(GIS) to determine optimum forest cover for minimizing runoff in a

International Forestry Review Vol.11(3), 2009
375
Utilizing geographic information system (GIS) to determine
optimum forest cover for minimizing runoff in a degraded
watershed in Jamaica1
O. B. EVELYN
Forestry Department, 173 Constant Spring Road, Kingston 8, Jamaica
Email: [email protected]
SUMMARY
Optimum forest cover is not a new concept. It was part of the suggested policies to achieve sustainable forest management at a 1996 meeting
of the UN/FAO Intergovernmental Panel on Forests. It has not been actively pursued by many countries however, but since deforestation is
being singled out as one of the major contributing factors to degrading global climatic conditions, this concept may take on new meaning.
This research devised a methodology for determining the optimum forest cover for a degraded watershed in south-central Jamaica. The
model suggested a spatial pattern in the upper portions of the watershed which increased the forest cover from 16.76 % to 37.47 %. This
significantly reduced runoff where the forest cover was optimized and simulated in a hydrological model. This technique will give direction
to forest management and conservation in Jamaica and will form part of the decision support system for the country’s planners.
Keywords: optimum forest cover, watershed rehabilitation, run-off
Utilisation du Système d’information géographique (GIS) pour déterminer le couvert forestier
optimal afin de minimiser l’écoulement dans un bassin versant dégradé en Jamaique.
O. B. Evelyn
Le couvert forestier optimal n’est pas un nouveau concept. Il faisait partie de politiques suggérées pour parvenir à une gestion forestière
durable lors d’une réunion du corps intergouvernemental de l’ONU/FAO pour les forêts en 1996. Peu de pays les ont toutefois poursuivies,
mais, comme la déforestation est à présent présentée comme l’un des facteurs majeurs contribuant aux conditions climatiques globales en
dégradation, il se peut que le concept prenne une nouvelle signification. Cette recherche a développé une méthodologie pour déterminer le
couvert forestier optimal dans un bassin versant dégradé dans la Jamaique centrale du sud. Le modèle suggérait une organisation spatiale
dans les portions du haut du bassin versant qui étendait le couvert forestier de 16.76% à 37.47%. Cela réduisait sensiblement l’écoulement
quand le couvert forestier était optimisé et simulé dans un modèle hydrologique. Cette technique va donner une direction à la gestion et à la
conservation forestière en Jamaique et va tenir un rôle dans le système d’aide aux prises de décisions pour les planificateurs du pays.
Utilización del sistema de información geográfica (GIS) en aras de determinar la cobertura
forestal ideal para minimizar la escorrentía de superficie en una línea divisoria de aguas
degradada en Jamaica.
O. B. EVELYN
El concepto de la cobertura forestal ideal no es nada nuevo, ya que formó parte de las políticas propuestas para lograr la gestión forestal
sostenible en una reunión del Panel Intergubernamental sobre los Bosques de la ONU/FAO en 1996. Pocos países han proseguido esta
política, sin embargo, pero el concepto puede adquirir un mayor significado ahora que la deforestación ha sido señalada como uno de los
factores que más contribuyen a la degradación de las condiciones climáticas mundiales. Esta investigación concibió una metodología para
determinar la cobertura forestal ideal para una línea divisoria de aguas degradada en la zona sudcentral de Jamaica, proponiendo un modelo
espacial en las zonas superiores de la línea divisoria de aguas que aumentó la cobertura forestal desde un 16.76 % hasta un 37.47 %, lo cual
redujo la escorrentía de superficie de forma significativa cuando la cobertura forestal fue optimizada y simulada en un modelo hidrológico.
Esta técnica proporcionará pautas para la gestión forestal y la conservación en Jamaica, y formará parte del sistema de apoyo para la toma
de decisiones en la planificación nacional.
1
This article is a condensed version of the research project from the unpublished thesis submitted by the author in partial fulfilment of the
requirements for the degree of Master of Science in Geographic Information, City University, London, UK, (2007).
376
O.B. Evelyn
INTRODUCTION
Watershed degradation is one of the most complex
environmental problems facing nations globally. It threatens
the livelihood of millions of people, especially in Third
World countries. Watershed degradation can be viewed as a
loss of value over time. Human decision-making processes,
influenced by bio-physical and socio-economic factors
have been identified as the main drivers. Some of these
anthropogenic activities have resulted in the loss of habitats
for important flora and fauna, increased local flooding which
sometimes results in the loss of human life and property,
reduced water quantity and quality in rivers and streams, and
destructive climate change.
Equally, watershed rehabilitation is a complex and
challenging problem (Ziemer 1981, Shaojun et al. 2004).
Rehabilitation usually requires changes in human habits
and customs which take time and are sometimes resisted.
Oftentimes, compromises are necessary in order to obtain
a near optimal solution (Wang et al. 2004). Sometimes
conditions are so critical that countries are forced to
implement rehabilitation strategies through government
policies and legislation.
This project used existing land use optimization and
hydrological models coupled with Geographic Information
System (GIS) to optimize forest cover on a degraded watershed
in Jamaica. Figure 1 shows the modelling process flow.
FIGURE 1 Model process flow
The objectives are summarized as follows:
• to devise a methodology to determine the optimum
forest cover for a degraded watershed using GIS
software,
• to identify and use existing land use optimization
techniques in the development of the methodology,
• to reduce soil loss by erosion,
• to evaluate the spatial impacts of the derived forest
land use pattern on runoff,
• to identify and use existing GIS-based distributed
hydrological model in determining the effects on
runoff,
• to identify and prioritise target areas for reforestation
and conservation, and
• to produce guidelines for forest conservation and
planning.
During the 1980s and the 1990s the extent to which
deforestation was taking place in Jamaica became quite
acute. The country was experiencing several environmental
disasters and deforestation was blamed for some of these
events. One such event occurred in 1985 when it was reported
that “rains associated with hurricane Kate caused flooding in
the central parishes of St Ann, Clarendon, Manchester and
St Elizabeth. Five persons lost their lives, and some 60,000
farming families in southern Clarendon were affected” (IDB
2006). This precipitated calls for policies and measures to
combat deforestation (Eyre 1986, Eyre 1987). The country
had lost 25% of its forests between the years 1491 and
2000 (Evelyn and Camirand 2003) and a National Forest
Management and Conservation Plan (NFMCP) was one of
the measures being put in place to address this problem.
During the preparation of the NFMCP the question was
raised as to what should be the optimum forest cover for
the island. The answer was critical to the determination of
a forest cover target and the restoration and conservation
strategies necessary to achieve this. This question was never
answered because no research had been done in Jamaica on
this subject.
Optimum forest cover was strongly suggested as part of
the policies necessary for sustainable forest management
in many countries at a UN/FAO forum on The Underlying
Causes of Deforestation and Forest Degradation in 1996
(FAO 1996a). Delegates at the forum argued that “not all
changes in forest cover are necessarily harmful, (but) it is
only possible to decide what changes are or are not harmful
against a background of national policies that make a best
judgement of optimum forest cover”. The forum concluded
with a call to the delegates to include “in (their) national
forest and land-use plans... targets on the optimum forest
cover and utilization”.
Yeo et al. (2003) define optimization as “a procedure
for finding the best or optimal solution of a problem that
can be formulated mathematically with decision variables,
an objective function and constraints.” Forest optimization
could therefore be defined as finding the most desirable
forest distribution pattern for a region while taking
into consideration, among other things, the physical,
Utilizing GIS for minimizing runoff in a watershed in Jamaica
environmental, social and economic constraints.
Several land use optimization models are in use and are
classified according to the methodologies used for simulating
land use allocation (Briassoulis 1999, Agarwal et al. 2001,
Parker et al. 2002, Verburg et al. 2004). This project sought
to determine optimal forest land use patterns, therefore only
one set of models was considered – the optimization models.
Briassoulis (1999) stated that these models are making
significant contribution to land use analysis and planning
and have helped to solve many issues relating to sustainable
development worldwide. She identified five principal
categories according to the particular mathematical
techniques which they employed. These categories are:
a. linear programming (LP) models
b. dynamic programming (DP) models
c. goal programming, hierarchical programming,
linear and quadratic assignment, and nonlinear
programming (GP/HP/LQA/NLP) models
d. utility maximization (UM) models
e. Multi-Objective/Multi-Criteria Decision Making
(MODM/MCDM) models
Successful application of some of these models include
Shukla et al. (2001), who demonstrated the integration
of LP modelling techniques with GIS to solve a complex
land transformation problem, Hopkins et al. (1978 cited in
Briassoulis, 1999) and Okumu et al. (1999) who applied
DP techniques in making a sequence of interrelated land
use decisions, Shaikh et al. (2003) who used “an empirical
random utility maximization model to examine factors
affecting farmers’ decisions to accept a tree-planting
program on their marginal land and Wang et al. (2004)
who successfully employed the MODM/MCDM model to
allocate future land uses in Lake Erhai basin, China.
Briassoulis (1999) observed that MODM/MCDM
optimization modelling “involves the combination of
optimization techniques with elaborate, multidimensional
techniques of land use assessment/evaluation in a spatially
explicit modelling environment”. Grabaum and Meyer
(1998) in applying this model took the approach of imposing
constraints on a single objective function using goals derived
from a GIS landscape analysis and general aims.
Generally, there are two classes of hydrological models.
These are the lumped and the distributed models (Ward and
Robinson 2000). The former represents the collective effects
of land use changes in a watershed and is not spatially
explicit while the latter represents the spatial variability of
factors that control runoff, thus improving the predictability
of hydrologic processes.
Hydrologic responses to forest cover changes in a
watershed have been studied by several researchers. Burt and
Swank (2002), for example, reported that in a paired basin
experiment at the Coweeta Hydrologic Laboratory in North
Carolina, stream flow increased significantly (40%) after the
forest in one of the watersheds was cleared and then decreased
steadily during regrowth. Similar results have also been found
elsewhere by other researchers (e.g. Jones and Post 2004).
Two other studies of interest are Yeo et al. (2003), who
377
integrated a GIS-based distributed hydrological model and a
GIS-based land-use model to demonstrate that by optimizing
land use patterns in a watershed, a reduction (15-20%) in
peak runoff rates could be achieved with a consequential
decrease in nonpoint source pollution. Shrestha (2003), in
a different approach applied a semi-distributed hydrologic
model to evaluate runoff changes caused by land use patterns
in a watershed in the Kathmandu Valley basin in Nepal.
INTRODUCTION TO THE STUDY AREA
For management purposes, Jamaica is divided into twenty six
(26) Watershed Management Units (WMUs). Each WMU is
based on the drainage system of the major river or stream
in the area and is named according to either the former or
the latter. The Rio Minho River is the major river in the Rio
Minho Watershed Management Unit and it drains to the
sea. For this study, the Rio Minho Watershed is the drainage
system of the Rio Minho River within the Rio Minho WMU
(Figure 2). This system has an area of 522.3 km2.
The Rio Minho River has three major tributaries: the
Thomas, Pindars and Rock Rivers. These tributaries have
steep gradients hence run-off is of short duration and during
the dry season, flow is significantly reduced. The upper
north-eastern portions of the watershed consist of Basal
FIGURE 2 The Rio Minho Watershed
378
O.B. Evelyn
Aquiclude while the middle is mostly Limestone Aquifer.
The southern portions are mainly Alluvium Aquifer.
The average annual rainfall is 1 530mm , with a high of
more than 1 780mm occurring in the mountainous areas and
the limestone plateau in the North and a low of 1 000mm on
the low alluvial plains.
Soils in the mountainous areas consist manly of shales,
conglomerates and igneous rocks. These soils usually occur
on steep slopes, are lithosolic, well drained, shallow and
easily eroded. Soils on the limestone plateau have slow
internal drainage while soils on the alluvial plains are very
fertile and capable of intensive use.
The watershed falls within the parish of Clarendon, one
of the 14 parishes in Jamaica. Clarendon is one of the largest
parishes with an area of 1 196.3 km2 and had a population
of 236 995 in 2001. It is estimated that the population is
growing at a rate of 1.2% per annum.
In 1997, the Natural Resources Conservation Authority
(NRCA) in its Jamaica: State of the Environment Report
(NRCA 1997) considered most of the upper portions of the
Rio Minho WMU to be moderately to severely degraded
(Figure 3). This was attributed to the intense hillside
agricultural practices taking place there.
FOREST OPTIMIZATION METHODOLOGY
The technique of Grabaum and Meyer (1998) was modified
and adopted for the land use optimization modelling. The
techniques consists of two major parts: Use of GIS software
to assess the data for the land-use alternatives then using the
results from this process as input to an optimization model to
solve for the optimal forest land-use pattern as measured by
the objective function, subject to the necessary constraints.
Multi-Objective/Multi-Criteria
Decision
Making
optimization technique was used. This was chosen because
FAO pointed out that “the specification of a single objective
function does not adequately reflect the preferences of
decision-makers, which are of a multi-objective nature in
many practical problems of land resources optimization”
(FAO 1996b). The technique of Shukla et al. (2001) was
adopted to set up the optimisation model. The software
chosen was Microsoft Excel Solver (Solver) add-in. The
structure of the optimization model is outlined in Table 1.
Present Land use Assessments
The Jamaican Forestry Department developed a standard
broad classification system for land use/cover for the Island
when satellite imagery was used (Table 2). These were further
broken down into finer details when aerial photographs were
used (Camirand and Evelyn 2003).
Using IKONOS imagery, it was determined that in 2001
16.76% of the Rio Minho watershed was in Forest, 35.28%
in Mixed Forest and 47.96% was Non-Forest (Figure 4 and
Table 3).
One of the goals of this project was to identify and target
areas for forests conservation. The land use assessment
therefore suggested the conservation of all areas that are
presently classified as Forest (8 753.7 ha). The Mixed and
Non Forest areas that were degraded will be the only areas
to be analysed through the assessments outlines below with
a view to convert them to Forest. However, residential areas
and water bodies would remain unchanged.
The Plantations (PC) areas in the lowlands are export
crops such as sugar cane and banana. These fertile lands are
FIGURE 3 Conditions of the Watershed Management Units in the 1990s
Utilizing GIS for minimizing runoff in a watershed in Jamaica
TABLE 1 Structure of the Optimization Model
Maximize
forest cover
Objectives
Minimize
- soil loss (erosion)
- runoff
-ground water loss
-agricultural
production loss
Constrains
- land availability
- land capability
- urban/industria
- slope
- soil type/depth
- ecological
reserved by the government of Jamaican for economic reasons
and would not normally be allowed to be converted to forest.
The suitability of lands for conversion will therefore depend
on the sociological, economical or biophysical interests.
Assessment of Potential Forest Land Use
Land capability classification is the first step towards
proper land use (UNDP/FAO 1973). Slope is considered
the most important factor in determining land capability
classification because severe erosion from cultivated slopes
is a major watershed problem. Camirand and Evelyn (2003)
in a review of the land classification systems that have been
379
developed in Jamaica since 1954 recommended a scheme
which used slope and soil depth as the two most important
variables for the classification of suitable lands for forest
development, conservation and management. Guidelines for
the development of the system are outlined in Table 4.
To create a spatial GIS layer following the guidelines
outlined in Table 4, the required datasets are a digital
elevation model (DEM) and a soil layer. The process flow is
outlined in Figure 5.
This analysis was done using ArcGISTM and an area
distribution by land use classes was calculated (Figure 6 and
Table 5).
The analysis identified several areas which could
potentially be converted to Forest. These areas would mainly
consist of those areas classified as protective forest for
watershed management – Class 7 (FP) that are not already
under Forest cover (or residential). The analysis showed that
3,128.40 hectares from the Mixed and Non-Forest areas had
potential for conversion.
Assessment of Potential Soil Erosion
Several methods are available for assessing erosion due
to water (FAO 1985). The main purpose of these soil-loss
equations is to guide decision makers in soil conservation
TABLE 2 Broad land use/cover types for the island.
Type (Code)
Definition
Forest ( > 75 %, Minimum unit: 25 ha )
Closed primary forest with broadleaf trees at least 5 m tall and crowns interlocking, with
Closed Broadleaf (PF)
minimal human disturbance
Disturbed broadleaf forest with broadleaf trees at least 5 m tall and species-indicators of
Disturbed Broadleaf (SF)
disturbance such as Cecropia peltata (trumpet tree)
Bamboo (BB)
Bambusa vulgaris (bamboo brakes) on the lower shale hills (disturbed forest)
Open natural woodland or forest with trees at least 5 m tall and crowns not in contact, in drier
Tall Open Dry (WL)
part of Jamaica with species-indicators such as Bursera simaruba (red birch)
Open scrub, shrub, bush or brushland with trees or shrubs 1-5 m tall and crowns not in
Short Open Dry (SL)
contact, in drier part of Jamaica with species-indicators such as Prosopis juliflora (cashaw) or
Stenocereus hystrix (columnar cactus)
Edaphic forest (areas with brackish water) composed of trees with stilt roots or pneumatophores,
Mangrove (MG)
species-indicators such as Rhizophora mangle (red mangrove)
Mixed Forest
Disturbed Broadleaf Forest and
>50% Disturbed Broadleaf forest; >25% fields
Fields (SC)
Bamboo and Disturbed Broadleaf
>50% bamboo; >25% Disturbed Broadleaf forest
Forest (BF)
Bamboo and Fields (BC)
>50% bamboo; >25% fields
Fields and Disturbed Broadleaf
>50% fields; >25% Disturbed Broadleaf forest
Forest (CS)
Bauxite Extraction and Disturbed
>50% bauxite extraction; >25% Disturbed Broadleaf forest
Broadleaf Forest (BS)
Non Forest
Plantations (PC)
Tree crops, shrub crops like sugar cane, bananas, citrus and coconuts
Fields (FC)
Herbaceous crops, fallow, cultivated grass/legumes
Water Bodies (WA)
Lakes, rivers
Buildings and Other Infrastructure
Buildings and other constructed features such as airstrips, quarries, etc.
(BA)
380
O.B. Evelyn
FIGURE 4 2001 land use map of Rio Minho Watershed
Four primary GIS data layers are required to develop the
RUSLE factors. These are a rainfall, a soil type coverage, a
DEM and a land use layer, (Figure 7).
Each factor in the RUSLE is derived as follows:
R-Factor
The R factor is a summation of the various properties of
rainfall including intensity, duration and amount. It is
computed using rainfall energy and the maximum 30 minute
intensity. The equation for R can be written as follows
(Camirand 2005):
R= ∑EI30 /100.
Where, EI is “the erosion index for a given storm (and) is a
product of the kinetic energy of the falling raindrops and its
maximum 30 minute intensity. The sum of these EI values
over a year divided by 100 gives the annual R factor”3.
Camirand (2005), after reviewing past calculations of
R-Factors for several locations in Jamaica, plotted these
factors against mean annual rainfall which resulted in a
high correlation (R2 = 0.8113). However, when compared
against similar work done in the United States, Australia
and Puerto Rico, he concluded that his calculations were
too high. He attributed the higher values to too few years
of measurements. The Puerto Rican model was therefore
adopted for Jamaica and used to successfully calculate the
R-Factor for the Buff Bay/Pencar Watershed Management
Unit in Jamaica. The Puerto Rican equation is given as (Del
Mar López et al. 1998):
planning. For this assessment, the Revised Universal Soil
Loss Equation (RUSLE) was used (Van Remortel et al.
2001). The equation incorporates six factors which influence
soil erosion as outlined by Wischmeier and Smith (1978 cited
in RUSLEFAC 1997);
A = R x K x L x S x C x P, where,
A = average annual soil loss rate (in tonnes per hectare
per year);
R = rainfall erosivity factor (MJ-mm/ha-h-y);
K = soil erodibility factor (t-ha-h/ha-MJ-mm);
L and S = the slope length and gradient factors,
(dimensionless);
C = crop management factor or vegetative cover
(dimensionless) and
P = erosion control practice factor (dimensionless).
The revision replaces two factors: the vegetative cover (C)
and control practice (P), with a vegetation management
factor (VM) which is defined as the “…ratio of soil loss from
land managed under specific conditions to that from the
fallow condition on which the factor K is evaluated”2.
2
3
R = -11.06 + 0.2629 * P (R2 = 0.9511)
Where, P = mean annual precipitation in mm
K-Factor
The K-Factor is “the rate of soil loss per unit area as measured
on a 3.7m x 22m plot. ‘K’ is a quantitative measure of a soil’s
inherent susceptibility/resistance to erosion and the soil’s
influence on runoff amount and rate” (RUSLEFAC, 1997).
The K value can be calculated for a specific soil, using
the following equation (Wischmeier et al. (1969 cited in
Camirand, 2005)):
100 K= 2.1 M1.14 (10-4) (12 - a) + 3.25(b - 2) + 2.5(c - 3)
where:
M = (silt (%) + very fine sand (%)) x (100 - percent clay)
a = percent organic matter
b = the soil structure code used in soil classification
c = profile permeability class.
Wischmeier et al. (1971) cautioned however that the formula
http://www.fao.org/docrep/006/T0099E/T0099e01.htm (7 January 2007).
http://pasture.ecn.purdue.edu/~sedspec/sedspec/doc/usleapp.doc (7 January 2007).
Utilizing GIS for minimizing runoff in a watershed in Jamaica
381
TABLE 3 Area of land use types in the Rio Minho Watershed (2001)
Type (Code)
Forest ( > 75 %, Minimum unit: 25 ha )
Closed Broadleaf (PF)
Disturbed Broadleaf (SF)
Bamboo (BB)
Tall Open Dry (WL)
Short Open Dry (SL)
Mangrove (MG)
Total (Ha)
%
Sub-total
26.0
5999.8
146.8
1912.8
584.2
84.1
8753.7
0.05
11.49
0.28
3.66
1.12
0.16
16.76
Sub-total
9741.7
1146.1
53.4
7306.5
177.8
18425.6
18.65
2.19
0.10
13.99
0.34
35.28
Sub-total
TOTAL
6546.9
14686.0
121.7
2978.3
614.5
25047.3
52226.6
12.54
28.12
0.42
5.70
1.18
47.96
100.00
Mixed Forest
Disturbed Broadleaf Forest and Fields (SC)
Bamboo and Disturbed Broadleaf Forest (BF)
Bamboo and Fields (BC)
Fields and Disturbed Broadleaf Forest (CS)
Bauxite Extraction and Disturbed Broadleaf Forest (BS)
Non Forest
Plantations (PC)
Fields (FC)
Water Bodies (WA)
Buildings and Other
Infrastructure (BA)
Residential - High Density (BH)
Residential - Low Density (BL)
TABLE 4 Guidelines for assessing potential forest land
SOIL DEPTH
Gently to Moderately
Sloping
<15° (<27%)
SLOPE
Strongly Sloping to
Moderately Steep
15º - 30º (27% - 58%)
Steep to Very Steep
>30º (>58%)
Deep (D)
1 ( FI/P, C )
(>100cm)
6 ( FS, FP, AF )
Moderately Deep (MD)
2 ( FI/P, AF, C )
(50-100cm)
Shallow(S)
3 ( FI, AF, C )
4 ( FI, AF, PA )
20-50cm)
7 ( FP )
Very Shallow(S)
5 ( FS, AF, PA
6 ( FS, FP, AF )
(<20 cm)
Potential Land Use:
FI/P - Forest for industrial production, including intensive site preparation and plantation establishment; possible mechanisation
Forest for industrial production (e.g. selective cutting, enrichment planting, seeding and coppicing); possible mechanisation,
FI but normally excluding intensive site preparation and plantation preparation and plantation establishment.
Selection forest for environmental protection and limited wood extraction: selective logging only, no clear-cutting, no road
FS construction, no mechanised site preparation, no mechanised ground skidding
Protection forest for watershed management, ecosystem protection, and/or recreation: no road construction, no timber
FP extraction.
Agroforestry: trees or shrubs grown in association with herbaceous plants under an approved system involving soil
AF conservation measures or
PA - Pastures.
C - Cultivable land.
Source: Modified from Camirand and Evelyn (2003) and Forestry Department (2001)
the formula is not valid for soil with sand > 65% or clay
> 35% and this would apply to several of the soils in the
The K-Factor is “the rate of soil loss per unit area as measured
Rio Minho watershed. An alternative to the equation is a
on a 3.7m x 22m plot. ‘K’ is a quantitative measure of a
nomogram based on soil texture, organic matter content, soil
soil’s inherent susceptibility/resistance to erosion and the
structure and soil permeability (Foster et al. 1981). Using
382 O.B. Evelyn
soil’s influence on runoff amount and rate” (RUSLEFAC,
this nomogram, K-Factors were calculated for the forty-four
(44) soils types found in the Rio Minho watershed.
“is calculated as the product of the slope length and
FIGURE 5 Process flow for creating potential land use
L
and S-Factors
steepness
constituents converging onto a point of interest”.
RUSLEFAC (1997) also stated that “the LS factor represents
van
et under
al. (2001)
stated
that the
LSoffactor
a ratioRemortel
of soil loss
the given
conditions
to that
a site
“is
as the
of of
the9%slope
lengthlength
and
withcalculated
the ‘standard’
slopeproduct
steepness
and slope
steepness
constituents converging onto a point of interest”.
of 22.13 m”.
RUSLEFAC
stated
that “the LS factor
represents
Without (1997)
doing also
field
measurements,
a GIS-based
aapplication
ratio of soilwas
lossproduced
under thetogiven
conditions
thatSoffactors
a site
generate
the Ltoand
from a DEM. This application is available as an ArcGIS
extension4.
TABLE
useland
by present
land
use in
the Rio Minho Watershed
FIGURE56Potential
Potentialland
forest
use – Rio
Minho
Watershed
K-Factor
Land use Code
1
2
Forest ( > 75 %, Minimum unit: 25 ha )
PF
26,0
SF
1613,9
181,8
BB
28,2
6,5
WL
235,9
SL
297,2
0,3
MG
84,1
Mixed Forest
SC
2001,2
383,7
BF
293,7
70,4
BC
0,6
CS
1518,0
157,4
BS
34,6
8,3
Non Forest
PC
5739,3
29,7
FC
7183,4
376,0
WA
221,7
BH
1550,7
7,4
BL
300,9
12,2
TOTAL (HA)
21129,3
1233,9
3
1033,4
25,2
7,1
3,3
2488,5
394,7
2,5
1484,1
44,3
420,3
2180,4
401,3
173,8
8658,9
Modified from Morgan, (2005)
is not valid for soil with sand > 65% or clay > 35% and
this would apply to several of the soils in the Rio Minho
watershed. An alternative to the equation is a nomogram
based on soil texture, organic matter content, soil structure
and soil permeability (Foster et al. 1981). Using this
nomogram, K-Factors were calculated for the forty-four
(44) soils types found in the Rio Minho watershed.
C and PLand
Factors
Potential
use
4
5
6
7
TOTAL
The C and P factors were replaced with a vegetation
management factor (VM). Camirand (2005) observed that
26,0of the
“originally the VM-factor was determined on the basis
1338,7
645,1
512,0
674,9
5999,9
product of three forest effects on erosion, namely vegetative
62,5
24,0
canopy, ground cover and0,3
roots and residues”.
The146,8
VM was
expanded 1611,9
later to include
9
other
factors.
Adopting
some
50,7
7,3
1912,8
of the values
derived
by
Camirand
(2005),
the
VM-Factor
276,7
6,8
584,3
values were calculated for the land use types found in the
84,1
Rio Minho watersh
A qualitative ranking system according to soil loss was
2818,1
551,7
442,1 Morgan1056,5
adopted and
modified from
(2005) to fit9741,7
the soil
erosion classes in the RUSLE
for applications
in
Canada
306,9
11,0
69,3
1146,1
(RUSLEFAC
1997).
This
system
places
greater
emphasis
30,6
0,6
19,2
53,4
on the relative implications of soil loss (e.g. severe vs.
2333,2
664,3
323,1
826,2
7306,5
negligible impact) and less on the actual calculated soil loss
77,7
8,4are defined177,8
rate. Five erosion classes 4,5
were used and
in Table
6 (Morgan 2005).
206,1 The areas
114,5
22,3 classes 14,9
6547,0
by soil erosion
and land use
(Figure
8
and
Table
7)
were
calculated
using
GIS
according
to the
2650,0
766,9
395,3
1133,9
14685,9
processes outlined above.
221,7
Most of the forested areas (Table 7) have low to very low
4,3
1006,4
8,2
2978,2
erosion potential. However, a significant area in the mixed
30,3
79,1
14,8
3,2
614,4 soil
forest and non forest areas have moderate to severe
9858,3
5716,6
1791,7
3837,8
52226,6
erosion potential.
The RUSLEFAC (1997) stated that “… tolerable soil
loss is the maximum annual amount of soil which can be
removed before the long-term natural soil productivity
of a hillslope is adversely affected”. Camirand (2005)
determined that a tolerable soil loss figure for Jamaica is 10
tonnes per hectares per year. It is also stated that “vegetation
is the ultimate, long-term erosion control”5. Thus the goal
from the above analysis was to convert to Forest, most of
the Mixed Forest and Non Forest areas which have high and
severe soil erosion potential. Table 7 shows that a total of
6 425.8 ha and 13 620.7 ha, respectively for each type had
potential for conversion.
Assessment of Ground Water Regeneration
L and S-Factors
Van Remortel et al. (2001) stated that the LS factor
4
5
There was a water shortage of 58.3 106 m3/a in the year 2000
and it is projected to increase to 161.0 106 m3/a by the year
http://www.yogibob.com/slope/slope.html (31 July 2006).
http://www.ceres.ca.gov/foreststeward/html/unneeded.html (7 January 2000)
Utilizing GIS for minimizing runoff in a watershed in Jamaica
383
TABLE 5 Potential land use by present land use in the Rio Minho Watershed
Land use Code
1
2
Forest ( > 75 %. Minimum unit: 25 ha )
PF
26.0
SF
1613.9
181.8
BB
28.2
6.5
WL
235.9
SL
297.2
0.3
MG
84.1
Mixed Forest
SC
2001.2
383.7
BF
293.7
70.4
BC
0.6
CS
1518.0
157.4
BS
34.6
8.3
Non Forest
PC
5739.3
29.7
FC
7183.4
376.0
WA
221.7
BH
1550.7
7.4
BL
300.9
12.2
100
100
TOTAL (HA)
21129.3
1233.9
3
Potential Land use
4
5
7
512.0
0.3
50.7
6.8
674.9
24.0
7.3
442.1
11.0
0.6
323.1
4.5
1056.5
69.3
19.2
826.2
8.4
9741.7
1146.1
53.4
7306.5
177.8
6547.0
14685.9
221.7
2978.2
614.4
52226.6
1033.4
25.2
7.1
3.3
1338.7
62.5
2488.5
394.7
2.5
1484.1
44.3
2818.1
306.9
30.6
2333.2
77.7
551.7
420.3
2180.4
206.1
2650.0
114.5
766.9
22.3
395.3
14.9
1133.9
401.3
173.8
8658.9
4.3
30.3
9858.3
1006.4
79.1
5716.6
8.2
14.8
1791.7
3.2
3837.8
FIGURE
FIGURE77Process
Processflow
flowfor
fordetermining
determiningpotential
potentialsoil
soilloss
loss
645.1
6
1611.9
276.7
664.3
26.0
5999.9
146.8
1912.8
584.3
84.1
TABLE
TABLE66 Potential
Potentialsoil
soilerosion
erosionclasses
classes
Soil
SoilErosion
ErosionClass
Class
11very
verylow
low(i.e.
(i.e.tolerable)
tolerable)
22Low
Low
33Moderate
Moderate
44High
High
55Severe
Severe
2015
(Table
8) in theslope
Rio
Minho
Watershed
Unit
with
steepness
of
and
with the
the ‘standard’
‘standard’
slope
steepness
of 9%
9%Management
and slope
slope length
length
(WMU).
This
indicates
that
the
demand
is
increasing
and
the
of
22.13
m”.
of 22.13 m”.
greater
need
is
for
ground
water.
Maintaining
ground
water
Without
doing
field
measurements,
a
GIS-based
Without doing field measurements, a GIS-based
isapplication
therefore was
critical
in the Unit
and it could
argued
that
application
to
the
and
SS factors
was produced
produced
to generate
generate
the LLbe
and
factors
the
water
balance
needs
to
be
tipped
more
to
ground
water
from
a
DEM.
This
application
is
available
as
an
ArcGIS
from a DEM. This application is available as an ArcGIS
regeneration
extension
extension44 through the reduction of runoff. This indicates
some sort of vegetative cover but evapotranspiration is quite
large
which suggests that more tree cover would
C
PPFactors
Cand
and(68%)
Factors
negate any increase in ground water (WRA 1990). This is a
dilemma
and P
could
be anwere
area for
furtherwith
research.
The
replaced
The CC and
and
P factors
factors
were
replaced
with aa vegetation
vegetation
management
factor
(VM).
Camirand
(2005)
management factor (VM). Camirand (2005) observed
observed that
that
“originally
“originallythe
thevM-factor
vM-factorwas
wasdetermined
determinedon
onthe
thebasis
basisof
ofthe
the
product
productof
ofthree
threeforest
foresteffects
effectson
onerosion,
erosion,namely
namelyvegetative
vegetative
canopy,
canopy,ground
groundcover
coverand
androots
rootsand
andresidues”.
residues”.The
ThevM
vMwas
was
expanded
later
to
include
9
other
factors.
Adopting
expanded later to include 9 other factors. Adopting some
some
TOTAL
Potential
PotentialSoil
SoilLoss
Loss
(tonnes/hectare/year)
(tonnes/hectare/year)
<<22
22--55
55--10
10
10
10--50
50
>>50
50
Grabaun
and Meyer
(1998) indicated
that productivity
to
maintain
removed
before
the
natural
removed
before
the long-term
long-term
natural soil
soil
productivity
ground
water
regeneration
at
a
quantitatively
high
level,
of
a
hillslope
is
adversely
affected”.
Camirand
of a hillslope is adversely affected”. Camirand (2005)
(2005)
lands
with
soils
which
are
permeable
and
are
in
agricultural
determined
that
a
tolerable
soil
loss
figure
for
Jamaica
determined that a tolerable soil loss figure for Jamaicaisis10
10
use
should
not
be converted
determine
the areas
tonnes
per
per
ItItforest.
isisalso
stated
that
tonnes
perhectares
hectares
peryear.
year.to
alsoTo
stated
that“vegetation
“vegetation
55
which
meet
these
criteria, erosion
the
soilscontrol”
of the Rio
Minho
.. Thus
the
goal
is
long-term
Thus
thewere
goal
is the
the ultimate,
ultimate,
long-term
erosion
control”
classified
according
to
the
USDA
Soil
Survey
Hydrologic
from
the
above
analysis
was
to
convert
to
Forest,
most
from the above analysis was to convert to Forest, most of
of
Groups
(USDA-SCS
The soils
and
the land
layers
the
Forest
Non
areas
which
have
high
and
theMixed
Mixed
Forestand
and1972).
NonForest
Forest
areas
which
haveuse
high
and
TM
were
then
using ArvGIS
determine
areas
severe
soil
erosion
Table
shows
that
total
of
severe
soilintersected
erosion potential.
potential.
Table 77to
shows
that aathe
total
of
should
be
left
unconverted.
The
USDA
classes
were
6which
425.8
ha
and
13
620.7
ha,
respectively
for
each
type
had
6 425.8 ha and 13 620.7 ha, respectively for each type had
converted
to aconversion.
more meaningful grade as follows: A – High, B
potential
potentialfor
for
conversion.
– Medium, C – Low and D – Very Low (Figure 9 and Table 9).
The results
suggest Water
that an
area greater than 5 700
Assessment
of
Regeneration
Assessment
ofGround
Ground
Water
Regeneration
hectare (CS) in the Mixed Forest and an area greater than
66 33Forest should not
8There
900 was
hectare
(PCshortage
and FC)of
the10
Non
There
aawater
58.3
m
was
water
shortage
ofin
58.3
10
m /a/ain
inthe
theyear
year2000
2000
66some
33 measure of
be
converted
to
forest.
These
figures
put
and
it
is
projected
to
increase
to
161.0
10
m
/a
by
and it is projected to increase to 161.0 10 m /a by the
the year
year
constraint
on8)
areas
for conversion
to forest
in
2015
in
Rio
Minho
Management
Unit
2015(Table
(Table
8)the
inthe
the
Rioavailable
MinhoWatershed
Watershed
Management
Unit
these
classes
suggested
by
the
previous
assessments.
(WMU).
This
indicates
that
the
demand
is
increasing
and
the
(WMU). This indicates that the demand is increasing and the
greater
greater need
need isis for
for ground
ground water.
water. Maintaining
Maintaining ground
ground water
water
isis therefore
critical
in
the
Unit
and
it
could
be
therefore critical in the Unit and it could be argued
argued that
that
the
the water
water balance
balance needs
needs to
to be
be tipped
tipped more
more to
to ground
ground water
water
regeneration
regeneration through
through the
the reduction
reduction of
of runoff.
runoff.This
This indicates
indicates
some
sort
of
vegetative
cover
but
evapotranspiration
some sort of vegetative cover but evapotranspirationisisquite
quite
large
large (68%)
(68%) which
which suggests
suggests that
that more
more tree
tree cover
cover would
would
384
O.B. Evelyn
101
FIGURE 8 Potential soil erosion
FIGURE 8 Potential soil erosion
Non Forest (NF) were used in the model to simplify the
process.
In setting up their formulation, Shukla et al. (2001) gave
the following guidance:
Each constraint (should be) formulated as a linear
equation, whose right hand side constant represents
the limit of the available resources ….the objective
functions and all the constraints must be strictly
linear over the domain of each activity ….each
linear variable can assume any real value including
both real and integers and fractions. All right hand
side values are assumed to be known constant.
All (values) must be at least equal to zero that is
negative assignment should not be included in the
model.
They also used, “coefficients of the objective functions”
which are “secondary objectives (which) act as a set of
constraints in limiting the value of the objective function”
(ibid). No coefficients were used in this analysis. The
problem was therefore formulated as follows:
Objective Function:
Max {(F to F) + (M to F) + (NF to F)}
Subject to the following constraints:
Objective function ≥ 0
Conservation forest:
F to F ≤ 8 606.90 ha.
Protective forest to minimize erosion:
M to F ≤ 6 425.80 ha.
NF to F ≤ 4 537.40 ha.
F to M ≥ 146.80 ha.
M to M ≥ 11 999.80 ha.
F to NF ≥ 0
Assessment of Domestic Food Crop Production
For this assessment, only the domestic food crop production
will be assessed. The Ministry of Agriculture does annual
assessments of the domestic food crop production by
parishes. In the 2003 report, the parish of Clarendon
produced 45 543 tonnes of food crops on a total area of 2
876 ha. This amounts to 9% of the Island’s domestic food
crop production. This area should be retained but not on
steep degraded lands. The high erosion figures stated above
could account for this low production.
Ground water regeneration and crop production:
M to NF ≥ 0
NF to NF ≤ 10 361.38 ha.
NF to M ≤ 10 148.52 ha.
Forest:
(F to F) + (F to M) + (F to NF) ≤ 8 753.7 ha.
Mixed:
(M to F) + (M to M) + (M to NF) ≤ 18 425.60 ha.
Non-Forest:
(NF to F) + (NF to M) + (NF to NF) ≤ 25 047.30 ha.
Forest Optimization
Multi-Objective/Multi-Criteria
Decision
Making
optimization modelling techniques which involve some
amount of linear programming methodologies was used to
determine the forest optimization. In a linear programming
problem, an objective function and the decision variables
are first defined. The problem is then formulated so that the
objective function can be either maximized or minimized
while satisfying a set of constraints.
The GIS assessments suggest the possible conversion
matrix (Table 10). This matrix was used as the decision
variables in the linear programming optimisation. The
broad land use groups Forest (F), Mixed Forest (M) and
The problem was set up in an Excel 2000 spreadsheet
(Figure 10) and the Excel Solver was used to calculate the
optimized forest cover value.
Results
After several iterations, the Solver reported that it found the
globally optimal solution, i.e. there was no other solution
satisfying the constraints which has a better value for the
Utilizing GIS for minimizing runoff in a watershed in Jamaica
102
385
TABLE 7 Potential soil loss area by erosion classes and land use
Potrntial Soil Eroion Classes (tonnes/hectares/year)
Very Low
Low
Moderate
High
Land Use Code
<2
2-5
5 - 10
10 - 50
Forest ( > 75 %, Minimum unit: 25 ha )
PF
25.9
0.0
0.0
0.0
SF
4168.9
1112.3
588.3
96.7
BB
0.6
0.3
64.0
0.6
WL
1911.0
1.1
0.1
0.6
SL
583.1
0.2
0.3
0.4
MG
84.1
0.0
0.0
0.0
Mixed Forest
SC
9704.3
1.9
10.5
6.6
BF
1142.6
0.1
0.9
0.7
BC
53.0
0.1
0.1
0.1
CS
271.3
3.6
605.8
1782.3
BS
176.7
0.1
0.3
0.2
Non Forest
PC
537.4
112.3
1359.8
4179.1
FC
399.5
710.8
4492.4
3393.4
WA
221.0
0.0
0.0
0.7
BH
2883.8
75.8
13.4
4.1
BL
483.4
78.6
42.9
6.6
TOTAL (HA)
22646.8
2096.9
7178.7
9472.1
Severe
>50
Total (HA)
0.0
33.6
81.3
0.0
0.2
0.0
26.0
5999.8
146.8
1912.8
584.2
84.1
18.5
1.8
0.1
4643.5
0.5
9741.7
1146.1
53.4
7306.5
177.8
358.3
5689.9
0.0
1.3
2.9
10832.0
6546.9
14685.9
221.7
2978.3
614.4
52226.6
TABLE 8 Annual water resource and demand 2000 – Rio Minho WMU
Annual Water Resources and Demand - Rio Minho WMU (106 m3/a )
Rainfall
2420.0
Evapotranspiration
1641.0
Water Resources
Surface
Water
225.0
Ground
Water
554.0
Exploitable Water
Total
779.0
Surface
Water
32.0
Ground
Water
439.0
Water Demand - 2000 (Mcm/yr)
Total
Irrigation
Industrial
Domeatic
Total
471.0
486.0
19.0
24.3
529.3
Source: WRA Water Resources Development Master Plan for Jamaica. 1990
Assessment of Domestic Food Crop Production
objective (Target Cell)6.
reports were
Answer,
Sensitivity
and
For Three
this assessment.
onlyproduced:
the domestic
food crop
production
Limits.
Answer
(Table
11) shows does
the optimal
will
be The
assessed.
TheReport
Ministry
of Agriculture
annual
value of the objective
function food
and the
decision
variables
assessments
of the domestic
crop
production
by
and how theInconstraints
optimalofvalues.
The
parishes.
the 2003 affected
report. these
the parish
Clarendon
solution suggested
that of
thefood
optimal
that area
couldofbe
produced
45 543 tonnes
cropsvalue
on a total
2
achieved
wasamounts
19 570.10
hectares
Target
Cell, Final
876
ha. This
to 9%
of the (see
Island’s
domestic
food
Value).
This increases
forests from
16.76
crop
production.
This the
areapercentage
should beofretained
but not
on
% to 37.47 %.
Theaccount
Adjustable
section of the Answer Report
could
for thisCells
low production.
showed the derived land use allocations. For example, the
Non-Forest
area has been significantly reduced by 56.91
Forest
Optimization
% from 25 047.3 hectares (Table 3) to 10 361.38 hectares
(Adjustable Cells, Cell $D$48, FinalDecision
Value, Table 11).
Multi-Objective/Multi-Criteria
Making
optimization modelling techniques which involve some
amount of linear programming methodologies was used to
determine
the forest optimization. In a linear
programming
6
http://www.solver.com/suppstdmsgresult2.htm
(8 January
2007).
problem. an objective function and the decision variables
The Sensitivity Report (Tables 12) provided useful
information
for interpreting
the results
and indicated
how
objective
function
can be either
maximized
or minimized
changing
the constraints
or the objective function would
while
satisfying
a set of constraints.
change
the results.
This suggest
report the
provided
twoconversion
types of
The GIS
assessments
possible
information:
sensitivity
of the
solution
the decision
decision
matrix
(Tablethe
10).
This matrix
was
used asto the
andthe
thelinear
sensitivity
of the solution
to the constraints.
variables in
programming
optimisation.
The broad
For example,
the report
onlyForest
very small
land
use groups
Forestshows
(F). that
Mixed
(M) decreases
and Non
were allowed
in the
value
of model
F to F,toM-to-F
andtheNF-to-F
Forest
(NF) were
used
in the
simplify
process.if
the In
current
to remain Shukla
optimal.et al. (2001) gave
settingsolution
up theirwas
formulation.
The Limitsguidance:
Report (Tables 13) indicates how changing
the following
the valueEach
of aconstraint
decision (should
variable be)
impacts
the solution.
The
formulated
as a linear
Lower equation.
Limit column
the smallest
valuerepresents
that the
whose gave
right hand
side constant
decisionthe
variable
be without
changing
other
limit ofcould
the available
resources
….theany
objective
decisionfunctions
variables and
or violating
any constraints.
Upper
all the constraints
must The
be strictly
linear over the domain of each activity ….each
linear variable can assume any real value including
both real and integers and fractions. All right hand
386
O.B. Evelyn
103
FIGURE 9 Ground watter regeneration
FIGURE 9 Ground water regeneration
Limit showed the largest value that the decision variable
could be without changing any other decision variables or
violating any constraints, while the Target Result column
showed what the objective value would be if this decision
variable took on a limiting value and the other decision
variables retained the values they had at the optimal solution.
The solution is very tight with most of the adjustable cells
fixed at their target values.
The values from the analysis in Tables 11 - 13 were used
to reconstruct an optimized land use distribution pattern for
the watershed. The GIS assessments layers were overlayed
and intersected to derive the disaggregated land use classes
that are shown in a visual form (Figure 11). These optimized
classes along with the original 2001 classes were used in the
hydrological modelling exercise.
HYDROLOGICAL
METHODOLOGY
MODELLING
AND
TESTING
For the hydrological simulation, the semi-distributed
hydrologic model Soil and Water Assessment Tool (SWAT)
was used. The main reasons this software was chosen was
that the data required to run it was available, it is designed to
simulate ungaged basins (a common situation in Jamaica) and
therefore “does not require calibration” (Arnold et al. 1998)
and it was freeware.
SWAT is integrated with several readily available
Geographic Information Systems such as ArcViewTM 3.X and
GRASS. The version of SWAT that was used in this project
is the SWAT2005 with its ArcViewTM interface AVSWAT–X
(Di Luzio et al. 2005). AVSWAT-X takes advantage of
some of ArcViewTM extensions and computation capabilities
and its Windows-based user interface. This includes the
creation of river networks, catchments and sub-basins using
Spatial Analysis. The model utilizes a ‘daily time step’ and
is developed to simulation processes over long time periods
(Arnold et al. 1998).
The basic data inputs required for AVSWAT-X to perform the
hydrological modelling was a DEM to delineate the watershed
boundary, a stream network layer to help define the streams,
a digital soil layer, a land use layer to define the Hydrologic
Response Units (HRUs), and empirical rainfall and weather
data to predict precipitation. Weather data, weather stations
location and daily rainfall, obtained from the Meteorological
Office in Jamaica were imported to the model. The format and
assembly of these data are described in the SWAT user manual
(Neitsch et al. 2002).
For the project area, ten sub-basins were delineated by
AVSWAT-X (Figure 12). On completion, a Watershed View
was created and the following themes were added to it: Subbasins, Streams and Outletst. A Topographic Report containing
a detailed summary by sub-basin of distinct locations in the
watershed was also added to the current ArcViewTM project.
Each sub-basin in the Sub-basins theme was numbered.
The names and areas of the ten sub-basins that were created
in the delineation process are outlined in Table 14.
Of the 10 sub basins, runoff analysis was carried out on
numbers 1, 2, 3, 4 and 10 because the forest optimization
process resulted in changes in these basins.
After the land use and soils theme databases were loaded,
“Reclassified” and overlayed, the HRUs were determined
by AVSWAT-X in order to capture differences in variations
within the spatial units. This way the runoff can be spatially
determined and summed for the entire watershed to give a more
accurate picture of the hydrological processes that are taking
place (Stahr and Gaiser, 2004). Neitsch et al. (2001) indicates
that “HRUs are intended to be summed areas of similar land
use/land cover and soils within a subbasin” .
On completion of the HRU distribution process, a SWAT
View (Figure 13) was automatically generated and the option
to input the weather data activated.
The sole purpose of the hydrological simulation in this
project was to test for significant differences between the
runoff from the two land uses: the 2001 land use and the
derived optimized land use. Therefore it was not necessary to
calibrate the model since only a comparison between the two
sets of runoff was required.
Results
After all the required data was entered into SWAT and
validated by the software, the simulation option was activated.
Utilizing GIS for minimizing runoff in a watershed in Jamaica
104
387
TABLE 9 Ground water regeneration
Land Use Code
HIGH
Forest ( > 75 %, Minimum unit: 25 ha )
PF
0.0
SF
4089.5
BB
110.3
WL
1767.0
SL
337.0
MG
76.0
Mixed Forest
SC
7341.2
BF
781.0
BC
52.3
CS
5798.4
BS
129.8
Non Forest
PC
1992.3
FC
6999.3
WA
0.0
BH
1122.1
BL
279.0
TOTAL (HA)
30875.4
Ground Water Regeneration
MEDIUM
LOW
VERY LOW
0.0
659.6
17.9
0.0
1.0
0.0
17.0
982.4
7.5
60.3
26.3
3.8
8.9
268.2
11.1
85.4
219.9
4.4
26.0
5999.7
146.8
1912.8
584.2
84.1
1208.0
247.0
0.0
649.2
19.6
795.5
104.9
1.1
362.7
0.2
397.0
13.3
0.0
496.2
28.2
9741.7
1146.1
53.4
7306.5
177.8
176.2
1977.7
0.0
111.8
134.1
5201.9
1708.1
1934.7
33.1
24.3
107.2
6169.2
2670.3
3774.4
188.7
1720.0
94.1
9980.1
6546.9
14686.1
221.7
2978.3
614.4
52226.6 105
values are
assumed
be known to
constant.
TABLE side
10 Suitable
areas
(ha) fortoconversion
forest All
(values) must be at least equal to zero that is negative
assignment should not be included
in the model.
FROM/TO
FOREST
(F)
FOREST (F)
8753.70
MIXED (M)
6425.80
which are “secondary objectives (which) act as a set of
NON-FOREST (NF)
4537.40
constraints in limiting the value of the objective function”
(ibid
FIGURE
10 therefore
Solver
in the
Excel
2000
problem
was
formulated
as
follows:
Daily
simulation
wassetup
run
for
period
May 1, 2002 to May
31, 2002. SWAT is designed to run for a longer period but
Objective
continuous
data forFunction:
the watershed was limited. For surface
{(F rain/CN/Daily”
to F) + (M to F)option
+ (NFwas
to F)}
runoff theMax
“Daily
selected for the
Subject
the following
constraints:
precipitation
timetostep,
runoff calculation
method and routing
time step.Objective function ≥ 0
A series of reports were produced in dbase table format after
Conservation
forest:
the simulation
was run successfully.
A display of the simulated
F
to
F
≤
8
606.90
ha. can be easily modified. The
runoff was then presented which
before and after runoff for all the sub-basins are also present in
forest14).
to minimize erosion:
histogramProtective
format (Figure
M
to
F
≤
6
425.80
ha. to whole numbers) runoff
The percentage (rounded
NF
to
F
≤
4
537.40
ha. by the optimized land use
reductions that were achieved
F
to
M
≥
146.80
ha.
was calculated and presented by sub-basins and showed in
Table 15.M to M ≥ 11 999.80 ha.
F to NF number
≥0
A significant
of the days in the sub-basins where
Ground
water
regeneration
and crop
production:
the treatment was applied
have reduced
runoff
above 10%.
M
to
NF
≥
0
Some experienced runoff reductions of 100% on some days.
NF to NF ≤ 10 361.38 ha.
is missing
76
TOTAL (HA)
http://www.solver.com/suppstdmsgresult2.htm
January 2007)
http://faculty.vassar.edu/lowry/wilcoxon.html
(6 (8
September
2006).
NF to M ≤ 10 148.52 ha.
Forest: (M)
MIXED
NON-FOREST (NF)
(F to- F) + (F to M) + (F to NF) ≤ 8 753.7
ha.
Mixed:
11999.8
(M to F) + (M to M) + (M to NF) ≤ 18 425.60 ha.
20509.90
Non-Forest:
(NF to F) + (NF to M) + (NF to NF) ≤ 25 047.30 ha.
It should be noted however, that some of the values which are
problem
was setin up
in anwhen
Excel
0% The
showed
an increase
runoff
the2000
runoffspreadsheet
values are
(Figure
10) than
and the
ExcelofSolver
wasAlso,
usedjust
to calculate
the
set to more
4 places
decimal.
as there was
optimized
coverflow,
value.
a decrease forest
in surface
there was also an increase in base
flow. SWAT produced a report for this but further analysis was
Results
not done by the researcher because it was beyond the scope of
this research.
After
several
iterations.
Solver reported
found the
The
next step
was tothe
establish
whether that
theseit reductions
globally
optimal
solution.
i.e.
there
was
no
other
solution
were statistically significant. A two sample test was therefore
satisfying
the
constraints
which
has
a
better
value
the
conducted on the two sets of runoff data. A Wilcoxon for
signed6
.
objective
(Target
Cell)
rank test was chosen instead of a two-sample t-test or a MannThreeUreports
were produced:
Answer.
Whitney
Test because
the runoff data
from Sensitivity
the two setsand
of
Limits.
The
Answer
Report
(Table
11)
shows
the present).
optimal
data did not have a normal distribution (outliers were
value
the values
objective
function
and
the decision
variables
Theofmean
and the
p-values
(Table
16) were calculated
and
how
the
constraints
affected
these
optimal
values.
The
after submitting the two runoff datasets produced by SWAT
suggested
that
the
optimal
value
that
could
be
solution
to the Wilcoxon signed-rank test at the 95% confidence level
®
achieved
was
19
570.10
hectares
(see
Target
Cell.
Final
using the Analyse-it statistical software add-in for Microsoft®
7
Value).
This
increases
of forests
from 16.76
Excel® for
Windows
andthe
thepercentage
on-line VassarStats
software
.
% to 37.47 %.
388
O.B. Evelyn
FIGURE 10 Solver setup in Excel 2000
ACKNOWLEDGEMENT
Special thanks to by friends and colleagues, who are too
numerous to mention, for their valuable advise and support.
I am truly grateful.
REFERENCES
For all of the basins, the P values for both the 1–tail and
2–tail tests are smaller than the significance level (p<0.05).
The null hypothesis that there is no significant difference
between the runoffs from the two sets of land uses was
therefore rejected.
CONCLUSION
The land use optimization model suggested a spatial pattern
in the upper watershed which had the following features: 1)
retain as conservation forests, forested areas on steep slopes
and shallow soils, 2) convert to forest, areas being used for
agriculture on steep slopes and shallow soils and areas with
high to severe erosion problems, 3) preserve areas with high
agricultural production and high ground water regeneration
potential. Features 1 and 2, if implemented, would help to
solve some of the problems being faced in the watershed.
While the problem of determining the optimum forest
cover and the desired spatial distribution to minimize runoff
was solved by this study, implementation of the conservation
and reforestation measures may prove to be a challenge since
it would require some amount of dislocation. Currently,
land use practices on the upper slopes of the watershed are
mainly agricultural crop production and are done mostly on
small holdings by subsistent farmers.
The results of the hydrological modelling point to
significant reductions in runoffs in the upper watersheds
where forest cover is restored and the NRCA indicated that
there were medium to severe degradation problems.
AGARWAL, C., GREEN, G. M., GROVE, J. M., EVANS,
T. P. and SCHWEIK, C. M. 2001. A Review and
Assessment of Land-Use Change Models: Dynamics
of Space, Time, and Human Choice. Bloomington
and South Burlington, Center for the Study of
Institutions, Population, and Environmental Change,
Indiana University and USDA Forest Service. CIPEC
Collaborative Report Series 1.
ARNOLD, J. G., SRINIVASAN, R., MUTTIAH, R. S.
and WILLIAMS, J. R. 1998. Large Area Hydrologic
Modeling and Assessment: Part I. Model Development.
Journal of the American Water Resources Association.
34(1): 73-89.
BRIASSOULIS, H. 1999. Analysis of Land Use Change:
Theoretical and Modeling Approaches. Regional
Research Institute, West Virginia University.
BURT, T. and SWANK, W. 2002. Forests or Floods?
Geography Review. 15(5): 37-41.
CAMIRAND, R. and EVELYN, O.B. 2003. Ecological
land classification for forest management and
conservation in Jamaica. Jamaica-Canada Trees for
Tomorrow Project Phase II, Forestry Department and
Tecsult International, Kingston, Jamaica. 40 p.
CAMIRAND, R. 2005. Water and Soil Conservation
Plan for Buff Bay Pencar Watershed Management
Unit, Jamaica. Draft Report Prepared for the Forestry
Department, Ministry of Agriculture Kingston,
Jamaica. Wood and Forest Sciences Department Laval
University, Québec, Canada December 2005
DEL MAR LÓPEZ, T., MITCHELL AIDE, T. and
SCATENA, F.N. 1998. The effect of land use on soil
erosion in the Guadiana watershed in Puerto Rico.
Caribbean Journal of Science 34(3-4): 298-307.
DI LUZIO, M., MITCHELL, G. and SAMMONS, N.
2005 AVSWAT-X short tutorial. Third Conference
on Watershed Management to Meet Water Quality
Standards and Emerging TMDL, March 5 - 9, 2005.
Sheraton Atlanta, Georgia, USA.
EVELYN, O. and CAMIRAND, R. 2003. Forest cover and
deforestation in Jamaica: An analysis of forest cover
estimates over time. International Forestry Review 5(4)
354-363.
EYRE, L. A. 1986. Deforestation in Jamaica: its rate and
implications. Department of Geology, University of the
West Indies, Kingston, 29 pp
EYRE, L.A. 1987. Jamaica: a test case for tropical
deforestation. Ambio. 16(6) 336-343.
FAO. 1985. Guidelines: land evaluation for irrigated
Utilizing GIS for minimizing runoff in a watershed in Jamaica
389
TABLE 11 Excel Solver Answer Report
Target Cell (Max)
Cell
$B$49
Adjustable Cells
Cell
$B$46
$C$46
$D$46
$B$47
$C$47
$D$47
$B$48
$C$48
$D$48
Constraints
Cell
$B$49
$B$46
$B$46
$C$46
$D$46
$B$47
$B$47
$C$47
$D$47
$B$48
$B$48
$C$48
$D$48
$C$46
$C$47
$C$48
$D$46
$D$47
$D$48
Name
TOTAL FOREST
Original Value
28800.20
Final Value
19570.10
Name
FOREST FOREST
FOREST MIXED
FOREST NON-FOREST
MIXED FOREST
MIXED MIXED
MIXED NON-FOREST
NON-FOREST FOREST
NON-FOREST MIXED
NON-FOREST NON-FOREST
Original Value
8753.70
0.00
0.00
6425.80
11999.80
0.00
13620.70
0.00
11426.60
Final Value
8606.90
146.80
0.00
6425.80
11999.80
0.00
4537.40
10148.52
10361.38
Name
TOTAL FOREST
FOREST FOREST
FOREST FOREST
FOREST MIXED
FOREST NON-FOREST
MIXED FOREST
MIXED FOREST
MIXED MIXED
MIXED NON-FOREST
NON-FOREST FOREST
NON-FOREST FOREST
NON-FOREST MIXED
NON-FOREST NON-FOREST
FOREST MIXED
MIXED MIXED
NON-FOREST MIXED
FOREST NON-FOREST
MIXED NON-FOREST
NON-FOREST NON-FOREST
Cell Value
19570.10
8606.90
8606.90
146.80
0.00
6425.80
6425.80
11999.80
0.00
4537.40
4537.40
10148.52
10361.38
146.80
11999.80
10148.52
0.00
0.00
10361.38
Formula
$B$49>=$B$67
$B$46<=$B$58
$B$46<=$B$68
$C$46<=$B$68
$D$46<=$B$68
$B$47<=$B$59
$B$47<=$B$69
$C$47<=$B$69
$D$47<=$B$69
$B$48<=$B$60
$B$48<=$B$70
$C$48<=$B$70
$D$48<=$B$70
$C$46>=$B$61
$C$47>=$B$62
$C$48>=$B$63
$D$46>=$B$64
$D$47>=$B$65
$D$48<=$B$66
agriculture. Soil resources development and
conservation service, FAO Land and Water Development
Division, FAO Soils Bulletin 55, FAO, Rome.
FAO. 1996a. Agro-ecological zoning Guidelines. FAO
Soils Bulletin 73, FAO, Rome
FAO. 1996b. E/CN.17/IPF/1996/2 - Report of the Secretary,
General Commission On Sustainable Development,
Ad Hoc Intergovernmental Panel on Forests, Second
session: 11-22 March 1996.
FORESTRY DEPARTMENT. 2001. National forest
management and conservation plan. Ministry of
Agriculture, Kingston, Jamaica. 100 p. and appendices.
Status
Not Binding
Binding
Binding
Not Binding
Not Binding
Binding
Binding
Not Binding
Not Binding
Binding
Binding
Not Binding
Binding
Binding
Binding
Binding
Binding
Binding
Binding
Slack
19570.10
0
0
8606.9
8753.7
0
0
6425.8
18425.6
0
0
14898.78
0
0.00
0.00
0.00
0.00
0.00
0
FOSTER, G.R., Mc. COOL, D. K., RENARD, K.G. and
Moldenhauer, W.C. 1981. Conversion of the universal
soil loss equation to SI metric units, Journal of Soil and
Water Conservation 36(6): 355-359.
GRABAUM, R. and MEYER, B. C. 1998. Multicriteria
optimization of landscapes using GIS-based functional
assessments, Journal: Landscape and Urban Planning,
43(1): 21-34.
IDB. 2006. Information on disaster risk management case
study of five countries - Case study Jamaica. InterAmerican Development Bank & Economic Commission
for Latin America and the Caribbean, 2006.
390
O.B. Evelyn
TABLE 12 Excel Solver Sensitivity Report
Adjustable Cells
Cell
Name
Final Value
$B$46
$C$46
$D$46
$B$47
$C$47
$D$47
$B$48
$C$48
$D$48
Constraints
FOREST FOREST
FOREST MIXED
FOREST NON-FOREST
MIXED FOREST
MIXED MIXED
MIXED NON-FOREST
NON-FOREST FOREST
NON-FOREST MIXED
NON-FOREST NON-FOREST
Cell
Name
Final Value
$B$49
TOTAL FOREST
19Ê570.10
8606.90
146.80
0.00
6425.80
11999.80
0.00
4537.40
10148.52
10361.38
Reduced
Cost
1.00
0.00
0.00
1.00
0.00
0.00
1.00
0.00
0.00
Objective
Coefficient
1
0
0
1
0
0
1
0
0
Allowable
Increase
1.00E+030
0
0
1.00E+030
0
0
1.00E+030
0
1.00E+030
Allowable
Decrease
1
1.00E+030
1.00E+030
1
1.00E+030
1.00E+030
1
1.00E+030
0
Shadow
Price
-
Constraint
R.H. Side
0
Allowable
Increase
19570.1
Allowable
Decrease
1.00E+030
Lower
Limit
-10963.20
146.80
0.00
-13144.30
11999.80
0.00
-15032.70
10148.52
#N/A!
Target
Result
0.00
19570.10
19570.10
0.00
19570.10
19570.10
0.00
19570.10
#N/A!
Upper
Limit
8606.90
8753.70
8753.70
6425.80
18425.60
18425.60
4537.40
25047.30
10361.38
Target
Result
19570.10
19570.10
19570.10
19570.10
19570.10
19570.10
19570.10
19570.10
19570.10
TABLE 13 Excel Solver Limits Report
Cell
$B$49
Target Name
TOTAL FOREST
Value
19570.10
Cell
Adjustable Name
Value
$B$46
$C$46
$D$46
$B$47
$C$47
$D$47
$B$48
$C$48
$D$48
FOREST FOREST
FOREST MIXED
FOREST NON-FOREST
MIXED FOREST
MIXED MIXED
MIXED NON-FOREST
NON-FOREST FOREST
NON-FOREST MIXED
NON-FOREST NON-FOREST
8606.90
146.80
0.00
6425.80
11999.80
0.00
4537.40
10148.52
10361.38
JONES, J. A. and POST, D. A. 2004. Seasonal and
successional streamflow response to forest cutting and
regrowth in the northwest and eastern United States,
Water Resources Research, 40.
MORGAN, R.P.C. 2005. Soil erosion and Conservation.
Third edition, Blackwell Publishing, Oxford, UK.304p
NEITSCH, S.L., ARNOLD, J. G., KINIRY, J. R.,
SRINIVASAN, R. and WILLIAMS, J. R. 2002. Soil and
Water Assessment Tool User’s Manual, Version 2000.
TWRI Report TR-192. Texas Water Resources Institute,
College Station, TX.
NEITSCH, S.L., ARNOLD, J.G., KINIRY, J.R.,
WILLIAMS, J.R. and KING K.W. 2001. Soil and Water
Assessment Tool Theoretical Documentation, Version
2000. Grassland, Soil & Water Research Laboratory,
Temple, Texas GSWRL Report 02-01 and Blackland
Research and Extension Center, Temple, Texas BRC
Report 02-05, pp506.
NRCA. 1997, State of the Environment 1997 Report,
Government of Jamaica
OKUMU, B.N., JABBAR, M.A., COLMAN, D. and
RUSSELL N. 1999. Bio-Economic Modelling
of Watershed Resources in Ethiopia. Paper
presented at the Annual Meeting of the American
Agricultural Economics Association, Nashville,
Tennessee, USA, 8-11 August 1999.
PARKER, D.C., MANSON, S.M., JANSSEN, M.A.,
HOFFMAN, M. and DEADMAN, P. 2002. Multi-Agent
Systems for the Simulation of Land-Use and LandCover Change: A Review. Annals of the Association of
American Geographers, 93(2): 314–337.
RUSLEFAC. 1997. Revised Universal Soil Loss Equation
for Application in Canada. A Handbook for Estimating
Soil Loss from Water Erosion in Canada. (Final version;
based on 1997 draft).
SHAOJUN, C., YUE, W. and YIJIE, W. 2004. The Loess
Plateau Watershed Rehabilitation Project, A case study
from Reducing Poverty, Sustaining Growth. What
Utilizing GIS for minimizing runoff in a watershed in Jamaica
FIGURE 11 Optimized land use classes of Rio Minho
Watershed
FIGURE 11 Optimized land use classes of Rio Minho Watershed
391
TABLE 14 Sub-basins created by SWAT in the delineation
process
107
Sub-basin #
Name
Area (Ha.)
1
Thomas River
7967.8
2
Pindars River
9883.8
3
Suttons
3224.3
109
4
May
Pen
4444.8
indicates that “HRUs are intended to be summed areas of
TABLE 14 Sub-basins created by SWAT in the delineation
land use/land cover and soils within a subbasin” .
process
5
Panassus similar
4657.8
On completion of the HRU distribution
process, a SWAT
Sub-basin #
Name
Area (Ha.)
View (Figure 13) was automatically generated and the option
1
Thomas River
7967,8
to input the weather data activated.
6
Webbers
Gully
1801.9
2
Pindars River
9883,8
The sole purpose of the hydrological simulation in this
project was to test for significant differences between the
3
Suttons
3224,3
7
New
Yarmoth
runoff
from the two land uses: 2492.5
the 2001 land use and the
4
May Pen
4444,8
derived optimized land use. Therefore it was not necessary
5
Panassus
4657,8
to calibrate the model since only a2781
comparison between the
8
Hayes
6
Webbers Gully
1801,9
two sets of runoff was required.
7
New Yarmoth
2492,5
9Hayes
Brokenbank
1924
Results
8
2781
9
Brokenbank
1924
After
all
the
required
data
was
entered into SWAT and
10
Upper
Rio
Minho
13048.5
10
Upper Rio Minho
13048,5
by AvSWAT-X in order to capture differences in variations
within the spatial units. This way the runoff can be spatially
determined and summed for the entire watershed to give a
more accurate picture of the hydrological processes that are
taking place (Stahr and Gaiser, 2004). Neitsch et al. (2001)
validated by the software, the simulation option was
activated. Daily simulation was run for the period May 1,
2002 to May 31, 2002. SWAT is designed to run for a longer
period but continuous data for the watershed was limited.
For surface runoff the “Daily rain/CN/Daily” option was
selected for the precipitation time step, runoff calculation
method and routing time step.
A series of reports were produced in dbase table format
by AVSWAT-X in order to capture differences in variations
FIGURE
Watershed
Viewway
withthe
sub-basins
delineated
within
the13
spatial
units. This
runoff can
be spatially
FIGURE 13 SWAT View
determined and summed for the entire watershed to give a
more accurate picture of the hydrological processes that are
taking place (Stahr and Gaiser. 2004). Neitsch et al. (2001)
108
HYDROLOGICAL
METHODOLOGY
FIGURE 13 SWAT View
MODELLING
AND
TESTING
For the hydrological simulation, the semi-distributed
hydrologic model Soil and Water Assessment Tool (SWAT)
was used. The main reasons this software was chosen was
that the data required to run it was available, it is designed
to simulate ungaged basins (a common situation in Jamaica)
and therefore “does not require calibration” (Arnold et al.
1998) and it was freeware.
SWAT is integrated with several readily available
Geographic Information Systems such as ArcviewTM 3.X and
GRASS. The version of SWAT that was used in this project
is the SWAT2005 with its ArcviewTM interface AvSWAT–X
(Di Luzio et al. 2005). AvSWAT-X takes advantage of some
of ArcviewTM extensions and computation capabilities and
its Windows-based user interface. This includes the creation
of river networks, catchments and sub-basins using Spatial
Analysis. The model utilizes a ‘daily time step’ and is
developed to simulation processes over long time periods
(Arnold et al. 1998).
The basic data inputs required for AvSWAT-X to perform
the hydrological modelling was a DEM to delineate the
watershed boundary, a stream network layer to help define
the streams, a digital soil layer, a land use layer to define the
Hydrologic Response Units (HRUs), and empirical rainfall
and weather data to predict precipitation. Weather data,
weather stations location and daily rainfall, obtained from
the Meteorological Office in Jamaica were imported to the
model. The format and assembly of these data are described
in the SWAT user manual (Neitsch et al. 2002).
For the project area, ten sub-basins were delineated
by AvSWAT-X (Figure 12). On completion, a Watershed
View was created and the following themes were added to
it: Sub-basins, Streams and Outletst. A Topographic Report
containing a detailed summary by sub-basin of distinct
locations in the watershed was also added to the current
ArcviewTM project. Each sub-basin in the Sub-basins theme
was numbered.
The names and areas of the ten sub-basins that were
created in the delineation process are outlined in Table 14.
Of the 10 sub basins, runoff analysis was carried out on
numbers 1, 2, 3, 4 and 10 because the forest optimization
process resulted in changes in these basins.
After the land use and soils theme databases were loaded,
“Reclassified” and overlayed, the HRUs were determined
FIGURE 12 Watershed View with sub-basins delineated
FIGURE 12 Watershed View with sub-basins delineated
110
after the simulation was run successfully. A display of the
simulated runoff was then presented which can be easily
modified. The before and after runoff for all the sub-basins
are also present in histogram format (Figure 14).
The percentage (rounded to whole numbers) runoff
reductions that were achieved by the optimized land use was
calculated and presented by sub-basins and showed in Table
15.
A significant number of the days in the sub-basins where
the treatment was applied have reduced runoff above 10%.
Some experienced runoff reductions of 100% on some days.
It should be noted however, that some of the values which are
0% showed an increase in runoff when the runoff values are
set to more than 4 places of decimal. Also, just as there was
a decrease in surface flow, there was also an increase in base
flow. SWAT produced a report for this but further analysis
was not done by the researcher because it was beyond the
scope of this research.
FIGURE 13 SWAT Reports
FIGURE 14 SWAT Reports
7
http://faculty.vassar.edu/lowry/wilcoxon.html (6 September 2006))
The next step was to establish whether these reductions
were statistically significant. A two sample test was therefore
conducted on the two sets of runoff data. A Wilcoxon
signed-rank test was chosen instead of a two-sample t-test
or a Mann-Whitney U Test because the runoff data from the
two sets of data did not have a normal distribution (outliers
were present).
The mean values and the p-values (Table 16) were
calculated after submitting the two runoff datasets produced
by SWAT to the Wilcoxon signed-rank test at the 95%
confidence level using the Analyse-it® statistical software
add-in for Microsoft® Excel® for Windows and the on-line
vassarStats software7.
For all of the basins, the P values for both the 1–tail and
2–tail tests are smaller than the significance level (p<0.05).
The null hypothesis that there is no significant difference
between the runoffs from the two sets of land uses was
therefore rejected.
indic
simil
O
View
to inp
T
runof
deriv
to ca
two s
Resu
After
valid
activa
2002
perio
For s
selec
meth
A
392
O.B. Evelyn
111
TABLE 15 Percentage runoff reductions by sub-basins
DATE
01/05/2002
02/05/2002
03/05/2002
04/05/2002
05/05/2002
06/05/2002
07/05/2002
08/05/2002
09/05/2002
10/05/2002
11/05/2002
12/05/2002
13/05/2002
14/05/2002
15/05/2002
16/05/2002
17/05/2002
18/05/2002
19/05/2002
20/05/2002
21/05/2002
22/05/2002
23/05/2002
24/05/2002
25/05/2002
26/05/2002
27/05/2002
28/05/2002
29/05/2002
30/05/2002
31/05/2002
1
0%
12%
77%
14%
23%
100%
0%
11%
14%
10%
15%
15%
0%
0%
6%
6%
61%
41%
0%
0%
6%
2%
1%
2%
2%
1%
1%
2%
2%
2%
3%
≤
10%
SUB BASINS
2
3
4
0%
0%
0%
0%
9%
0%
27%
27%
30%
83%
12%
0%
0%
23%
0%
0%
0%
0%
26%
24%
30%
83%
11%
0%
100%
15%
0%
18%
12%
23%
0%
36%
0%
0%
33%
0%
0%
16%
0%
0%
2%
0%
0%
7%
0%
0%
67%
0%
0%
0%
0%
33%
29%
35%
89%
50%
0%
100%
24%
0%
18%
13%
23%
0%
3%
0%
5%
5%
6%
3%
4%
2%
2%
2%
1%
1%
2%
1%
4%
10%
0%
3%
2%
2%
3%
2%
2%
3%
2%
3%
0%
3%
0%
11 50 %
10
0%
9%
100%
11%
22%
100%
0%
11%
15%
9%
50%
65%
100%
0%
7%
60%
40%
100%
0%
0%
8%
3%
4%
7%
2%
2%
9%
2%
2%
3%
3%
51 100%
TABLE 16 Results of Wilcoxon signed-ranks test
StaBASINS
tis1
2
3
tics
Me0.29
0.430
2.000
dian
w
349
256
427
ns/r
26
24
29
z
4.43
3.65
4.61
p(1<0.0001
0.000
<0.0001
tail)
p(2<0.0001
0.000
<0.0001
tail)
ns/r = number of signed ranks
4
10
0.199
0.954
106
16
2.73
351
26
4.45
0.003
<0.0001
0.006
<0.0001
Works, What Doesn’t, and Why - A Global Exchange
CONCLUSION
for Scaling Up Success Scaling Up Poverty Reduction:
Global
Learning Process
Conference
Shanghai,
TheAland
use optimization
modeland
suggested
a spatial
pattern
May
25.27,
2004. which had the following features: 1)
in the
upper
watershed
SHAIKH,
S. L., SUCHÁNEK,
P., SUN,areas
L. and
retain as conservation
forests. forested
on CORNELIS
steep slopes
KOOTEN,
2003. Does
Inclusion
Landowners’
andVAN
shallow
soils. 2)G.convert
to forest.
areasofbeing
used for
Non-Market
Values
Lower
Costs ofsoils
Creating
Carbon
agriculture
on steep
slopes
and shallow
and areas
with
Sinks.
highForest
to severe
erosion problems. 3) preserve areas with high
SHRESTHA,
M.N. and
2003.
Spatially
Distributed
agricultural production
high ground
water regeneration
Hydrological
Land-use
potential.
FeaturesModelling
1 and 2. considering
if implemented.
wouldchanges
help to
using
Remote
GIS.
Map
Conference
solve
some
of the Sensing
problemsand
being
faced
in Asia
the watershed.
2003,
Resources.
While Water
the problem
of determining the optimum forest
SHUKLA,
S.,desired
YADAV,
P. D.distribution
and GOEL to
R.minimize
K. 2001.runoff
Land
cover and the
spatial
Planning
and Linear Programming.
Map
wasUse
solved
by thisUsing
study.GIS
implementation
of the conservation
2001 - measures
Conference
Environmental
andIndia
reforestation
mayProceedings,
prove to be a challenge
since
Planning.
it would
require some amount of dislocation. Currently.
STAHR,
K. and GAISER,
T. 2004.
Estimation
of impactare
of
land use practices
on the upper
slopes
of the watershed
landagricultural
use changescrop
on catchment
andmostly
sediment
mainly
productionhydrology
and are done
on
load
in southern
Benin, farmers.
A thesis submitted in partial
small
holdings
by subsistent
fulfillment
of Science
inAgricultural
The results ofof the
theMaster
hydrological
modelling
point to
Science, Food Security andNatural Resource
Management.
where
forest cover is restored and the NRCA indicated that
UNDP/FAO.
1973.to Watershed
Management
and Soil
there were medium
severe degradation
problems.
Conservation Activities in Jamaica: An Evaluation
Report. FO:SF/JAMN 5 Technical Report 9. Forestry
Development and Watershed Management in the Upland
ACKNOWLEDGEMENT
Regions, Jamaica, p121.
USDA-SCS.
U.S. Dept.
of Agriculture.
Special thanks1972.
to by friends
and colleagues.
who areSoil
too
Conservation
Service
Engineering
Handbook.
numerous
to mention.
for National
their valuable
advise and
support.
4. Hydrology.
I amSection
truly grateful.
VAN REMORTEL, R.D., HAMILTON, M.E. and HICKEY,
R.J. 2001. Estimating the LS Factor For RUSLE Through
Iterative Slope Length Processing of Digital Elevation
REFERENCES
Data Within ArcInfo GRID, Cartography 30(1): 27-35.
VERBURG,
P.H.,
SCHOT,
P., DIJST,
J. and
AGARWAL. C..
GREEN.
G. M..P.GROVE.
J. M..M.
EVANS.
T.
VELDKAMP,
A. C.
2004.
Land Ause
change
modelling:
P. and SCHWEIK.
M. 2001.
Review
and Assessment
current
practice
andModels:
research
priorities,
GeoJournal
of Land-Use
Change
Dynamics
of Space.
Time.
61:
2004. Bloomington and South Burlington.
and 309–324,
Human Choice.
WANG,
Y. and
HUANG, G.H.
2004. Land
CenterH.,forSHENG,
the Study
of Institutions.
Population.
and
allocation
based
on Indiana
integrated
GIS-optimization
Environmental
Change.
University
and USDA
modeling
at a CIPEC
watershed
level, Landscape
and Urban
Forest Service.
Collaborative
Report Series
1.
Planning (Elsevier
Science),
66(2):
ARNOLD.
J. G.. SRINI
VASAN.
R..61-74.
MUTTIAH. R. S.
WARD,
R. C. and ROBINSON,
2000.
Principles
of
and WILLIAMS.
J. R. 1998.M.
Large
Area
Hydrologic
Hydrology.
4thAssessment:
Edition. McGraw-Hill.
Modeling and
Part I. Model Development.
WISCHMEIER,
H., JOHNSON,
C. B. and Association.
CROSS, B.
Journal of theW.American
Water Resources
V.
1971.
A Soil Erodibility Nomograph for Farmland
34(1):
73-89.
and Construction
Sites.Analysis
Journalof of
soilUse
andChange:
water
BRIASSOULIS.
H. 1999.
Land
conservation
26:
189-192.
Theoretical and Modeling Approaches. Regional
WRA.
1990. Institute.
Underground
Water Authority
Final Report,
Research
West Virginia
University.
1990.
Water
Resources
Development
Master
Plan,
Final
BURT. T. and SWANK. W. 2002.Forests or
Floods?
Report,
Main
Volume,
Government
of
Jamaica,
March
Geography Review. 15(5): 37-41.
1990.
CAMIRAND.
R. and EVELYN. O.B. 2003. Ecological land
YEO, I., GORDON, S.I. and GULDMANN, J. 2003.
Optimizing
Patterns of LandTrees
Use to
Peak Project
Runoff
Jamaica. Jamaica-Canada
forReduce
Tomorrow
Flow
and
Nonpoint
Source
Pollution
with
an
Integrated
Phase II. Forestry Department and Tecsult International.
Hydrological
and Land-Use
Model. Issn: 1087-3562,
Kingston. Jamaica.
40 p.
Journal:
Earth
Interactions
8(6):1-20.
CAMIRAND. R. 2005. Water and Soil Conservation Plan for
ZIEMER,
R.Pencar
R. 1981. Watershed
rehabilitation:
process
Buff Bay
Watershed Management
Unit.aJamaica.
Utilizing GIS for minimizing runoff in a watershed in Jamaica
view; In: Robert N. Coats (ed). Proceedings of the
Symposium on Watershed Rehabilitation in Redwood
National Park and Other Pacific Coastal Areas, 24-28
August 1981, Arcata, California. Center for Natural
Resource Studies of JMI, Inc. and National Park
Service. p. 1-10.
393