semi-diurnal constituents to be augmented by resonance, while the diurnal constituents remain relatively unaltered. An MIT study (1967) by Harleman, Corona, and Partheniades utilised simultaneous salinity measurements in the navigation channel, and established a relation between a dimensionless intrusion length parameter and a characteristic estuary number. The parameters include the tidal prism, the magnitude of current in the inlet to the estuary, the average water depth, fresh water discharge, and the tidal period. This relation was then demonstrated using simple theoretical argument, and the assumption of a partial or complete mixed water column. If the tidal prism decreases as a result of the partial or complete closure of the inlet, the analysis predicted reduced saltwater intrusion. 16 Background and Literature Review 5.2 The Battelle Study Sponsored by Creole Petroleum Corporation, the Battelle study constituted the first major investigation into the ecology of Lake Maracaibo, south of the Straits. This study produced baseline data on water quality, primary productivity, macrofauna, sediments, and fishery resources of the lake, as well as data on major sources of industrial and domestic wastewater discharged into the lake (Battelle 1974). The salinity patterns obtained at one metre depth and one meter above the lake bottom were similar to those of Gessner (1956) and Redfield (1960), however the values were approximately twice as high. The average phosphorous concentration in the upper 5m of the water column ranged between 40-280 _g/L while the total nitrogen concentration ranged between 779-1580 _g/L. These concentrations were several times higher than those measured by Redfield (1960). Algal blooms were observed in late winter and early spring. During the wet season, it was observed that the river discharge contained a high concentration of suspended solids. The information gathered in the Battelle study supported the conclusion drawn by Redfield (1960), in that salt is primarily introduced into the epilimnion from the hypolimnion, largely near the lake centre, whereas the freshwater input is derived from the rivers on the margin of the lake. The cone-shaped hypolimnion previously described by Redfield (1960) was most apparent in the wet season (November 1973). There was also a positive correlation between the salinity in the hypolimnion and the high levels of PO4-P (up to 4900 _g/L) and NH3 (up to 24,800 _g/L) observed in this region. Battelle (1974) found that only a small portion of nutrients released by biological or chemical actions returned to the epilimnion through diffusion and were made available for biological utilisation. The majority of nutrients were retained in the hypolimnion until the next mixing event. Thus, mixing events impacted greatly on the chemistry of the lake water. Sediments were found to be rich in organic material, with extractable organic contents as high as 3% of the total dry weight. Phytoplankton abundance was observed to follow a cyclic pattern with the highest levels, among the highest found in the world, occurring after mixing events. The distribution and diversity of phytoplankton in the lake was approximately the same in winter and summer, and from station to station, with the blue-green algae being the most dominant group, the highest population densities being recorded along the Bolivar (north-east) coast. 17 Background and Literature Review The Battelle report compiled a summary of the domestic waste discharges (Battelle Appendix F, 1974). The study summarises non-petroleum industrial wastes such as NH3-N, BOD, suspended solids, COD, phosphorous, and sodium pentachlorophenate. Several large rivers (including the Catatumbo, Santa Ana, Escalante and Motatan) were also sampled during the dry and wet seasons in order to evaluate the river water quality. The levels of dissolved NH3-N and organic nitrogen were found to be high (~300 _g/L). The study warned of the potential adverse impacts of non-petroleum wastes, both domestic and industrial, on water quality. 5.3 The Parra Pardi Studies From 1974 to 1979, Parra Pardi and his associates undertook a comprehensive study aimed at understanding the lake ecosystem, evaluating its capacity to assimilate contaminants, establishing scientific bases for water quality standards, and making recommendations for corrective actions. The study was commissioned by la División de Investigaciones sobre Contaminación (DISCA). An array of methodologies was used: physical modelling of the strait and the bay, a mathematical model of the strait and the bay, intensive field observations in the strait, and a hydrologic analysis of the water balance. The work of Parra Pardi and colleagues was presented in two reports (1977 and 1979). The reports proposed a strategy for managing water quality in the system, collecting and treating the domestic and industrial wastewater from the City of Maracaibo, the Cities of La Silva and La Rosa, and re-utilising industrial and agricultural wastewater. 5.4 The NEDECO Study In the November - December 1992 wet season campaign and the March 25 to April 8 1993 dry season campaign, wind velocity, water velocity and other biochemical and ecological data were collected at 23 locations scattered in the gulf (3), the bay (4), the strait (4), and the lake (13). For the same period, tidal data was collected at 14 points (the gulf: 3; the bay: 4; the strait: 3; the lake: 4), and water flowing into the Lake from 15 major rivers was sampled. The total river flow into the lake, including flow from the Limón, was found to be 2167.3 m3/s in the wet season and 571.4 m3/s in the dry season. During the dry season campaign, the maximum flow through the strait was in the order of 25,000 m3/s. The purpose of this study was twofold, to quantify agricultural, domestic and industrial water 18 Background and Literature Review demand, and to estimate water supply. Intrinsic to the NEDECO study, an engineering evaluation of alternative methods of lake desalinisation was undertaken. The alternatives considered included reducing the depth of the navigation canal via the natural process of sedimentation, construction of parallel dikes, and partial or complete closure of the inlet. By reducing the Lake salinity, it was thought that the accelerating processes of contamination and eutrophication of the lake over the last 25 years could be stopped, or even reversed. This study investigated both the salinity and sediment transport through the strait under annual flow conditions typical of the period 1973-1983. The important factors contributing to eutrophication were found to include oil spills, industrial discharge of inorganic nutrients, ammonia compounds and toxic materials from some 104 point sources, agricultural discharges, and the sewage discharges of 1.8 million people living in Maracaibo. Rivers were estimated to carry 1943 m3/s water and 182,900 kg/day total nitrogen into the lake system. The domestic and industrial wastewater, on the other hand, delivers 16.9 m3/s water and 61.3 kg/day total nitrogen. The industrial organic pollutant loading, concentrated in the north-east region of the lake and in the strait, makes a lesser contribution in terms of quantity. The report analysed water demand and supply for 1992, projecting the future demand and supply to the year 2020. In order to meet this future demand for water, additional dams would need to be built upstream. Considering the number of dams that are already under construction, one might anticipate that in the near future the lake system would operate under a different hydraulic regime. As a part of the study, the Delft Hydraulic Laboratory conducted an investigation of the lake system by means of a 3-D numerical model. The model incorporated hydraulic processes (tides, currents, river discharges, and the mixing of fresh and salt water), physical-chemical processes (nutrients, contaminants, dissolved oxygen, and suspended solids), and ecological processes (dynamic cycle characteristics, eutrophication, and algal blooms). The numerical simulation conducted by Delft indicated that mixing in the strait was sufficiently intense that over the course of several years, contaminants released into the strait would be carried toward the bay and the gulf by net water discharge from the lake. Thus, leaving only a small portion of contaminants entering the lake (NEDECO 1993). 19 Background and Literature Review 5.5 The ICLAM Studies The Master Plan regarding the control and management of water quality in the Lake Maracaibo Basin prepared by ICLAM (1991) presents a considerable amount of information that has been tabulated in the context of water quality goals to be achieved by the year 2005. A comprehensive database on salinity and other water quality parameters has become available in the last 5-6 years, due to the data collection efforts of ICLAM. The work of analysing this data and correlating water quality parameters with dimensionless hydrodynamic parameters such as the estuarine number, the lake number, and the inflow Froude number has not yet begun. However, current research undertaken at the Centre for Water Research (University of Western Australia) may begin to elucidate the link between non-dimensional numbers and water quality in Lake Maracaibo. Using data collected in July 1996, Godoy (2000) estimated the balance of total nitrogen in the lake. The inputs to the lake are 278 ton/day, including 158 ton/day from river discharges, 80 ton/day from rainfall, and 40 ton/day from wastewater. An estimated 70-140 tons of total nitrogen is removed from the lake by outflow each day. This results in a net gain of 138 to 208 tons of total nitrogen, which remains in the lake water and sediment. 5.6 The INTEVEP Studies INTEVEP conducted several studies of the hydrodynamics and water quality of the system, as well as studies on bottom sediments. A good summary of these studies is provided in three recent reports by INTEVEP (1998a, b, c.) The sediment sampling campaigns conducted by INTEVEP in 1996 and 1997 in the watershed of the Catatumbo River and Lake Maracaibo focused on the content of heavy metals, such as vanadium, nickel, chrome, zinc, barium, mercury, and iron. In general, it was observed that the distribution patterns of metals in the sediments deposited in the lake system have the lowest values in the strait and the highest values in the lake. This work constitutes a reference baseline for future studies on sediment quality. The recent report titled “Evaluation of the Physiochemical Parameters of the Water Column and the Sediment Quality of Lake Maracaibo” (1998c) presents the results of measurements of 20 Background and Literature Review physicochemical water parameters and sediment quality in Lake Maracaibo, obtained during a sampling campaign performed in July 1994. The results indicate that epilimnion water of the lake has slightly higher temperatures than the hypolimnion, low salinity (4‰), elevated dissolved oxygen content, and alkaline pH in an oxidizing environment. The hypolimnion has cooler water, moderate salinity (>10‰), less alkaline pH and absence of dissolved oxygen. The sediment in the western region of the lake was characterised by lower organic matter and lower metal content than sediments from the central, eastern and southern regions. The report titled “Water Quality of Lake Maracaibo” (1998b) presents an analysis of the water quality of the Maracaibo System, and also discusses the salinity measurements taken between 1963 and 1971. The report suggests that the vertical stratification of chlorides is a typical estuarine stratification, and concludes that the stability of salinity values over the last 30 years demonstrates that a new equilibrium has been reached. 5.7 Studies at the University of Zulia Various studies on specific topics related to the environmental quality of the lake have been conducted over the years by faculty and students at the University of Zulia. In the continued effort to model numerically the Lake Maracaibo system, Molines et al. (1989) has demonstrated two important wind effects: the development of seiche and large scale eddies within hours of the initiation of wind. These processes, though having particular significance for mixing and transport of waters in the bay, however seem less relevant when considering mixing processes in the Lake. 5.8 Other Studies of Interest An issue that has received considerable attention in the literature, is the increase in salinity of the surface water over the years and the question of whether or not this process of salinisation in continuing. When historic annual average salinity data, taken at 1-m below the water surface is plotted, it becomes apparent that the salinity of the lake increased in the mid-50s (when the navigation channel was deepened.) Whether the salinity of the Lake has now reached a new equilibrium state or is rising to higher levels remains a subject of debate. Performing a linear regression on the data series taken between 1954 and 1996, Bautista concluded that the water in the lake has become steadily more saline (Bautista et al. 1997). In contrast, Parra Pardi (1996) 21 Background and Literature Review conducted a linear regression on the data taken between 1960 and 1995 and found no convincing evidence that the lake is in a process of salinisation. Parra Pardi further suggested that the changing hydrology of the lake watershed in recent years (as a result of deforestation,) may be the cause of some of the variations in salinity. 5.9 The Bechtel Integrated Environmental Remediation Study A multi-year study conducted by Bechtel International (2000a,b,c), under the sponsorship of Petroleos de Venezuela, S.A. (PDVSA), evaluated the current environmental condition of the Lake Maracaibo system, assessed the sources and quantity of nutrients and contaminants entering the Lake, and then applied two computer models (DYRESM-WQ and MIKE-3) to determine short and long-term changes in water quality due to modifications of the navigation channel, and a reduction in nutrient loading. The study included development and preliminary cost analysis of alternative pathways to continue the transport of oil, coal, petrochemicals and general cargo to and from the Lake. The evaluations and analyses of this study lead to the conclusion that the solution to the problem of anoxic conditions in the hypolimnion requires the reduction of dead organic matter in the Lake, which in turn requires a reduction in primary production (phytoplankton, i.e. algae). The natural conditions in the Lake indicate that this can not be easily accomplished. Two remediation options were analysed in depth during this study. Both options involved a 10% reduction in nutrient loading to the Lake through point source and non-point source pollution control, with the second option, additionally evaluating the effect of allowing the Lake system to return to its pre-dredging bathymetry, through closure of the navigation channel. The option of closing the navigation channel and allowing the system return to its pre-dredging bathymetry was found to have the greatest positive impact on salinity and nutrient levels. The models indicated that there would be a seasonal improvement in oxygen levels in the Lake, but during several (in some years up to seven) months per year, anoxic conditions would persist in the hypolimnion, just as they do today. Channel closure would also reduce stratification and lead to a subsequent reduction in the concentration of nutrients in the Lake. However, the effect on total chlorophyll-a and algal species composition was uncertain. MIKE 3 predicted a 35% reduction in chlorophyll-a if the navigation channel was allowed to 22 Background and Literature Review return to its pre-dredging bathymetry, while DYRESM-WQ predicted no change in chlorophylla. The difference in the predictions by the two models was due to the fact that each modelled only one species of algae. DYRESM-WQ simulations assumed the main bloom-forming species were cyanobacteria, while MIKE 3 simulations assumed chlorophytes were the dominant algal species. However, evidence from past studies indicates that both species are present at high concentrations within the Lake, and that during different periods of the year algal dominance changes between cyanobacteria and chlorophytes. Due to the differing assumptions, neither model could provide any meaningful insight into the likely impact of proposed management strategies on phytoplankton species composition, and thus water quality. 23 Background and Literature Review 6 Historical Salinity Trend The aim of this analysis of historical data is to determine whether salinity levels in the Lake have reached a new equilibrium following construction of the navigation channel. An analysis of the salinity data at 1m depth has been conducted to reduce any bias due to the uneven distribution of data in time. The data sources utilised in this analysis include those used by Bautista et al. (1997) together with additional data from the ICLAM database for the period 1996-1998 (Appendix A). Figure 6.1 shows the linear trend line obtained from the regression analysis of all data. The slope of this line, i.e. the rate of salinity increase, is 0.054‰ per year, giving the impression that Lake Maracaibo has been undergoing a process of salinisation since 1954. This relatively high average rate of salinity increase is due, in part, to the fact that this data set includes data collected prior to the deepening of the navigation channel, a discrete event that is generally believed to have contributed to the increase of salinity levels in the lake. To assess the current trend in salinity levels it is more appropriate to work with data obtained post completion of the navigation channel. Salinity of Lake Maracaibo at 1m depth (Data: 1954-96 from Bautista et al. (1997) and 1996-98 from ICLAM Database) 5.5 5 ] u s p [ y t i n i l a S salinity 50y r trend 20y r trend New Equilibrium Salinity 4.5 4 3.5 3 2.5 2 1.5 1 Jan54 Jan77 Date Figure 6.1: Salinity of Lake Maracaibo at 1m depth 24 Nov 98 Background and Literature Review To reduce the biases due to uneven distribution of data in time, it was decided to test the time series from 1977 to 1998, a period for which a significant data series exists. Prior to 1973 there was only one sampling event per year, and between 1973 to 1977 only two sampling events per year. The average of all salinity values between 1977 and 1998 is 3.81‰, with a root mean square error of 0.059‰. Thus, it appears that the lake has reached a new state of equilibrium over the past twenty years, with salinity in the surface waters stabilising at approximately 3.8 g/L. To better understand salinity changes in the lake, it is important to examine the processes giving rise to saline-fresh water exchange at the lake entrance. 6.1 Behaviour of the Underflow Prior to the deepening of the channel, periods of high rainfall flushed the Lake, and salt stratification was replaced by a weak thermal stratification (Redfield 1961). The deepening of the channel in the mid 1950's allowed more seawater to enter the Lake, increasing both the epilimnetic salinity, and the salt stratification in the bottom waters (Parra-Pardi 1983). Salinity stratification is now a permanent feature of the Lake (Laval et al. 2001a). The salt water in the Gulf of Venezuela enters the Bay of Tablazo through the mouth of the estuary. In the Bay of Tablazo, which has an average depth of 2.5m, the water is generally mixed over the entire depth (NEDECO 1993). A definite stratification has been documented in the navigation channel leading to the Strait of Maracaibo (Jha et al. 1999). In the strait, fresher water from Lake Maracaibo meets the saline water from the bay, forming a salt wedge along the length of the strait. The exact nature of the salt wedge, the strength of the stratification, and the positions of the isohalines are determined by both the baroclinic and barotropic flows induced by a combination of the: • Barotropic tides; • Nett outflow due to the fresh water inflows to the lake; • Baroclinic pressure gradient due to the salinity difference between the bay and the lake; and • Vertical diffusion of salt. When these influences combine favourably, salt water from Tablazo Bay is able to negotiate its way through the strait and flow down the incline into the bottom of Lake Maracaibo. Two 25 Background and Literature Review inferences follow directly from this general description. First, the underflows of salt water into Lake Maracaibo are most likely tidally modulated, and therefore of short duration. Second, the periods during which spring tides can combine with other influences to produce substantial salt water inflows into Lake Maracaibo will be seasonally modulated, with major underflow events occurring on a small number of occasions each year (Bechtel 1999a). Freshwater B E Outflow D A C Figure 6.1.1 Schematic diagram of the salt flux path in the Maracaibo System along the navigation channel. A) Transfer across the mouth between the Gulf of Venezuela and Tablazo Bay; B) mixing and secondary transport in Tablazo Bay; C) Unsteady salt wedge dynamics in Maracaibo Strait; D) Plunging salt water underflow into Lake Maracaibo; E) Vertical mixing in Lake Maracaibo. As can be observed from Figure 6.1.1 lower density water flows out of the lake on the surface, while denser water from the Bay of Tablazo flows towards the lake. It has been suggested that hydraulic discharges control the balance of salinity in the lake (Laval pers. com. 2001). During the wet season, when discharge velocities are high, the mouth of the strait is flooded with water from the lake, preventing salt water from entering. When discharges drop during the dry season, and aided by tidal effects, the saline water enters the lake. It was observed that when fresh water flows into the ocean, a current of deep salt water flows in the opposite direction (Bechtel 1999a). Tides thus produce cyclical oscillations resulting in vertical and longitudinal salinity gradients. 26 Background and Literature Review 7 Phytoplankton Composition & Succession Seasonal changes in the abundance and composition of phytoplankton may be observed to occur on temporal scales measured in weeks or months, during which mean specific population densities may increase or decrease through 6-9 orders of magnitude (i.e. from 10-3 to 106 cells ml-1 ) (Reynolds 1984). The time scale is such that the amplitude range of environmental oscillations is itself likely to alter significantly. In temperate lakes, for instance, the environmental variables that are subject to such profound changes include day-length, total irradiance and its attenuation within the underwater light field, temperature, thermal structure and nutrient availability, as well as the abundance of herbivorous animals. Reynolds (1984) stipulates that Tropical lake communities will be influenced by broadly similar constraints to those that operate in temperate waters. The ecological responses of the phytoplankton to these changes are generally well defined, although factor interaction often prevents a clear appraisal of cause and effect, and may even lead to misinterpretation. Correct interpretation is essential if year-to-year variations and long-term responses of phytoplankton composition to climate change, hydrological change, and eutrophication effects are to be adequately understood and predicted. There have been a large number of studies made and descriptions given of phytoplankton periodicity, though few have seriously attempted to account for the patterns observed. Reynolds (1984) cites several classical studies on individual lakes, or groups of temperate lakes (e.g. Birge and Juday 1922; Ruttner 1930; Pearsall 1932; Grim 1939; Findenegg 1943) which established the clear year-to-year similarities in changes in abundance and composition, however there is a dearth of such literature elucidating the patterns occurring in Tropical Lakes. In large tropical lakes chlorophytes and cyanophytes dominate, and in some instances diatoms also (Lewis 2000). Cyanobacteria are generally absolutely and/or relatively more abundant in tropical lakes (Reynolds 1984) than in those at higher latitudes. Water blooms of cyanobacteria develop during periods of stratification, low turbulence, and high temperature. The absence of real water blooms of cyanophytes and dinoflagellates in deep large lakes may be due to the lack of one of the prerequisites for their development; minimal water turbulence. A given wind velocity leads to a greater turbulence per unit area in large lakes than in small water bodies. Consequently, in low-latitude lakes, chlorophytes and diatoms dominate during the mixing 27 Background and Literature Review period (Reynolds 1984). 7.1 Historical Phytoplankton Composition & Succession Due to the high nutrient inputs, nearly constant sunlight, and elevated temperatures, the densities of phytoplankton in Lake Maracaibo are extremely high. According to Dr. Fritz Gessner (1956) “Lake Maracaibo may well be the richest in plankton of all the waters of the earth.” It is important to understand the composition of the phytoplankton community as this provides an indication of water quality. Phytoplankton blooms within water bodies tend to be indicators of cultural eutrophication, and can have serious toxicological and aesthetic implications for the water, and for organisms that inhabit those ecosystems. Phytoplankton form the basis of the food web for aquatic fauna; however, high to excessive levels of cyanobacteria (blue-green algae) are indicative of nutrient-rich, eutrophic conditions. The secretion of toxic nuerotoxins is more often associated with cyanobacteria than any other algal group (Margalef 1983). Thus, high concentrations of cyanobacteria may be harmful to some vertebrates. However, at moderate levels the phytoplankton provides a rich food base for the entire aquatic community, thus contributing to its richness and diversity. Nutrient rich river inputs combined with large volumes of untreated residential waters and the occasional petroleum spillage have resulted in Lake Maracaibo showing signs of accelerated eutrophication. Throughout the lake system, the phytoplankton distribution correlates with the nutrient levels in the water. At the centre of the lake, the phytoplankton population in the surface water is low in the dry season, increasing considerably during the wet season. Algal populations are generally higher along the lakeshore, where the water depth is shallow. In the northeastern part of the lake, especially along the shoreline, there are high concentrations of blue-green algae, believed to be due to a combination of river discharges, lake currents, tidal mixing, and prevailing wind direction (Battelle 1974). By presenting the characteristics of the successional dynamics and the temporal and spatial variations of the lacustrine phytoplankton populations, we can attempt to establish possible relationships between the phytoplankton blooms and the process of eutrophication in Lake Maracaibo. Researchers at ICLAM reviewed variations in phytoplankton populations in the lake for the 28 Background and Literature Review years 1973, 1974, 1978, 1986, 1988, and the period of 1992 through 1994 (Latchinian et al. date unspecified). Phytoplankton data obtained during 1973, 74, 78, and 86-88 was collected it such a fashion that it was not possible to establish consistent seasonal patterns. Figure 7.1.1 presents the general trend of phytoplankton development for the period 1973 -1994. Phytoplankton Evolution (1973-1994) 100 90 n o i t i s o p m o C 80 Cy anobacteria Chlorophy tes Chry sophy tes Other 70 60 50 40 % 30 20 10 0 1973 1994 Y ear Figure 7.1.1: Phytoplankton Evolution in Lake Maracaibo (1973-1994) It is evident from Figure 7.1.1 that chlorophytes have become relatively more abundant over the period 1973-1994, however, cyanobacteria remain the dominant species throughout this period. It is interesting to note that cyanobacteria were far more dominant in the early 70’s comprising approximately 63% of the total phytoplankton standing crop. Thus, if the navigation channel were to be filled, it would be valid to assume that the phytoplankton composition would shift back towards conditions that favour cyanobacteria. Battelle (1974) also documented the dominance of cyanobacteria during the early 70’s. On analysing data from 1973-1994 Latchinian et al. (date unspecified) identified four dominant groups of phytoplankton: cyanobacteria, chlorophytes, chrysophytes, and others (euglenophytes, pyrophytes, and chryptophytes). Figures 7.1.2 shows typical periodic phytoplankton population succession between the two most dominant groups in the lake, cyanobacteria and chlorophytes. 29 Background and Literature Review Phytoplankton Composition at 1m Depth (1992-1994) 7000 Cy anobacteria Chlorophy tes Chry sophy tes Other Total 6000 5000 l m / s 4000 l l e 3000 c 2000 1000 0 Mar92 Jun92 Oct92 Dec92 Mar93 Jun93 Date Sep93 Nov 93 Apr94 Jul94 Figure 7.1.2: Phytoplankton composition at 1m depth (1992-1994). Latchinian et al. (date unspecified) found that areas in the northwest and central locations were dominated by cyanobacteria, with Anabaena and Microcystis being the dominant representatives of this group. Conversely, chlorophytes were dominant in the vicinity of the discharge zones of the rivers along the southern and central shorelines of the lake and towards the Motatán. This finding is in accordance with Konopka et al. (1978), who observed that Anabaena and Microcystis tend to occupy the epilimnion layers in large eutrophic lakes. These microorganisms are particularly efficient at fixing organic nitrogen and N2, which allow them to maintain their growth even when concentrations of dissolved nitrogen diminish in the lake. According to Garlick et al. (1977) the exceptional capacity of realising oxic, as well as anoxic photosynthesis, allows these organisms to exist in conditions that are uninhabitable by other alga. For the period before 1992 Anabeana dominated over Microcystis, a situation that was reversed after 1992, probably as a result of the increase in available nutrients, principally Nitrogen, as a result of the progressive increase in anthropogenic contributions. Observations by Parra Pardi (1996) support this theory. During the NEDECO field campaign of 1992 and 1993 (NEDECO 1993), the lake waters were dominated by green algae (chlorophytes), with associated changes due to the wet and dry seasons. The green algae measured in the lake were species characteristic of freshwater systems. 30 Background and Literature Review The most common genus measured was Selenastrum, however during the dry season Scenedesmus predominated in the north and northeastern part of the lake. NEDECO (1993) attribute the changing phytoplankton composition within Lake Maracaibo to the increase in salinity since the deepening of the navigation channel. Latchinian et al. (date unspecified) postulated that the change in composition could be attributed to a combination of increasing lake salinisation, and an increase in available nutrients, primarily nitrogen, arising from increased river and anthropogenic contributions. Recently, Gardner et al. (1998) examined nitrogen cycling rates and light effects on the algae growth. They found that the nutrient cycles are dominated by biological control, with nitrogen being the most limiting nutrient for phytoplankton growth. Thus, it is likely that the increase in anthrogenic nutrient inputs has contributed to the change in phytoplankton composition, as suggested by Latchinian et al. (date unspecified). 31 Background and Literature Review 8 Predictive Water Quality Modelling The need for predictive water quality modelling has arisen largely as a result of increased eutrophication of lakes throughout the world. Eutrophication describes the biological reaction of aquatic systems to nutrient enrichment, the eventual consequence of which is the development of primary production to nuisance proportions (Marsden 1989). Eutrophication can arise under natural conditions, but is more commonly recognised as a consequence of anthropogenic activities (Moss 1988), and frequently results in large high-density growths of planktonic algae which significantly reduce the financial and aesthetic value of lakes and reservoirs (Marsden 1989). Eutrophication also results in a reduction in species diversity in waterbodies at all trophic levels. The frequent dominance of cyanobacteria in eutrophic waters is an additional concern due to the common ability of these organisms to produce toxins (Codd 2000). As a corollary, the fundamental aim in restoring eutrophic waterbodies, such as Lake Maracaibo, should be the reduction of total phytoplankton biomass and a shift in species dominance away from cyanophytes. Understanding mechanisms that control lake eutrophication is an issue of both theoretical and applied concern. Initially, it was thought that a reduction in external loading of nutrients would be sufficient to curb lake productivity. However, in many lakes, the remineralisation of deposits accumulated during eutrophication produces significant internal loading (Ostrovsky et al. 1996). The release of nutrients from the sediments, and their subsequent transport upward can play a decisive role in lake trophic status (Vollenweider 1968, Nürnberg 1984). On the other hand, in lakes characterised by stable stratification, the metalimnetic barrier restricts direct vertical exchange of material. Therefore, in stratified eutrophic lakes, primary producers inhabiting the epilimnion are often restricted by the lack of dissolved mineral biogenic elements (e.g. phosphorus), while in the hypolimnion, concentrations of these elements can be much higher due to their release from bottom sediments (especially in anaerobic conditions) and to decomposition of settling organic matter (Wetzel 1983). Thus, any physical mechanism responsible for significant cross-metalimnetic upward mass transport will be important in controlling lake productivity (Ostrovsky et al. 1996) 32 Background and Literature Review Restoration programs must be based on a theoretical framework that will allow a prediction of change. Numerical models can provide this necessary theoretical framework. They can also assist in understanding the origins and mechanisms of water quality problems and can be important tools in the development of management strategies for lakes and reservoirs (OECD 1982). Multitudes of models are available for simulating water quality problems, ranging from simple empirically derived equations, to multi-dimensional ecosystem models (Orlob 1984). 8.1 Input-Output Models Deterioration of recreational and drinking water quality arising from cultural eutrophication prompted the development of a number of quantitative relationships between nutrient loading to lakes and in-lake concentrations (Vollenweider 1968, Vollenweider 1975, Dillon and Rigler 1974), and indices of plant biomass and availability of nutrients to primary producers (Sakamato 1966). In most temperate lakes of North America and Europe phosphorus (P) has been implicated as the limiting nutrient for algal growth (Schindler 1977, Nürnberg 1984, Marsden 1989, Hamilton & Schladow 1997). Hence a cornerstone strategy for the restoration of eutrophic lakes has been the reduction in the external inputs of P (Edmondson 1970) which are commonly derived from controllable point sources (Vollenweider 1968). The most common modelling approach is exemplified by the empirical derivation of input-output (mass balance) models, based on correlations between P loading and indicators of trophic status (Hamilton & Schladow 1997, Marsden 1989, Nürnberg 1984). The effect of P-loading reduction strategies on total phosphorous can be estimated with empirical loading models (Vollenweider 1968, 1975; Dillon & Rigler 1974) from the external P-load, hydraulic residence time, average depth, and estimates of P lost to the sediments. These models provide long-term, steady-state predictions after internal P alterations have reached equilibrium with an assumed constant external loading. However, they neglect the response time to attain steady-state concentrations after an external loading reduction (Robson et. al. in progress). This response time can be highly variable due to differences in internal P-loading, the magnitude of the former loading, and the hydraulic retention time of the reservoir (Jeppesen et al. 1991). Additionally, though reductions in external P-loading may reduce concentrations of the limiting nutrient, lower algal biomass may not necessarily result because of disruptions via food web interactions (Jeppesen et al. 33 Background and Literature Review 1991). Input-Output models make a number of simplifying assumptions, which make them inappropriate in modelling the long-term changes in phytoplankton (chlorophyll-a) abundance in Lake Maracaibo, resulting from the proposed reduction in river-borne nutrient loading and closure of the navigation channel. The empirical P-loading models discussed above assume that: • The system is at steady state; • The lake is well mixed; • The lake is P limited; • The in-lake P concentrations are < 0.2 mg/L (OECD 1982) • The external loading rate remains constant (i.e.- no intra or inter-annual variations in loading); • The outflow concentrations are equal to the average in-lake concentration; The Lake Maracaibo system fails to meet each and every one of these assumptions. To highlight the inadequacy of the use of input-output models in assessing the long-term changes in water quality it is worth noting three salient points regarding Lake Maracaibo: 1. The system rarely fully mixes due to the almost permanent salt stratification. 2. Phosphorous concentrations in the surface waters are approximately 0.4 mg/L (ICLAM database). 3. Nitrogen and light limit the system (Gardner, et al. 1998). For these reasons it is obvious that a simple input-output model will have no utility in evaluating and analysing the management options under consideration in this study, and that a more complex process based model is required. 34 Background and Literature Review 8.2 Process Based Modelling/Ecological Models The development and refinement of deterministic ecological models has furthered the understanding of biological and chemical processes in aquatic ecosystems, and have been used as a vehicle to synthesise and evaluate knowledge of natural waters. An important objective of ecological models is the ability to predict the outcome of a series of possible management alternatives for a given aquatic system. The majority of deterministic phytoplankton models are ordered frameworks of mechanistic or semi-empirical sub-models. Each sub-model describes a particular process, whether it be lake circulation, nutrient limited growth rate of phytoplankton, or the respiration rate of phytoplankton. The success or indeed failure of such models is dependent not only upon the inclusion of all important sub-models, but also the way in which given processes are mathematically described (Scavia and Robertson 1979). Effects of nutrient reduction strategies on the water quality of a lake or reservoir must account for both external loading (rivers and precipitation), internal loading (release from sediments) and fate (uptake by phytoplankton or settling with burial to sediments). Thus, it is imperative that a water quality model capable of accurately simulating the seasonal and inter-annual hydrodynamics, biogeochemistry and ecology of lakes and reservoirs be utilised as a tool to understand long-term process dynamics, and to formulate and/or evaluate optimal management strategies (Robson et. al. in progress). 35 Model Selection 9 Model Selection The complex biogeochemical processes that control the nutrient cycles within Lake Maracaibo present particular challenges to any attempt at modelling the ecology of the lake. Initial investigations show a strong link between peak river inflows and high nutrient concentrations in the hypolimnion. The short lag between the peak inflows and peak concentrations suggests high sedimentation rates, which must be captured by the numerical models. The sedimentation of organic matter from the epilimnion to the hypolimnion, and the subsequent accumulation of the unoxidized fraction of this material on the floor of the lake, may be an important sink for nutrients within the Lake Maracaibo. In addition to external loadings described above, there are a number of important internal loadings within the lake. These processes are controlled by vertical fluxes of salt, oxygen, and nutrients through the water column. Due to the density stratification, these fluxes are limited by entrainment across the halocline. It is important that any hydrodynamic model capture these entrainment processes. As well as determining the volume of the saline hypolimnion, the rate of entrainment across the halocline will determine the dissolved oxygen profile, and the rates at which nutrients are recycled from the nutrient rich waters of the hypolimnion to the photic waters of the epilimnion. The release of nutrients from the sediments underlying the anoxic saline hypolimnion may be an important internal loading, and any model must include such fluxes. Importantly, these fluxes are dependent on the dissolved oxygen concentration, reiterating the interdependence of many of these processes, and the importance of capturing vertical fluxes through the water column. The observed persistent low concentrations of dissolved oxygen and nitrates in the saline hypolimnion suggests that denitrification may be a sink for nitrogen in Lake Maracaibo. The horizontal spatial variability of nutrients and chlorophyll-a over the surface of the lake is dependent on many factors, including the location of nutrient point sources, the primary counterclockwise currents, local wind-driven currents and the secondary radial currents. It would be unlikely to capture these factors in sufficient detail to allow for direct comparison between model simulations and field data over such a large domain without introducing some horizontal spatial 36 Model Selection averaging. However, from a review of the literature it appears that the nutrient cycles within Lake Maracaibo are dominated by vertical fluxes. Thus a one-dimensional hydrodynamic/ecological model such as DYRESM-CAEDYM provides the best possible methodology of simulating the horizontally averaged fluxes and concentrations. 9.1 DYRESM-CAEDYM DYRESM-CAEDYM is a one-dimensional coupled hydrodynamic and water quality model for lakes and reservoirs. It is used to predict the variation of hydrodynamics and ecology with depth and time. The hydrodynamic component (DYRESM) is a process-based model that simulates the vertical distribution of temperature and salinity in lakes and reservoirs through a series of Lagrangian layers. The success of the model applications can be attributed to the high level of process description, which means that the model is free from calibration, and to the Lagrangian layer structure, which optimises computation times and negates any requirement to solve vertical flow velocities. The water quality component (CAEDYM), however, is empirically based and as such has many parameters requiring calibration. 9.2 CAEDYM CAEDYM (Computational Aquatic Ecosystem DYnamics Model) is a self-contained ecological model, which has been designed to link to a suite of hydrodynamic models. The model has been set up largely for assessments of eutrophication, being of the ‘N-P-Z’ (nutrients-phytoplanktonzooplankton) model format, but it also contains process descriptions of oxygen dynamics, nutrient cycling, and several other state variables. CAEDYM is a substantial advance on the traditional N-P-Z models, in that it serves not only as a general ecosystem model (e.g. resolving various biogeochemical processes), but also as a species- or group-specific model (e.g. resolving various phytoplankton species) (Hamilton and Herzfeld 1999). CAEDYM has been modified specifically to allow easy linkage to the hydrodynamic models used within the Centre for Water Research, University of Western Australia. One of the prime objectives of the hydrodynamic links is to allow user flexibility in resolving the appropriate temporal and spatial scales of interest. It has been coupled previously with process representations of biogeochemical processes (DYRESM-WQ; Hamilton and Schladow, 1997; Schladow and Hamilton, 1997). However, the CAEDYM model represents a substantial advance in process representation and flexibility of state variable choices over the traditional DYRESM37 Model Selection WQ model, which was used in previous modelling of Lake Maracaibo, or other N-P-Z models. By simulating phytoplankton at the species level, CAEDYM represents the ideal model to aid in better understanding the ecology of Lake Maracaibo. In addition, coupling CAEDYM with the one-dimensional hydrodynamic model (DYRESM) allows for analysis and assessment of the effect of the proposed management options on annual and long-term (>10 years) variations in water quality. The major biogeochemical state variables included in CAEDYM are illustrated in Figure 9.2.1, with differentiation of ecological state variables to species level shown in Figure 9.2.2. Figure 9.2.1: Major State Variables included in the CAEDYM model 38 Model Selection Species Group Phytoplankton Zooplankton Fish Macroalgae Invertebrates Seagrass Jellyfish 1 Dinoflagellates 44-100 mm Hardyheads Gracilliara Bivalves Halophila ovalis Phyllorhiza punctata 2 Freshwater Cyanobacteria 100-300 mm Perth Herring Cystosira Polychaetes 3 Marine Cyanobacteria > 300 mm (Gladioferans) Yellow Tail Trumpeter Chaetomorpha Crustacean Grazers 4 Chlorophytes > 300 mm (Sulcanus) Black Bream Ulva 5 Cryptophytes > 300 mm (Acartiura) Sea Mullet 6 Marine Diatoms 7 Freshwater Diatoms Figure 9.2.2: Species representation within the major biological state variables in CAEDYM The model offers substantial flexibility to the user in the choice of state variables. The objective in setting up the model has been to offer the user a very wide choice of options, which can then be reduced according to the specific application. This approach is considered to provide greater flexibility to the user; input files of calibration parameters can be adjusted rather than modifying the source code (Hamilton and Herzfeld 1999). In order to predict the effect on phytoplankton composition and abundance resulting from closure of the navigation channel or a reduction in nutrient loading, and to further the understanding of phytoplankton dynamics within Lake Maracaibo, it is necessary to discuss the way CAEDYM represents phytoplankton growth. 9.2.1 REPRESENTATION OF PHYTOPLANKTON GROWTH WITHIN CAEDYM There is general agreement that algal production and photosynthesis is dependent on the effects of light, nutrients and temperature (Lehman et al. 1975). However, the interactions between these effects vary between models. Chen and Orlob (1975) assumed that the limiting factors are independent of each other and are of equal weight, and thus applied a multiplicative relationship between light, temperature and nutrient limitation functions. CAEDYM’s representation of phytoplankton growth limitation differs from Chen and Orlob, in 39 Model Selection that it assumes that more than one limitation value (e.g. for light or nutrients) is not limiting at the same time. In this approach, the minimum of two or more of the limitations is used (e.g. Hamilton and Schladow 1997). P = µ max ↔f (T ) ↔min[ f ( I ), f (nutrients )] ↔[phyt ] (Equation 9.2.1) The maximum growth rate (µmax) is thus modified by the temperature function f(T) and by the minimum of the light limitation function f(I), and the nutrient limitation functions, f(nutrients), which can include nitrogen, carbon, phosphorus and silica as is appropriate, and then multiplied by the phytoplankton concentration, [phyt], to get the new phytoplankton biomass P. Most recent models use a single multiplicative function or use only the minimum limiting function (CWR book unpublished). Gardner et al. (1998) suggested that nitrogen and light are the factors limiting phytoplankton growth within Lake Maracaibo. Light Limitation A wide range of light limitation models have been used in the past (e.g. Jørgensen 1983). Most include the reduction of light with depth and a Michaelis-Menten expression to describe the relationship between photosynthesis and irradiance (Robson et. al. in progress). CAEDYM uses the simple model of Scavia and Park (1976), in which the fractional limitation of the growth rate by light, f(I), is described as: f (I ) = 1 − exp (Equation 9.2.2) I √ IK √ ↵ where, I is irradiance and IK is a light saturation parameter. 40 Model Selection Nitrogen Limitation CAEDYM uses a simple Michaelis-Menten term to represent nitrogen limitation: f (N ) = NH 4 + NO3 NH 4 + NO3 + K N (Equation 9.2.3) where KN is the half saturation constant for nitrogen, NO3 is the nitrate concentration and NH4 is the ammonium concentration.. The value of KN can be set to be negligible or zero for the case of nitrogen fixing cyanobacteria (Hamilton and Herzfeld 1999). 41 Model Selection 9.3 DYRESM: REVIEWING THE 1-D ASSUMPTION The earliest limnological studies of thermal structure of lakes (Hutchinson 1957) recognized that during the period of greatest stratification, especially with the formation of a distinct thermocline, there was comparatively little variation in temperature over a horizontal plane parallel to the water surface. Strong stratification inhibits vertical motions and reduces turbulence to intermittent bursts. Thus, for strongly stratified lakes, such as Lake Maracaibo, the transverse and longitudinal directions play a secondary role, and only variations in the vertical enter the first order mass, momentum, and energy balances (Imberger & Patterson 1981). DYRESM is based on the assumption of one-dimensionality, that is, variations in the lateral directions are small when compared with variations in the vertical. The assumption is based on observations that the density stratification usually found in lakes inhibits vertical motions, while horizontal variations in density are quickly relaxed by horizontal advection and convection. Horizontal exchanges generated by weak temperature gradients are communicated over several kilometres on time scales of less than a day, suggesting that a one-dimensional model such as DYRESM is applicable for simulations over daily time scales. It is this assumption that gives rise to the model’s layer construction, in which the lake or reservoir is represented as a series of horizontal layers. There is no lateral or longitudinal variation in the layers, and vertical profiles are obtained from the property values of each layer. The layers represented in DYRESM are of differing thickness; as inflows and outflows enter or leave the lake, the affected layers expand or contract, and those above move up or down to accommodate the volume change. The vertical movement of the layers is accompanied by a corresponding change in layer thickness as the area occupied by each layer varies according to its vertical position. Mixing is modelled by an amalgamation of adjacent layers, with individual layer thickness dynamically set by the model to ensure an adequate resolution is obtained for each process A criterion for assessing the validity of the one-dimensional assumption, and thus the applicability of using DYRESM to model a certain lake or reservoir, was first suggested by Water Resources Engineers, Inc. (1969), who proposed the use of the densimetric Froude number. 42 Model Selection The use of the densimetric Froude number as a guide to the applicability of the one dimensional approximation was superseded by Patterson et al. (1984) who developed a set of criteria based on the Wedderburn number, the inflow and outflow Froude number and the Rossby radius. The Wedderburn number represents the ratio of the buoyancy force acting on the surface layer to the shear stress due to the wind, and is defined as: (Equation 9.3.1) g'h2 W = 2 u* L Where L is the fetch length, h is the thickness of the surface mixed layer, u* is the surface wind shear velocity, and g’ is the modified gravitational acceleration across the base of the surface mixed layer defined as g'= g (Equation 9.3.2) ∆ρ ρ0 where ∆ρ is the density difference between the bottom of the surface layer and the hypolimnion, and ρ 0 is the density of the hypolimnion. When the Wedderburn number is W >> 1 the buoyancy force is greater than applied wind stress, resulting in only a small tilting of the isopycnals and negligible horizontal variations. For this case the processes are essentially one-dimensional and the algorithms of DYRESM are valid. When W ~ 1, the forces due to the applied wind stress are similar to buoyancy forces. This is the critical condition and can be described physically as the point where the base of the surface layer tilts toward the surface at the upwind end, defined as upwelling. W << 1 results in broad upwelling at the upwind end of the lake with vertical motions occurring on smaller time scales than horizontal advection. For the two latter cases the assumption of one-dimensionality is not valid. The Wedderburn number appears to impose the restriction that DYRESM only be applicable to “small to medium size” lakes and reservoirs. As, if L >> h then W << 1 unless the applied wind stress is always low. When the wind stress is large, small lakes may also violate W >> 1. However, when this occurs the resulting mixing is large and vertical density gradients are smoothed. Although the upwelling processes are not explicitly modelled by DYRESM the mixed layer algorithm will predict large vertical deepening for this case, resulting in DYRESM 43 Model Selection accurately predicting the same resulting density profile. Figure 9.3.1 illustrates schematically that the same wind stress applied to a large and small lake will result in a large amount of mixing due to upwelling in the small lake, while only causing small mixing due to upwelling in the large lake. Figure 9.3.1: Upwelling occurring in a small and large lake with the same applied wind stress. For large reservoirs the angle of deflection of the thermocline necessary to cause upwelling is small compared with that for small reservoirs. The flow field resulting from upwelling is highly complicated, with horizontal motions becoming highly important. DYRESM seems to be unsuitable in this situation, as the processes necessary to describe this flow field are not considered in the model. However, for lakes and reservoirs of “small to medium size” the end result of upwelling is to thoroughly mix the water body, and although DYRESM does not explicitly model the upwelling process, its mixed layer algorithm predicts a very similar mixing efficiency. Thus, DYRESM models reservoirs and lakes of “small to medium size” accurately, even if the Wedderburn number criteria is violated. This situation is complicated in most large lakes and reservoirs by the fact that upwelling may not result in complete mixing. The inflow Froude number may be written as Finf low = (Equation 9.3.3) U g' H Where U is the inflow velocity, g’ defined similarly to (Equation 9.3.2) with ∆ρ now being the difference between the inflow water and the reservoir water and H is the reservoir depth. The one-dimensional assumption is valid when Finflow < 1. Physically this means that the buoyancy 44 Model Selection force will dampen any non-vertical motions in the density structure resulting from inflow disturbances. Similarly for the outflow Froude number defined as Foutflow (Equation 9.3.4) Q = 1/ 2 5 / 2 g' H where Q is the outflow discharge and g’ is the reduced gravity between the surface and the bottom water. Again, the one-dimensional assumption is valid if Foutflow < 1. The Rossby radius is defined as (Equation 9.3.5) R= g' H f Where g’ is the effective reduced gravity: g'= (Equation 9.3.6) ∆ρ g ρ0 H = mean Lake depth = 25.9m (Battelle 1974) f = Coriolis parameter = 2Ωsin(latitude) ; latitude = 10° Ω = Earth’s angular velocity = 7.292 × 10-5 rads-1 For one dimensionality R > 1 More recently, the validity of the one-dimensional assumption has been tested against the Lake number (LN) (Imberger & Patterson 1990). The extent of hypolimnetic mixing in lakes and reservoirs is strongly dependent upon the balance between the forces supplying energy to mix the system and the forcing acting to inhibit mixing. Forces active in mixing primarily include wind, cooling, inflow, outflow, and artificial destratification devices; whereas, density stratification is the primary force acting to inhibit mixing. If the forces acting to mix the water column are insufficient, as is often the case during stratification, bottom waters becomes isolated from surface mixing. In many systems, this temporal isolation is associated with hypolimnetic oxygen depletion, and deterioration in water quality (Robertson and Imberger 1994) 45 Model Selection Several parameters have been used to try to describe changes in the strength or intensity of stratification, i.e., the dynamic stability of the system. The dimensionless Lake Number, LN, formalised and discussed by Imberger and Patterson (1990) is one parameter that represents the dynamic stability and extent of deep mixing within a lake or reservoir. LN is a quantitative index of the dynamic stability of the water column; it is defined as the ratio of the moments about the water body’s centre of volume, of the stabilising force of gravity (resulting from the density stratification) to the destabilising forces from wind, cooling, inflow, outflow, and artificial destratification devices (Robertson and Imberger 1994). If wind is assumed to be the dominate mixing force, i.e., inflow, outflow, and any artificial destratification devices have minimal destabilising force, LN can be defined by Equation 9.3.7 (similar to Imberger and Patterson 1990; Robertson et al. 1990). (Equation 9.3.7) S ( H − h2 ) LN = 2 t3 / 2 u* A ( H − hv ) where H is the maximum depth of the water body, h2 is the thermocline depth from the bottom of the lake, hv is height to the centre of volume of the lake, As is the lake’s surface area, St the Schmidt stability, and u* is the water friction velocity [cm s-1] due to wind stress approximated by the bulk aerodynamic formula in Equation 9.3.8. u* = 2 Where: U10 CD (Equation 9.3.8) Qa 2 ↔C D ↔U 10 Qm = Wind velocity at 10m above the water surface [cm s-1] = Drag Coefficient = 1.3× 10-3 [dimensionless] Qa/Qm = ratio between air and water densities = 1.2 × 10-3 [dimensionless] For LN » 1, stratification is strong and dominates the forces introduced by surface wind energy. Under these circumstances, the isopycnals are expected to be primarily horizontal. Little seiching of the seasonal thermocline and little turbulent mixing in the hypolimnion are expected. Above a value of 1.0 increases in LN represent very little difference in terms of mixing below the seasonal thermocline. For LN « 1, stratification is weak with respect to wind stress. Under these 46 Model Selection circumstances, the seasonal thermocline is expected to experience strong seiching and the hypolimnion is expected to experience extensive turbulent mixing due to internal shear (Imberger 2001). Table 9.3.1 summarises the non-dimensional parameters discussed above in order to validate the assumption of one-dimensionality and describe the dynamics of Lake Maracaibo. Table 9.3.1: Non-Dimensional Numbers for Lake Maracaibo (adapted from Laval et al. 2000a) Non-Dimensional Number Value [O (of the order of)] Interpretation Wedderburn number (W) O (10) Upwelling of the metalimnion induced by wind stress is unlikely Lake Number (N) O (10) Upwelling of the hypolimnion induced by wind stress is unlikely Burger Number (S) O (1) Suggests that rotation and stratification are of equal importance Based on the non-dimensional numbers presented above, it is appears valid to apply the assumption of one-dimensionality when modelling Lake Maracaibo. 47 Applying DYRESM to Lake Maracaibo 10 APPLYING DYRESM-CAEDYM TO LAKE MARACAIBO The one-dimensional coupled hydrodynamic-ecological model DYRESM-CAEDYM is a powerful tool, which can be used to investigate the interactions between physics, chemistry, and biology in aquatic ecosystems. The model uses a number of text files as input. A substantial quantity of data must be collected and compiled in order to prepare the text files (see Figure 10.1). For the majority of lakes, the vast amount of data required is not available in the form required. Various assumptions are often required in order to determine the physical and morphometric characteristics, and create continuous time-series of inflow quantity and quality, and meteorological data. The sources of the input data required to model Lake Maracaibo using DYRESM-CAEDYM, and the assumptions made in manipulating such data, are presented in the following sub-sections. Figure 10.1: Inputs Required for DYRESM-CAEDYM 48 Applying DYRESM to Lake Maracaibo 10.1 DYRESM SETUP All data used in preparing the DYRESM-CAEDYM files was obtained from papers, records, and other documentation relating to “The Integrated Study for the Environmental Remediation of Lake Maracaibo” or staff from within the Centre for Water Research who were involved in the aforementioned project. The initial set-up required by DYRESM-CAEDYM includes the following information: (a) physical and morphometric data of the lake basin, inflows and outflows, (c) water quality groups/processes and corresponding rate coefficients, and (c) initial profiles of temperature, salt and water quality variables. 10.1.1 PHYSICAL AND MORPHOMETRIC DATA The values of the physical characteristics of Lake Maracaibo used in DYRESM-CAEDYM include lake length and width at maximum water level, hypsographic data consisting of height, area and cumulative volume (Table 10.1.1), and geometric characteristics of major rivers. The hypsographic data were based on DMAHTC Sea Chart No 24481, Lago De Maracaibo. Table 10.1.1Surface Area and Volume of Lake Maracaibo as a Function of Depth Height from the bottom (m) 0 1 2 3 4 5 10 15 20 25 30 Area (x1000 m2) Volume (x1000 m3) % of Total Volume 725,760 1,614,240 3,541,000 4,402,000 5,076,000 5,736,000 6,926,000 8,585,000 10,230,000 11,220,000 11,956,320 0 945,400 2,559,600 6,531,100 11,270,100 16,676,100 48,331,100 87,108,600 134,146,100 187,771,100 245,711,900 0.0% 0.4% 1.0% 2.7% 4.6% 6.8% 19.7% 35.5% 54.6% 76.4% 100.0% Although there are thirty catchments draining into Lake Maracaibo, 81 percent of the total runoff is accounted for by just four major rivers, with the largest of these, Rio Catatumbo, contributing approximately 60 percent of the total inflow to the Lake. Since these rivers all flow into the south-western part of Lake Maracaibo, DYRESM-CAEDYM modelled the river inflow as a single source. The dynamics of the river inflow, including the entrainment of Lake water by 49 Applying DYRESM to Lake Maracaibo the inflow, is governed by the longitudinal slope of the streambed and by its cross-sectional shape. DYRESM assumes that the river valley is approximately triangular with some average streambed half-angle (defined as the angle of the side slope of the stream to the vertical). Slopes and half angles for this single inflow stream were based on bathymetry data for the Rio Catatumbo and are shown below in Table 10.1.2. Details of the inflow time series are given later in Section 10.2.2. Table 10.1.2: Stream inflow parameters Inflow Parameter Streambed Slope (degrees) Streambed half angle(degrees) Streambed Drag Coefficient River Inflow 0.001 85 0.016 Saline Underflow 0.005 85 0.016 The saline underflow from Maracaibo Strait (as discussed in section 8.1) was included in DYRESM-CAEDYM as a separate stream inflow and the streambed slope and half angle were based on the bathymetric data for the southern end of the Strait. The outflow through the Strait was not specified for modelling Lake Maracaibo, but rather a maximum water elevation specified and the outflow calculated by DYRESM-CAEDYM. The outflow calculated by DYRESM was compared with MIKE-3 outflow and demonstrated good agreement on a seasonal time-scale. In the absence of detailed water levels over the Lake it was assumed that variations in Lake water level could be neglected for the purposes of assessing long-term effects on water quality. 10.1.2 INITIAL PROFILE Initial profiles of salinity, temperature, dissolved oxygen, and nutrients were derived by linear interpolation of field profiles collected by ICLAM at station C11 on September 22nd 1998. 10.2 INPUT DATA Input data required by DYRESM-CAEDYM for the duration of each simulation includes: meteorological forcing, and inflow volumes and properties (e.g. temperature, salinity and water quality. 50
© Copyright 2025 Paperzz