This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright Author's personal copy Journal of Hydrology 369 (2009) 165–174 Contents lists available at ScienceDirect Journal of Hydrology journal homepage: www.elsevier.com/locate/jhydrol The influence of historical and potential future deforestation on the stream flow of the Amazon River – Land surface processes and atmospheric feedbacks Michael T. Coe a,*, Marcos H. Costa b, Britaldo S. Soares-Filho c a The Woods Hole Research Center, 149 Woods Hole Rd., Falmouth, MA 02540, USA The Federal University of Viçosa, Viçosa, MG, 36570-000, Brazil c The Federal University of Minas Gerais, Belo Horizonte, MG, Brazil b a r t i c l e i n f o Article history: Received 18 June 2008 Received in revised form 27 October 2008 Accepted 15 February 2009 This manuscript was handled by K. Georgakakos, Editor-in-Chief, with the assistance of Phillip Arkin, Associate Editor Keywords: Amazon Discharge Numerical models Deforestation s u m m a r y In this study, results from two sets of numerical simulations are evaluated and presented; one with the land surface model IBIS forced with prescribed climate and another with the fully coupled atmospheric general circulation and land surface model CCM3-IBIS. The results illustrate the influence of historical and potential future deforestation on local evapotranspiration and discharge of the Amazon River system with and without atmospheric feedbacks and clarify a few important points about the impact of deforestation on the Amazon River. In the absence of a continental scale precipitation change, large-scale deforestation can have a significant impact on large river systems and appears to have already done so in the Tocantins and Araguaia Rivers, where discharge has increased 25% with little change in precipitation. However, with extensive deforestation (e.g. >30% of the Amazon basin) atmospheric feedbacks, brought about by differences in the physical structure of the crops and pasture replacing natural vegetation, cause water balance changes of the same order of magnitude as the changes due to local land surface processes, but of opposite sign. Additionally, changes in the water balance caused by atmospheric feedbacks are not limited to those basins where deforestation has occurred but are spread unevenly throughout the entire Amazon by atmospheric circulation. As a result, changes to discharge and aquatic environments with future deforestation of the Amazon will likely be significant and a complex function of how much vegetation has been removed from that particular watershed and how much has been removed from the entire Amazon Basin. Ó 2009 Elsevier B.V. All rights reserved. Introduction The Amazon River system provides habitat for one of the world’s most diverse aquatic environments and is central to the existence of millions of people. It provides nearly all, domestic and commercial transportation in the region. For example, there are no roads connecting Manaus, an important industrial city of about 1.2 million people in the center of the Amazon, with the major population centers in northeastern and southern Brazil. The river also provides drinking water, livelihood, and protein in the form of fish to the majority of the population of the Amazon. Additionally, major infrastructure investments are planned for the Amazon basin, including hydroelectric dams, mining and industrial development, and waterways for barge traffic (Carvalho et al., 2001; Laurance et al., 2001). These social, ecological, and economic actors all depend on a stable Amazon River system. Despite the popular view that the Amazon contains a near limitless supply of water, extreme climate events such as droughts * Corresponding author. Tel.: +1 508 540 9900; fax: +1 508 540 9700. E-mail address: [email protected] (M.T. Coe). 0022-1694/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jhydrol.2009.02.043 and floods routinely disrupt communication, commerce, economics, health and ecology over wide areas of the watershed (Marengo et al., 2008a,b). With increasing population and infrastructure within the Amazon, future variability and changes to stream flow are likely to create more frequent and larger disruptions. Global economic and regional population and development pressures have resulted in high rates of deforestation in the Amazon River basin (Achard et al., 2002; Fearnside and Graça, 2006; Kaimowitz et al., 2004) and about 17% of the Amazon basin (excluding the Tocantins) has been deforested by 2007, mostly in the eastern and southern portion of the basin (Fearnside, 1993; INPE, 1999, 2000; Nepstad et al., 1999; Skole et al., 1994; Skole and Tucker, 1993) (Fig. 1). Cattle ranching is the single largest use of cleared land in the Amazon, covering about 75% of total deforested area (Faminow, 1998; Margulis, 2003) and herd size in the Amazon has doubled since 1996 from about 15 million to 30 million head (Nepstad et al., 2006; Simon and Garagorry, 2005). However, soybean production, primarily for export as animal feed to Europe and China (Nepstad et al., 2006), has become more important in the last decade and new land is now being converted directly from cerrado and forest for soybean production. Recent deforestation Author's personal copy 166 M.T. Coe et al. / Journal of Hydrology 369 (2009) 165–174 Fig. 1. Vegetation cover of the Amazon basin with three land covers classes, tropical evergreen forests (green), Cerrado (beige), and agriculture (yellow) shown for potential vegetation (CTL) as reconstructed by Ramankutty and Foley (1998), vegetation distribution of the year 2000 (MOD) as estimated by Eva et al. (2002), and two scenarios for the year 2050 by Soares-Filho et al. (2006) with strong governance of deforestation (GOV) and relatively weak governance (BAU). rates reflect the global demand for Brazilian beef and soybeans with about 22,000 km2 deforested each year between 2000 and 2004 (INPE, 2004). Economic conditions and recent trends suggest that high rates of Amazonian deforestation for cattle and soybean production will continue well into the future (Alencar et al., 2004; Carvalho et al., 2001; Kaimowitz et al., 2004; Laurance et al., 2001; Soares-Filho et al., 2004, 2006). The last 5 months of 2007 saw a record 7000 km2 of deforestation in the Amazon. The land surface of the Amazon is coupled to its rivers, streams and wetlands, through hydrological processes. Human land cover and land use changes influence the quantity of surface water resources by: Changing how incoming precipitation and radiation are partitioned among sensible and latent heat fluxes, runoff, and river discharge (Bonan et al., 2004; Costa and Foley, 1997; Li et al., 2007) and altering regional and continental scale precipitation patterns (Costa and Foley, 2000; Delire et al., 2001; Dickinson and Henderson-Sellers, 1988; Malhi et al., 2008b; Nobre et al., 1991). Changes in the water and energy balance work at a variety of time and space scales and the combined influences on the river discharge are complex. Observations at micro (<1 km2) or meso (100 s km2) spatial scales in the global tropics and extra-tropics indicate that deforestation reduces evapotranspiration and increases stream flow because of the reduced leaf area index, decreased root density and depth, and increased soil compaction (Bruijnzeel, 1990; Costa, 2005; Sahin and Hall, 1996; Scanlon et al., 2007). Two microscale, paired experiments in Amazonia confirm the results obtained outside of the Amazon. The first paired experiment compares two catchments of areas <0.01 km2 in Fazenda Vitória, eastern Amazonia (Moraes et al., 2006), and found that the ratio of runoff to precipitation increases from 3% in the forest catchment to 17% in the pasture catchment. The second paired experiment compares two 1.2 km2 catchments, Colosso and AçúMirim, in Central Amazonia (Trancoso, 2006), concluding that the runoff coefficient increases from 21% in the forest catchment to 43% in the pasture catchment. At the large scale, studies in the Tocantins and Araguaia Rivers of eastern Amazonia conclude that rapid land cover changes since 1960 are associated with about a 25% increase in the annual mean discharge despite no significant change in precipitation (Coe et al., 2008; Costa et al., 2003). Therefore, in eastern Amazonia land cover change appears to have already significantly reduced evapotranspiration and increased runoff and discharge. This reduction in evapotranspiration is a consequence of the land surface processes involved in the exchange of energy and water from the biosphere to the atmosphere. The pastures, with higher albedo, lower leaf area, lower roughness length and shallower roots, usually have a lower evapotranspiration rate than the forest, particularly during the dry season. Global and meso-scale climate model studies indicate, however, that once deforestation in the Amazon basin occurs on a very large scale (>100,000 km2), atmospheric feedbacks may significantly reduce regional precipitation. Replacement of forest with higher albedo, less water-demanding crops and pastures leads to reduced net surface radiation, decreased moisture convergence over the basin, decreased water recycling, and reduced precipitation (Costa and Foley, 2000; Dickinson and Henderson-Sellers, 1988; Malhi et al., 2008b; Nobre et al., 1991). Recent studies that evaluated the effects of partial deforestation on the climate of the region (Costa et al., 2007; Sampaio et al., 2007) found that significant changes in precipitation occur only after more than 40% of the Amazon basin is deforested. Author's personal copy M.T. Coe et al. / Journal of Hydrology 369 (2009) 165–174 These two processes lead to competing influences on the stream flow, with decreased local evapotranspiration consistent with increased discharge and reduced regional precipitation consistent with decreased discharge (Costa, 2005; D’Almeida et al., 2006, 2007). The combination of these two processes is therefore, likely to create complex and unexpected changes in stream flow that vary spatially as a function of local and non-local conditions. Brazilian law currently requires that 80% of any landholding in the Amazon biome must remain as undisturbed forest and that riparian buffer zones must be maintained from 30 to 500 m from a stream depending on stream size. Further, Brazil continues to set aside large expanses of the Amazon basin as protected reserves: 23 million hectares since 2004 (Campos and Nepstad, 2006). How laws are enforced, and protected areas augmented, in the future will have a significant effect on the magnitude and character of land use changes (Soares-Filho et al., 2006) and therefore on aquatic systems that support millions of people and vast biological diversity. More focused research on the influences of heterogeneous deforestation and differing management practices on the aquatic systems of the Amazon are needed to better understand the scale of changes that have occurred and are likely to occur, the relative importance of individual forest protected areas, and help guide future policy. In this study, regional land surface models and a coupled global climate and vegetation model are forced with historical potential vegetation, modern (year 2000) vegetation, and land cover scenarios for the year 2050 to address the complex ways in which deforestation, through local and non-local feedbacks, can influence the conversion of precipitation to runoff and discharge throughout the Amazon River. The focus of this study is on the response of discharge within individual watersheds because the interaction of local and non-local responses to deforestation can lead to consequences for river discharge that an analysis of precipitation alone cannot clarify and because of the obvious environmental and social importance of maintaining the integrity of the Amazon River system. Methodology Two sets of simulations are made; one with an offline land surface model and another with a fully coupled atmospheric general circulation and land surface model. The land surface model IBIS (Kucharik et al., 2000) is used with the river transport model THMB (Coe et al., 2007) to evaluate the influence of historical and potential future deforestation on local evapotranspiration and the discharge of the Amazon River system in the absence of atmospheric feedbacks to precipitation. The fully coupled CCM3/ IBIS global climate and land surface model (Delire et al., 2002, 2004) is used with THMB to evaluate the response of the discharge to the combined local evapotranspiration and regional precipitation feedbacks that result from deforestation. This combination of offline land surface and fully coupled global climate models helps clarify the scale of the individual local and non-local processes and how in combination they affect the river system. A suite of land cover scenarios is used: potential vegetation (natural vegetation with no anthropogenic change (Ramankutty and Foley, 1998) the vegetation distribution of the year 2000 (Eva et al., 2002), and two deforestation simulations for the year 2050, one with strict governance of deforestation and one with a continuation of business as usual deforestation practices (SoaresFilho et al., 2006). This suite of scenarios is useful for: (1) Understanding the location and scale of impact deforestation has already had on the Amazon River; observations alone cannot always clearly provide this understanding because of the difficulties in obtaining long data records and because of the variability and error 167 inherent in that data; (2) Understanding the importance of continued and future efforts to reduce deforestation on the local and regional health of the Amazon River; continued efforts to reduce deforestation can be strengthened by knowledge of the scale of disruption that can be avoided; (3) Beginning to quantify the importance of individual protected areas in maintaining the integrity of the Amazon; scenarios can provide more information on the strategic importance of protected areas in reducing regional deforestation rates and hence may be useful in informing policy (Soares Filho et al., 2008). Model descriptions All models used in this study have been extensively calibrated and validated for the Amazon River and are thoroughly described in documents listed in the sections below. Therefore, only a brief description of each model is provided here. IBIS terrestrial ecosystem model IBIS is a physically-based model that integrates a variety of terrestrial ecosystem processes within a single, mechanistic model to simultaneously calculate a wide range of processes, including the land surface water and energy balances (Kucharik et al., 2000). The model has two vegetation canopies with an upper layer of trees and a lower layer of shrubs, grasses and crops, and 15 types of natural vegetation cover comprised of a combination of 12 plant functional types including woody and herbaceous plants. The soil module has six soil layers (with a total of 8-m depth in this study). The dynamics of soil volumetric water content are simulated for each layer. The soil water infiltration rate is based on the Green–Ampt formulation (Green and Ampt, 1911). The soil moisture simulation is based on Richards’ flow equation, where the soil moisture change in time and space is a function of soil hydraulic conductivity, soil water retention curve, plant water uptake, and upper and lower boundary conditions. Plant transpiration is a mechanistic process governed by stomatal physiology, in IBIS it is tightly coupled to photosynthesis through the Ball–Berry formulation (Ball et al., 1986). The plant root-water uptake is a function of atmospheric demand, soil physical properties, root distribution, and soil moisture profile (Kucharik et al., 2000; Li et al., 2005). IBIS explicitly simulates surface and sub-runoff on a grid cell basis as a function of the soil, vegetation, and climate characteristics. Horizontal runoff transport between grid cells is subsequently simulated by the THMB hydrological routing model (described in ‘‘THMB terrestrial hydrology model”). IBIS has been validated and applied to the Amazon (Botta et al., 2002; Coe et al., 2002, 2007; Delire and Foley, 1999; Foley et al., 2002). CCM3/IBIS coupled global climate land surface model The National Center for Atmospheric Research (NCAR) Community Climate Model version 3 (CCM3) is an atmospheric general circulation model with spectral representation of the horizontal fields. It simulates the large-scale physics (radiative transfer, hydrologic cycle, cloud development, thermodynamics) and dynamics of the atmosphere. In this study, we operate the model at a spectral resolution of T42 (2.81° 2.81° latitude/longitude grid), 18 levels in the vertical, and a 15-min time step. The oceans are represented by monthly averaged fixed sea-surface temperatures and serve as boundary conditions for the atmosphere. The water, energy and carbon cycle of this version of the model has been globally validated (Delire et al., 2002) and validated for the Amazon by (Senna et al., submitted for publication). Simulated precipitation was compared (Fig. 2) against eight different precipitation databases (Senna et al., submitted for publication), including three climatological surface rain gauge datasets – CRU (Climatic Research Unit (New et al., 1999)), LW (Legates and Author's personal copy 168 M.T. Coe et al. / Journal of Hydrology 369 (2009) 165–174 Fig. 2. Comparison of mean annual precipitation averaged over the area in gray in thumbnail of South America. Precipitation is simulated by CCM3-IBIS and reported by eight different data sources: CRU (Climatic Research Unit (New et al., 1999)), (Legates and Willmott, 1990), CMAP (CPC Merged Analysis of Precipitation (Xie and Arkin, 1997)), GPCP (Global Precipitation Climatology Project (Huffman et al., 1997)), TRMM (Tropical Rainfall Measuring Mission (Kummerow et al., 1998)), NCEP/NCAR (Kalnay et al., 1996), and ERA-40 (Uppala et al., 2005). All available data in the time series were used to describe the precipitation climatology. Willmott, 1990), and LC (Leemans and Cramer, 1990); three that blend remote sensing data with surface rain gauges – CMAP (CPC Merged Analysis of Precipitation (Xie and Arkin, 1997)), GPCP (Global Precipitation Climatology Project (Huffman et al., 1997)), and TRMM (Tropical Rainfall Measuring Mission (Kummerow et al., 1998)); and two reanalysis datasets – NCEP/NCAR (Kalnay et al., 1996) and ERA-40 (Uppala et al., 2005). All available data in the time series were used to describe the precipitation climatology. The use of a large number of precipitation datasets is important because regional precipitation estimates in Amazonia vary considerably (Costa and Foley, 1998). Observed annual mean precipitation estimates vary considerably; from 4.98 to 6.70 mm/day. The CCM3-IBIS simulated estimate (6.20 mm/day) is in the middle of the precipitation dataset range and is within 5% of the ERA-40, Leemans and Cramer, Legates and Willmott and TRMM datasets. The largest difference is 24.5% from the CRU dataset, in large part due to the negative precipitation bias in the CRU dataset in western Amazonia discussed in ‘‘Offline IBIS and THMB simulations”. The simulated seasonal amplitude of the precipitation for the Amazon is within the amplitude of the datasets although one month in advance (Fig. 2). THMB terrestrial hydrology model The terrestrial hydrological model with biogeochemistry (THMB) is forced by climate data and surface runoff and sub-surface drainage provided by IBIS to simulate the water balance of the Amazon River system. THMB is a distributed grid model at 5-min horizontal resolution that has been applied in large to moderate scale watersheds throughout the world, including North America, Africa, and India (Coe, 1998, 2000; Coe and Foley, 2001; Donner et al., 2002; Li et al., 2005, 2007; Shankar et al., 2004; Suprit and Shankar, 2007). The version used here has been developed, calibrated, and validated specifically for the Amazon and Tocantins River basins, as part of a NASA LBA-ECO project (Coe et al., 2007). THMB uses prescribed river paths and floodplain morphology, derived from the SRTM digital elevation model, linked to a set of linear reservoir equations describing the change with time of the river and floodplain reservoirs. Mass in the river and floodplain reservoirs is explicitly conserved. Variation of the total water within the stream at any point in THMB is the sum of the land surface run- off, sub-surface drainage, precipitation and evaporation over the surface waters, and the flux of water from the upstream grid cells and to the downstream cell. The flux to downstream cells (discharge) is based on the volume of water in the river reservoir and river geomorphic characteristics, such as slope and hydraulic radius. The inundation of the river floodplain is a function of the flux of water from the stream channel to the floodplain, the vertical water balance, and the geomorphic characteristics of the river and floodplain. The equations are solved with a 1-h time step. The results of THMB are spatially explicit representations of the (in-channel) river volume, stage and discharge, and the (out-of-channel) floodplain volume, stage, and inundated area at the temporal resolution of the input data (1 h to 1 month) and spatial resolution of the topographic data (5-min in this case). Experimental designs Offline IBIS and THMB simulations Four simulations were made with IBIS and THMB forced with prescribed identical climate and vegetation representing the following (Fig. 1): historical (CTL), modern (MOD), year 2050 with strict governance of deforestation (GOV), and year 2050 with a business-as-usual scenario of deforestation (BAU). To generate mean monthly surface and sub-surface runoff for THMB first, two simulations were completed using IBIS with identical CRU-ts2 observed climate (Mitchell et al., 2005) forcing for the period January 1915–December 2000 at ½-degree horizontal resolution. The two simulations had differing vegetation cover, one with the potential vegetation (IBIS-POT) as depicted by (Ramankutty and Foley, 1998) and another in which all vegetation in the Amazon is replaced by C4 grass (IBIS-GRASS). Four simulations were completed with THMB using the surface and sub-surface runoff data from IBIS-POT and IBIS-GRASS interpolated to the 5-min resolution of THMB: (1) a simulation representing no anthropogenic vegetation changes (CTL); (2) a modern simulation (MOD) using the deforestation classification of (Eva et al., 2002) to represent the forest distribution as of 2000; (3) a year 2050 simulation with deforestation simulated under the strict governance scenario (GOV), in which moderate deforestation takes place; and (4) a year 2050 simulation with deforestation occurring assuming a ‘‘business-as-usual” scenario (BAU), in which wide- Author's personal copy M.T. Coe et al. / Journal of Hydrology 369 (2009) 165–174 spread deforestation takes place (Fig. 1). The land cover maps are at 1 km2 horizontal resolution and the total surface runoff (Rt) and sub-surface drainage (Dt) input to each 5-min THMB grid cell is determined as the sum of IBIS-POT and IBIS-GRASS over the undisturbed and disturbed fractions of the grid cell respectively. Rt ¼ F f Rp þ ð1 F f Þ Rg Dt ¼ F f Dp þ ð1 F f Þ Dg Ff is the forested fraction (0–1) in each of the land cover scenarios, Rp (Dp) is the surface (sub-surface) runoff from IBIS-POT and Rg (Dg) is the surface (sub-surface) runoff from IBIS-GRASS. Therefore in CTL where Ff = 1 the surface and sub-surface runoff come entirely from the IBIS-POT simulation, while in the MOD, GOV, and BAU simulations the runoff input to each THMB grid cell is some mix of both IBIS-POT and IBIS-GRASS. Linear mixing of simulated results to create individual scenarios is appropriate because it closely approximates the linear averaging of land cover types that occurs within each grid cell of IBIS. THMB was run for the period 1915–2000 with a 1-h time-step, and with the monthly mean climate data linearly interpolated to 1h for each of the four scenarios. The first 2 years of each scenario were considered model spin-up and were discarded. The differences in simulated discharge of MOD, GOV and BAU from CTL quantify the sensitivity of the surface hydrology to prescribed land cover changes. The CRU-ts2 dataset significantly underestimates the precipitation in the western portions of the Amazon outside of Brazil, as did the CRU05 dataset previously used by (Coe et al., 2007). The bias is due most likely to a lack of data and the interpolation technique used in the creation of the dataset (Coe et al., 2007). The bias is reduced in this study as by (Coe et al., 2007) by applying a discharge bias correction to the IBIS simulated surface and sub-surface runoff for the four tributaries affected by the bias and equal to the amount of discrepancy between simulated and observed annual mean discharge: Japurá-24%, Negro-16%, Solimões-49%, and Madeira-32% before (and including) December 1984 and Japurá-52%, Negro24%, Solimões-66%, and Madeira-50% after December 1984. The large change after 1984 is most likely a result of a change in the number of precipitation gauge stations reporting data in the 1980s and used in creating the CRU products (New et al., 2000). This constant correction was applied to all grid cells upstream of the gauge station nearest the Brazilian border for all months in the surface and sub-surface runoff files. The corrected runoff was used as input data to provide the estimates of discharge and flooding presented in this study. Coupled CCM3/IBIS simulations A second series of simulations were made, with the National Center for Atmospheric Research Community Climate Model-3 coupled to the IBIS land surface model. The simulations correspond to the CTL, GOV, and BAU simulations described above and were created in the following way. Twenty-year simulations were made with the potential vegetation and year 2050 governance and business-as-usual simulated deforestation maps (Fig. 1) prescribed within CCM3-IBIS. The land use maps at the CCM3 resolution were made by aggregation of the higher resolution land use datasets. The land use type at each CCM3 cell was calculated as the dominant land use type (deforestation/natural vegetation) from the corresponding area in the high-resolution dataset. Adjustments were made to keep the total deforested area equivalent between the two datasets. Three simulations were made for each scenario and ensembles were created to reduce the effects of variability. CO2 concentrations and SST patterns were kept constant in all simulations, so the results are the consequence of the changes in land use only. 169 The surface and sub-surface runoff for the last 10 years of the CCM3-IBIS ensemble runs were averaged to create monthly mean values and the differences of the individual experiments from the potential vegetation run (POT) were created (GOV-POT and BAUPOT). These difference files were interpolated to ½-degree resolution of IBIS and added to the IBIS-simulated runoff used as input in the CTL simulation in THMB (‘‘Offline IBIS and THMB simulations”). We chose to force THMB with the difference between simulations rather than the direct GCM output of each experiment, as in (Broström et al., 1998; Coe, 2000), in order to reduce the influence of any GCM bias and facilitate direct comparison with the results of the IBIS-offline experiments. These two datasets were used as input to THMB and the model was run for the period 1915– 2000, as above, to create CCM3 governance (CCM3-GOV) and business-as-usual (CCM3-BAU) river discharge. The difference between the discharge of the CCM3-GOV and CCM3-BAU experiments from IBIS-offline CTL is a measure of the influence of deforestation on the coupled climate-biosphere system. Results Results of offline IBIS simulations – no deforestation (CTL) The area of interest in this study is the entire Amazon Basin with specific emphasis on the major southern and western tributaries of the Tocantins, Xingu, Tapajós, Madeira, Purus and Juruá basins where the greatest part of the historical deforestation has taken place and where much of the future deforestation is predicted to occur (Figs. 1 and 3). The CTL simulation reproduces the discharge of the Amazon River in good agreement with the observations. The CTL results of this study differ somewhat from those of (Coe et al., 2007) because the version of IBIS used in this study includes more explicit representation of root and soil dynamics (Li et al.,2005, 2006, 2007) and as a result the evapotranspiration differs from the previous version of IBIS (Coe et al., 2002, 2007). The mean monthly hydrograph for 11 major streams of the Amazon is in excellent agreement with the observations. The correlation coefficient (r2) is greater than 0.85 at all stations (Table 1, Fig. 3). The discharge magnitude also agrees well with the observations at most locations as indicated by the low percent relative error (%RE, Table 1), which is within ±16% for all streams except the Xingu (+25%). There is a relatively strong negative bias for the Tocantins River also (RE = 15%) that will be discussed in ‘‘Results of offline IBIS simulations – year 2000 (MOD)”, and is most likely related to the fact that the Tocantins is largely deforested but in this control simulation has the potential vegetation of Cerrado and forest. There is also a negative bias in the magnitude of the discharge at the most downstream location on the Amazon (RE = 11%) that is due, in part, to underestimated flux on the Madeira River (RE = 16%). Results of offline IBIS simulations – year 2000 (MOD) The estimates of total deforestation by the year 2000 indicate that about 10% of the roughly 5 million km2 area of rainforest in the Amazon has been altered (Fig. 1, Table 2). The Tapajós is about 20% deforested and the Madeira, Xingu, and Japurá Rivers 12–13% deforested. The Purus and Juruá are relatively undisturbed with less than 5% of their rainforest affected. The Tocantins is the most disturbed with about 58% of the rainforest area in agriculture (Table 2). The simulated influence of deforestation as of 2000 is modest in the MOD simulation, with the exception of the Tocantins, but leads to improvement of the agreement with observations compared to Author's personal copy 170 M.T. Coe et al. / Journal of Hydrology 369 (2009) 165–174 Fig. 3. Major watersheds of the Amazon basin. Numbers correspond to the gauge stations for which results are presented in the text and tables. The cross-hatched area corresponds to the watershed of the Óbidos station (#33, Table 1). Table 1 Analysis of simulated discharge compared to observed for potential vegetation (CTL) and year 2000 vegetation (MOD) simulated with IBIS offline. n Amazon #33 Negro #21 Japurá #9 Solimões #10 Solimões #5 Juruá #8 Purus #17 Madeira #31 Tapajós #38 Xingu #44 Tocantins #56 350 256 259 120 297 308 153 337 227 331 246 CTL Table 2 Deforested area of entire Amazon and selected watersheds. Area (km2) MOD r2 RE% RMSE% r2 RE% RMSE% 0.9925 0.9887 0.9652 0.9219 0.8697 0.8581 0.8483 0.9252 0.9252 0.9757 0.9213 11 3 2 3 7 1 7 16 12 22 15 18 24 43 24 27 48 33 32 53 49 35 0.9938 0.9887 0.9660 0.9192 0.8637 0.8574 0.8486 0.9384 0.9305 0.9754 0.9665 10 3 2 2 5 0 6 11 17 27 7 17 24 43 24 28 48 33 29 55 51 30 Entire basin Amazon #33 Negro #21 Japurá #9 Solimões #10 Solimões #5 Juruá #8 Purus #17 Madeira #31 Tapajós #38 Xingu #44 Tocantins #56 4970079 3702481 582064 217367 1858883 867257 156376 333480 906552 285072 377175 189048 Deforested % MOD (%) GOV (%) BAU (%) 10 7 3 12 6 8 1 3 13 20 12 58 30 25 15 18 29 21 22 21 41 50 26 80 49 40 29 20 42 23 46 43 61 82 66 93 Number of months of observed and simulated discharge used in the statistical analysis, n, coefficient of correlation, r2, percent relative error, RE, and percent root mean square of the error, RMSE. Total watershed area (km2) and fraction deforested as of 2000 (from Eva et al., 2002) and the year 2050 GOV and BAU deforestation scenarios of Britaldo SoaresFilho et al. (2006). CTL for almost all locations (Tables 1 and 3). The discharge is increased by 7% on the Madeira compared to CTL and by 5% or less on the Tapajós and Xingu. The simulated discharge of the Tocantins is increased by 26% compared to CTL, the r2 is improved to 0.9665 (from 0.9213), and the RE is 7% compared to 15% for CTL. The results of the MOD simulation suggest that historical deforestation has had a large impact on the discharge of the Tocantins River and small impact elsewhere in the southern tributaries. The simulated results at the 175,000 km2 Porto Nacional subwatershed of the Tocantins River (not shown) illustrate the importance of the recent deforestation on the water balance and support the analysis of observations at that location made by Costa et al. (2003). Costa et al. (2003) compared the mean discharge for two 20-year periods at Porto Nacional; 1949–1968 when the basin was less than 30% deforested and 1979–1998 when the basin was about 50% deforested. In that study the authors found a 25% Table 3 Fractional change in simulated discharge, offline IBIS/THMB. Discharge Amazon #33 Negro #21 Japurá #9 Solimões #10 Solimões #5 Juruá #8 Purus #17 Madeira #31 Tapajós #38 Xingu #44 Tocantins #56 MOD (%) GOV (%) BAU (%) 2 0 0 1 2 0 1 7 4 5 26 5 0 0 3 5 2 4 18 9 9 32 7 1 0 4 6 5 6 23 13 15 34 Offline IBIS/THMB simulated change in discharge relative to the potential vegetation simulation (CTL). Author's personal copy M.T. Coe et al. / Journal of Hydrology 369 (2009) 165–174 increase in annual mean discharge about ½ of which, was attributed by the authors to a precipitation increase and about ½ to deforestation and the resulting increase in ET. A comparison of the CTL and MOD simulated discharge for the two periods shows results similar to Costa et al. (2003). In CTL the simulated mean annual discharge at Porto Nacional is 11% greater in the 1979–1998 period compared to the 1949–1968 period. Since the land cover is unchanged in that simulation and the ET difference is minimal the 11% increase in discharge can be attributed to a precipitation increase between these periods. The difference between the MOD simulated discharge for the period 1979–1998 and the CTL 1949–1968 is 25%, which is comparable to the observed change. Therefore, the simulations and observations suggest that a large part of the observed change in discharge over the last 50 years in the Tocantins is most likely due to deforestation. 171 Implications for the future be paved on time, compliance with laws regarding protected lands (public and private) is low, and no new protected areas are created. As a result by the year 2050 total deforestation is almost 50% of the tropical evergreen forests of the basin (Table 2). The Tocantins is the most deforested, about 93% by 2050. The total deforested area in the other basins ranges from about 20% of the Japurá to 61% of the Madeira, 66% of the Xingu, and 82% of the Tapajós by 2050. The two scenarios illustrate that in the future even under strict compliance with laws (GOV), deforestation is likely to be large: about 1.5 million km2 of the basin deforested by 2050. However, the difference between the GOV and BAU scenarios also indicates that compliance with existing laws on public and private lands and additions to protected areas can have significant impacts, avoiding in these scenarios, almost 1 million km2 of future deforestation. Given the sensitivity of local evapotranspiration and regional precipitation to large-scale deforestation (Bonan, 2008), the two scenarios suggest that there will be potentially large changes in river flow and aquatic ecosystems in the future. Future land use scenarios In the (Soares-Filho et al., 2006) GOV scenario no deforestation takes place on protected areas, new protected lands are created, and at least 50% of all private lands remain in forest. Despite the strict adherence to laws, by the year 2050, 30% of the Amazon evergreen forest is deforested, three times the year 2000 value (Table 2, Fig. 1). Deforested area varies from 15% of the Negro in the far north to 80% of the Tocantins in the southeast. The Xingu Basin is considerably less deforested than its neighbors (26%) because of the Xingu Indigenous Park and associated protected lands, which occupy about 55% of the basin. The (Soares-Filho et al., 2006) BAU scenario assumes that current deforestation trends will continue, all scheduled roads will Results of offline IBIS simulations – land surface processes only The simulations with IBIS offline with GOV and BAU land cover and identical prescribed climate illustrate the relative influence of land surface processes alone on the local evapotranspiration and water balance. The simulated river discharge in all tributaries is increased, relative to the control simulation, proportional to the area deforested, as a decrease in total LAI in any deforestation experiment results in reduced evapotranspiration and increased runoff (Fig. 4a, Table 3). The annual mean discharge of the Amazon River (station #33, Fig. 2) is increased by 5% in GOV relative to the CTL simulation with about 25% of the basin deforested and by 7% in BAU with 40% of the basin deforested (Tables 2 and 3). Greatest change in annual mean discharge (about 10–30% increase) is Fig. 4. Change in annual mean discharge relative to the IBIS-offline CTL simulation for the GOV (blue) and BAU (red) simulations for (a) IBIS-offline simulations and (b) coupled CCM3-IBIS simulations. Author's personal copy M.T. Coe et al. / Journal of Hydrology 369 (2009) 165–174 172 simulated in the southern tributaries consistent with the largest area deforested by 2050. The northern and western tributaries have 5% or less increase in GOV consistent with about 25% or less of those basins being deforested, and only a modest increase in BAU. The Tocantins River, with about 80% of the basin deforested in the GOV simulation, has a 32% simulated increase relative to CTL, and 34% increase in BAU, with 93% deforestation. The Madeira has an 18% increase in GOV that increases to 23% in BAU, while the Tapajós and Xingu both have a 9% increase in GOV that increases to 13% and 15% respectively in BAU (Table 3). The results of these offline simulations are consistent with small, and in the case of the Tocantins and Araguaia (Coe et al., 2008; Costa et al., 2003), large-scale observations of the response of river discharge to deforestation and conversion to agriculture: evapotranspiration decreases as native vegetation is replaced with less water demanding pasture and crops, and annual mean discharge increases. Therefore, as with the case of the Tocantins and Araguaia Rivers, which already show an increase in discharge of about 25%, in the absence of any significant atmospheric feedbacks to precipitation, future deforestation in any of these basins can be expected to lead to locally increased discharge. However, as has been demonstrated in numerous global climate model simulations, with large-scale deforestation regional precipitation is expected to decrease because of the combined influences of increased albedo, decreased surface roughness and decreased water recycling that accompany deforestation (Costa, 2005; Delire et al., 2001; Dickinson and Henderson-Sellers, 1988; Malhi et al., 2008a). Long-term discharge is the residual of the precipitation minus the ET, therefore, any decrease in precipitation will act to decrease discharge and offset some or all of the increase in discharge that may result from a local decrease in ET. (3%) and most of the tributaries (Fig. 4b, Tables 2 and 4). The net discharge decrease indicates that the regional precipitation changes that result from the feedback with deforestation are larger than the local ET decrease and dominate the discharge response. In the case of the Negro there is a moderate precipitation decrease of 8% (Table 4), relatively small land cover change (15% Tables 2 and 4) and therefore, a relatively large discharge decrease of 10%. The southern tributaries, where deforestation is greatest respond in a more complex way. The Xingu River has a large precipitation decrease (15%) due to the regional atmospheric changes but the deforestation within the basin is relatively low compared to its neighbors (26%). As a result, there is an 11% decrease in simulated discharge in the Xingu, which is greater than any other basin (Tables 2 and 4). The neighboring Madeira, Tapajós, and Tocantins basins have a similarly large precipitation decrease but the deforested area is much greater than the Xingu. As a result there is an increase in the discharge of the Madeira (3%) and Tocantins (12%) and a small decrease in the Tapajós (2%) (Tables 2 and 4). In the BAU-CCM3 simulation about 40% of the Amazon upstream of Óbidos is deforested and precipitation is further decreased in almost all basins, particularly where deforestation is greatest (Tables 2 and 4). The discharge is decreased in all tributaries except the Japurá, Madeira, and Tocantins Rivers relative to the GOV-2050 and CTL simulations (Table 4, Fig. 4b). Precipitation is decreased by an additional 4% in the Madeira River compared to GOV but the deforested area increases from 54% to 69% of the watershed, as a result there is a slight increase in the discharge to 4% greater than CTL. The discharge difference of the Tocantins River relative to CTL decreases to about 8% compared to about 12% in GOV-2050. Results of coupled CCM3/IBIS simulations – land surface processes and atmospheric feedbacks The GOV-CCM3 and BAU-CCM3 simulations illustrate the potential importance of the combined effects of the local ET decrease and regional precipitation decrease on the river discharge. The mean annual precipitation of the GOV-CCM3 simulation is decreased in all tributaries of the Amazon compared to the precipitation used in the CTL. The greatest change, compared to the control simulation, of as much as 15% occurs, in the southeastern tributaries where deforestation is also greatest (Tables 2 and 4). In contrast to the offline IBIS-THMB results presented in ‘‘Results of offline IBIS simulations – land surface processes only”, there is a small net decrease in simulated discharge for the mainstem of the Amazon Discussion and conclusions Table 4 Change in coupled CCM3/IBIS model simulated precipitation and discharge relative to the offline IBIS/THMB CTL. Precipitation Amazon #33 Negro #21 Japurá #9 Solimões #10 Solimões #5 Juruá #8 Purus #17 Madeira #31 Tapajós #38 Xingu #44 Tocantins #56 Discharge GOV-CCM3 (%) BAU-CCM3 (%) GOV-CCM3 (%) BAU-CCM3 (%) 9 8 5 6 5 5 9 14 12 15 15 12 12 2 7 5 13 15 17 16 20 14 3 10 3 1 0 8 5 3 2 11 12 4 10 0 2 1 13 8 4 5 17 8 The precipitation values are the input data derived for the GOV-CCM3 and BAUCCM3 experiments minus the precipitation used as input to the offline-IBIS CTL simulation. The discharge values are the differences of the simulated THMB discharge forced with the coupled model data from the CTL. At the micro scale to meso-scale, deforestation generally results in decreased ET and increased runoff, and discharge. At the largescale, atmospheric feedbacks may significantly reduce precipitation regionally and, if larger than the local ET changes, may decrease water yield, runoff and discharge. The results presented here, as with all modeling studies, are to some unknown-degree dependent on the experiment design and model assumptions. IBIS has been well-tested and validated for historical climate for tropical and non-tropical vegetation types and its sensitivity to land cover differences has been validated in previous studies (Li et al., 2007). However, as pointed out by (Li et al., 2007) the sensitivity of IBIS simulated ET to deforestation is, in part, a function of model specific parameters such as plant rooting depth and soil hydraulic properties, among others. The simulated climate of the coupled CCM3-IBIS model has been extensively validated and shown to reproduce historical climate of the Amazon in good agreement with the observations (Senna et al., submitted for publication). However, the exact response of the atmospheric circulation and climate simulated by CCM3-IBIS to deforestation is a function of numerous specific features of this model and experiment design such as resolution, convection scheme, cloud parameterizations, prescribed sea-surface temperatures, fixed CO2, etc. For example, the potential feedbacks between deforestation, globally increasing CO2 and temperature and plant physiological and soil respiration responses not included in this study (Cox et al., 2000; Knorr et al., 2005; Korner and Arnone, 1992; Melillo et al., 2002) are not well understood but could potentially affect hydrology through changes in plant water demand and ecosystem structure (Moorcroft, 2006). Despite the unknown aspects of the model sensitivity, the series of simulations presented in this manuscript clarify a few important points about the impact of deforestation on the Amazon River. The Author's personal copy M.T. Coe et al. / Journal of Hydrology 369 (2009) 165–174 simulations with IBIS and THMB offline indicate that the local ET decrease and subsequent discharge increase can be a significant fraction of the water balance when greater than 50% of a watershed is deforested. The results of this study agree with the findings reported for a 175,000 km2 section of the Tocantins River (Costa et al., 2003) and an 82,000 km2 section of the Araguaia (Coe et al., 2008): observed discharge has increased by about 25%, despite little precipitation change, in the last 50 years as these watersheds were converted from predominantly forest and Cerrado to pasture and agriculture. The model results suggest that in the absence of a continental scale precipitation change, large-scale deforestation can have a significant impact on a large river system and appears to have already done so, at least in the Tocantins and Araguaia Rivers. In the other large tributaries, where deforestation has not yet exceeded 25% of the watershed area, any changes to discharge are probably too small to be detected in the observations (<10%). However, smaller watersheds in which deforestation has already exceeded 50% (e.g. the southern Xingu, and Tapajós) may already be experiencing large and as yet undocumented changes. The coupled CCM3-IBIS results suggest that atmospheric feedbacks brought about by large-scale deforestation may be of the same order of magnitude as the changes to local land surface processes, but of opposite sign. Additionally, changes in the water balance caused by atmospheric feedbacks are not limited to those basins where deforestation has occurred but are spread unevenly throughout the basin by atmospheric circulation. As a result, changes to discharge and aquatic environments with future deforestation of the Amazon will likely be a complex function of how much vegetation has been removed from that particular watershed and how much has been removed from the entire Amazon Basin. For example, in the GOV simulation the Purus River is about 20% deforested, which the offline IBIS simulations suggest should result in a very small positive discharge anomaly (<5% increase) from the ET decrease. However, in the coupled simulations the Purus has a 5% decrease in mean discharge, illustrating the importance of precipitation changes that result from deforestation in tributaries other than the Purus. Finally, these simulations also illustrate that the response in individual tributaries is likely to change with time. For example, the Tocantins River, which is already largely deforested, is likely to be experiencing the greatest positive discharge anomaly now, because its discharge has increased due to the local land surface processes but atmospheric feedbacks do not appear to have affected precipitation. In the future, as the Amazon becomes progressively more deforested, precipitation in the Tocantins is likely to decrease but local deforestation will not change significantly, and hence the positive discharge anomaly in the Tocantins will decrease. Other tributaries that are more distant from the deforestation front, such as those in the west and north may first experience a decrease in discharge as the southeast Amazon becomes deforested and atmospheric feedbacks decrease precipitation but the negative anomalies may be reduced or reversed as deforestation begins to occur in the individual watersheds and the local land surface processes become a more important part of the water balance. The results of these simulations strongly support the idea that policies favoring conservation will have a potentially big influence on individual watersheds. For example, in the GOV simulation the Xingu River behaves very differently than its neighboring watersheds, the Tapajós and Tocantins, because maintenance of the Xingu Indigenous Park and other protected areas in the watershed limits deforestation to about 25% of the basin. One important issue that these simulations do not address is the potentially very large change to extreme events that would accompany any moderate and large shift in the mean discharge. The necessarily coarse temporal and spatial scales of these simulations make examination of changes to extreme events impossible. 173 However, as observations indicate (Bradshaw et al., 2007) it is reasonable to expect that a change in the mean state will result in an increase in the scale of flood and drought events. These simulations also do not address the potentially large morphological and biochemical changes to rivers, that accompany land cover change and that have important impacts on aquatic environments (Coe et al., 2008; Gordon and peterson, 2008). These changes to extreme events, morphology, and biochemistry will be responsible for most of the social, ecological, and economic disruptions to come from deforestation and must be addressed with higher resolution modeling and observational studies in the future. In summary, the exact hydrological future of the Amazon is uncertain but the results of this study suggest that a combination of land surface responses and atmospheric feedbacks from historical deforestation has already influenced watersheds in the southeastern Amazon basin significantly and that complex humaninduced impacts will spread throughout the basin as a larger area is deforested and converted to pasture and agriculture. Understanding future changes and mitigating future impacts in an individual watershed will only come by integrating continental-scale and local-scale information. Acknowledgements We gratefully acknowledge the contributions of Paul Lefebvre and Claudia Stickler and reviewer Dr. Christine Delire. This work was supported through grants from The National Aeronautic and Space Administration Large-Scale Atmosphere and Biosphere Experiment in Amazonia program and the Gordon and Betty Moore Foundation. References Achard, F. et al., 2002. Determination of deforestation rates of the world’s humid tropical forests. Science 297 (5583), 999–1002. Alencar, A., Nepstad, D., McGrath, D., Moutinho, P., Pacheco, P., 2004. Desmatamento na Amazônia: Indo Além da ‘Emergência Crônica. IPAM (Instituto de Pesquisa Ambiental da Amazônia), Belém, Brazil. Ball, J.T., Woodrow, I.E., Berry, J.A., 1986. A model predicting stomatal conductance and its contribution to the control of photosynthesis under different light conditions. In: Biggins, J. (Ed.), Progress in Photosynthetic Research. M. Nijhoff Publishers, Dordrecht, pp. 221–224. Bonan, G., 2008. Forests and climate change: forcings, feedbacks, and the climate benefits of forests. Science 320 (5882), 6. Bonan, G.B., DeFries, R.S., Coe, M.T., Ojima, D.S., 2004. Land use and climate. In: Gutman, G.e.a. (Ed.), Land Change Science. Kluwer Academic Publishers, Amsterdam, pp. 301–314. Botta, A., Ramankutty, N., Foley, J.A., 2002. Long-term variations of climate and carbon fluxes over the Amazon basin. Geophysical Research Letters 29 (9). Bradshaw, J.A., Sodhi, N.S., Peh, K.S.-H., Brook, B.W., 2007. Global evidence that deforestation amplifies flood risk and severity in the developing world. Global Change Biology 13, 17. Broström, A. et al., 1998. Land surface feedbacks and palaeomonsoons in northern Africa. Geophysical Research Letters 25 (119), 4. Bruijnzeel, L.A., 1990. Hydrology of moist tropical forests and effects of conversion: a state of knowledge review. UNESCO, Paris and Amsterdam. 226 pp. Campos, M.T., Nepstad, D.C., 2006. Smallholders, the Amazon’s new conservationists. Conservation Biology 20 (5), 1553–1556. Carvalho, G.O., Barros, A.C., Moutinho, P.R.S., Nepstad, D.C., 2001. Sensitive development could protect Amazonia instead of destroying it. Nature 409, 131. Coe, M.T., 1998. A linked global model of terrestrial hydrologic processes: simulation of modern rivers, lakes, and wetlands. Journal of Geophysical Research 103, 15. Coe, M.T., 2000. Modeling terrestrial hydrologic systems at the continental scale: testing the accuracy of an atmospheric GCM. Journal of Climate 13, 19. Coe, M.T., Foley, J.A., 2001. Human and natural impacts on the water resources of the Lake Chad basin. Journal of Geophysical Research 106 (D4), 3349–3356. Coe, M.T., Costa, M.H., Botta, A., Birkett, C., 2002. Long-term simulations of discharge and floods in the Amazon basin. Journal of Geophysical Research 107 (D20), 17. Coe, M.T., Costa, M.H., Howard, E.A., 2007. Simulating the surface waters of the Amazon River Basin: impacts of new river geomorphic and dynamic flow parameterizations. Hydrological Processes 21, 12. Costa, M.H., 2005. Large-scale hydrological impacts of tropical forest conversion. In: Bonell, M., Bruijnzeel, L.A. (Eds.), Forests, Water and People in the Humid Tropics. Cambridge University Press, New York, pp. 590–597. Author's personal copy 174 M.T. Coe et al. / Journal of Hydrology 369 (2009) 165–174 Costa, M., Foley, J., 1997. Water balance of the Amazon Basin: dependence on vegetation cover and canopy conductance. Journal of Geophysical Research 102 (D20), 17. Costa, M.H., Foley, J.A., 1998. A comparison of precipitation datasets for the Amazon Basin. Geophysical Research Letters 25, 155–158. Costa, M.H., Foley, J.A., 2000. Combined effects of deforestation and doubled atmospheric CO2 concentrations on the climate of Amazonia. Journal of Climate 13 (1), 18–34. Costa, M.H., Botta, A., Cardille, J.A., 2003. Effects of large-scale changes in land cover on the discharge of the Tocantins River, Southeastern Amazonia. Journal of Hydrology 283, 12. Costa, M.H., Yanagi, S.N.M., Souza, P., Ribeiro, A., Rocha, E.J.P., 2007. Climate change in Amazonia caused by soybean cropland expansion, as compared to caused by pastureland expansion. Geophysical Research Letters 34 (7). Cox, P.M., Betts, R.A., Jones, C.D., Spall, S.A., Totterdell, I.J., 2000. Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model (vol. 408, pp. 184, 2000). Nature 408 (6813), 750. D’Almeida, C. et al., 2006. A water balance model to study the hydrological response to different scenarios of deforestation in Amazonia. Journal of Hydrology 331 (1–2), 125–136. D’Almeida, C. et al., 2007. The effects of deforestation on the hydrological cycle in Amazonia: a review on scale and resolution. International Journal of Climatology 27 (5), 633–647. Delire, C., Foley, J.A., 1999. Evaluating the performance of a land surface/ecosystem model with biophysical measurements from contrasting environments. Journal of Geophysical Research-Atmospheres 104 (D14), 16895–16909. Delire, C. et al., 2001. Simulated response of the atmosphere – ocean system to deforestation in the Indonesian archipelago. Geophysical Research Letters 28 (10), 2081–2084. Delire, C. et al., 2002. Comparison of the climate simulated by the CCM3 coupled to two different land–surface models. Climate Dynamics 19 (8), 657–669. Delire, C., Foley, Jonathan A., Thompson, S., 2004. Long-term variability in a coupled atmosphere–biosphere model. Journal of Climate 17, 13. Dickinson, R.E., Henderson-Sellers, A., 1988. Modelling tropical deforestation: a study of GCM land–surface parameterizations. Quarterly Journal of the Royal Meteorological Society 114, 439–462. Donner, S.D., Coe, Michael T., Lenters, J.D., Twine, T.E., Foley, J.A., 2002. Modeling the impact of hydrological changes on nitrate transport in the Mississippi River Basin from 1955–1994. Global Biogeochemical Cycles 16 (3), 16. Eva, H.D. et al., 2002. A Vegetation Map of South America. Official Publications of the European Communities, Luxembourg. Fearnside, P.M., 1993. Deforestation in Brazilian Amazonia: the effect of population and land tenure. Ambio 22 (8), 537–545. Fearnside, P.M., Graça, P.M.L.d.A., 2006. BR-319: Brazil’s manaus-porto velho highway and the potential impact of linking the arc of deforestation to central Amazonia. Environmental Management 38, 705–716. Foley, J.A., Botta, A., Coe, M.T., Costa, M.H., 2002. El Nino-southern oscillation and the climate, ecosystems and rivers of Amazonia. Global Biogeochemical Cycles 16 (4). Gordon, L.J., Peterson, G.D., 2008. Agricultural modifications of hydrological flows create ecological surprises. Trends in Ecology & Evolution 23 (4). Green, W.H., Ampt, G.A., 1911. Studies on soil physics, 1. The flow of air and water through soils. Journal of Agricultural Science 4, 25. Huffman, G.J., Adler, R.F., Arkin, P., 1997. The global precipitation climatology project (GPCP) combined precipitation dataset. Bulletin of the American Meteorological Society 78, 16. INPE, 1999. Monitoring of the Brazilian Amazonian Forest by Satellite 1997–1998. Sao Jose dos Campos, Sao Paulo, Brazil. INPE, 2000. Desflorestamento 1998–1999. INPE, São José dos Campos, São Paulo, Brazil. INPE, 2004. INPE, São Paulo. <http://www.obt.inpe.br/prodes/>. Kaimowitz, D., Mertens, B., Wunder, S., Pacheco, P., 2004. Hamburger connection fuels Amazon destruction. Center for International Forest Research, Bangor, Indonesia. Kalnay, E., Kanamitsu, M., Kistler, R., 1996. The NCEP/NCAR 40-year reanalysis project. Bulletin of the American Meteorological Society 77, 35. Knorr, W., Prentice, I.C., House, J.I., Holland, E.A., 2005. Long-term sensitivity of soil carbon turnover to warming. Nature 433 (7023), 298–301. Korner, C., Arnone III, J.A., 1992. Responses to elevated carbon dioxide in artificial tropical ecosystems. Science 257, 1672–1675. Kucharik, C.J. et al., 2000. Testing the performance of a dynamic global ecosystem model: water balance, carbon balance, and vegetation structure. Global Biogeochemical Cycles 14 (3), 795–825. Kummerow, C., Barnes, W., Kozu, T., Shiue, J., Simpson, J., 1998. The tropical rainfall measuring mission (TRMM) sensor package. Journal of Atmospheric and Oceanic Technology 15, 9. Laurance, W.F. et al., 2001. The future of the Brazilian Amazon. Science 291, 438– 439. Leemans, R., Cramer, W.P., 1990. The IIASA database for mean monthly values of temperature, precipitation and cloudiness on a global terrestrial grid, Laxenburg, Austria. Legates, D.R., Willmott, C.J., 1990. Mean seasonal and spatial variability in gaugecorrected, global precipitation. International Journal of Climatology 10, 17. Li, K.Y., Coe, M.T., Ramankutty, N., 2005. Investigation of hydrological variability in West Africa using land surface models. Journal of Climate 18 (16), 3173–3188. Li, K.Y., De Jong, R., Coe, M.T., Ramankutty, N., 2006. Root-water-uptake based upon a new water stress reduction and an asymptotic root distribution function. Earth Interactions 10, 1–22 (10-014). Li, K.Y., Coe, M.T., Ramankutty, N., De Jong, R., 2007. Modeling the hydrological impact of land-use change in West Africa. Journal of Hydrology 337, 258–268. Malhi, Y. et al., 2008a. Climate change, deforestation, and the fate of the Amazon. Science 319 (5860), 169–172. Malhi, Y. et al., 2008b. Climate change, deforestation, and the fate of the Amazon. Science 319, 4. Marengo, J.A., Nobre, C.A., Tomasella, J., Cardosa, M.F., Oyama, M.D., 2008a. Hydroclimate and ecological behaviour of the drought of Amazonia in 2005, vol. 6. Philosophical Transactions of the Royal Society. Marengo, J.A. et al., 2008b. The drought of Amazonia in 2005. Journal of Climate 21 (3), 22. Melillo, J.M. et al., 2002. Soil warming and carbon-cycle feedbacks to the climate system. Science 298, 2173. Moorcroft, P.R., 2006. How close are we to a predictive science of the biosphere. Trends in Ecology & Evolution 21 (7), 8. Nepstad, D.C. et al., 1999. Large-scale impoverishment of Amazonian forests by logging and fire. Nature 398, 505–508. Nepstad, D.C., Stickler, C.M., Almeida, O.T., 2006. Globalization of the Amazon soy and beef industries: opportunities for conservation. Conservation Biology 20 (6), 1595–1603. New, M., Hulme, M., Jones, P., 1999. Representing twentieth-century space–time climate variability. Part I: development of a 1961–90 mean monthly terrestrial climatology. Journal of Climate 12, 28. New, M., Hulme, M., Jones, P., 2000. Representing twentieth-century space–time climate variability. Part II: Development of 1901–96 monthly grids of terrestrial surface climate. Journal of Climate 13 (13), 22. Nobre, C.A., Sellers, P.J., Shukla, J., 1991. Amazonian deforestation and regional climate change. Journal of Climate 4, 957–988. Ramankutty, N., Foley, J.A., 1998. Characterizing patterns of global land use: an analysis of global croplands data. Journal of Geophysical Research 103 (D22), 13. Sahin, V., Hall, M.J., 1996. The effects of afforestation and deforestation on water yields. Journal of Hydrology 178 (1–4), 293–309. Sampaio, G. et al., 2007. Regional climate change over eastern Amazonia caused by pasture and soybean cropland expansion. Geophysical Research Letters 34 (L17709). Scanlon, B.R., Jolly, I., Sophocleous, M., Zhang, L., 2007. Global impacts of conversions from natural to agricultural ecosystems on water resources: quantity versus quality. Water Resources Research 43 (3). Senna, M.C. et al., submitted for publication. Challenges of a coupled climate biosphere model to reproduce vegetation structure and dynamics in Amazonia. Global Change Biology. Shankar, D., Kotamraju, V., Shetye, S.R., 2004. A quantitative framework for estimating water resources in India. Current Science 86 (4), 10. Simon, M.F., Garagorry, F.L., 2005. The expansion of agriculture in the Brazilian Amazon. Environmental Conservation 32 (3), 203–212. Skole, D.L., Tucker, C.J., 1993. Tropical deforestation and habitat fragmentation in the Amazon satellite data from 1978 to 1988. Science 260, 1905–1910. Skole, D.L., Chomentowski, W.H., Salas, W.A., Nobre, A.D., 1994. Physical and human dimensions of deforestation in Amazoina. Bioscience 44 (5), 314–321. Soares Filho, B.S. et al., 2008. Reduction of carbon emissions associated with deforestation in Brazil: the role of the Amazon Region Protected Areas Program (ARPA), WWF-USA, WWF-Brazil, IPAM. Soares-Filho, B.S. et al., 2004. Simulating the response of land-cover changes to road paving and governance along a major Amazon highway: the Santarem-Cuiaba corridor. Global Change Biology 10 (5), 745–764. Soares-Filho, B.S. et al., 2006. Modelling conservation in the Amazon basin. Nature 440 (7083), 520–523. Suprit, K., Shankar, D., 2007. Resolving orographic rainfall on the Indian west coast. International Journal of Climatology 28 (5), 15. Trancoso, R., 2006. Changes in land cover and alterations in the hydrological response of catchments in Amazonia. INPA/UFAM, 139 pp. Uppala, S.M., Kallberg, P.W., Simmons, A.J., Andrae, U., 2005. The ERA-40 re-analysis. Quarterly Journal of the Royal Meteorological Society 131 (612), 52. Xie, P., Arkin, P.A., 1997. Global precipitation: a 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs. Bulletin of the American Meteorological Society 78, 20.
© Copyright 2025 Paperzz