CHAPTER 1 INTRODUCTION in Recent Years, Malaysia Has

CHAPTER 1

INTRODUCTION

In recent years, Malaysia has undergone very rapid development with population growth, urbanization, expansion of agriculture, logging activities, and industrialization. These changes have caused complex environmental problems and the most affected natural resource is water.

Rising concern over the degradation of the environment, such as our river water quality, has resulted in an increase in research on the identification and study of environmental problems. This situation is changing in parallel to a rapid rise in the volumes and quantity of data collected, and massive changes in technical capability have facilitated the development of Geographical Information System (GIS) and remote sensing to handle the diversity of information involved.

Inherent in the solution of the above problem and many environmental problems is the need to bring together dispersed data sets. The complexity and size of these databases make the requirement for application of GIS and remote sensing technology all the more necessary. The advantage of holding all the information together is to allow searches and queries to be made on a combination of information so that effective formulation and implementation of any strategies, policies and plans which are highly dependent on accurate, comprehensive and timely information can be achieved.

Statement of the Problem

In the study of watershed problems, it has been found that most of the distributed/physically based HWQ model from developed countries are not suitable for local use due to different atmospheric conditions and availability of data. Another fact is that most of those models are too complex and thus, too difficult to use or not user friendly. In addition, very few of them are applicable to basin scale study. Knowing these, a simple GIS interface physically based and computationally efficient distributed model has been chosen. Through data overlaying technique, classified remote sensing data has been applied in this study. Necessary data set representing hydrology, weather, soil, elevation and surface characteristic were integrated in a GIS in tabular, vector and grid formats. By bringing key data and analytical components together “under one roof”, it is hoped that the problems of lack of integration, limited coordination, and time- intensive execution typical of the more traditional assessment tools can be overcome.

By confirming the applicability of remote sensing and GIS in watershed management, similar studies in the future will become easier in terms of time, effort and cost saving to acquire and manage temporal as well as spatial data.

With the ability of selected model to predict or generate simulated rainfall, streamflow and water quality from a river basin, these provide a quantitative means to test alternatives and controls before expensive measures are implemented. These also enable

effective planning and decision making of any development of pollution control strategies, plan or policy for better water management of a humid tropical river basin.

It is envisaged that the database generated and the water quality output of the simulation as well as maps, charts, or tables generated can be used to advance the knowledge about the existing environment during the study period. This is to allow comparison with the present or future situation of the river basin and also allow a continual trend analysis for any further similar studies.

Objectives of the Study

In view of the issues and problems discussed above, this study aim to evaluate the effectiveness of a GIS interface physically based hydrologic and water quality (HWQ) model in predicting daily stream flow and water quality from a tropical river basin.

The specific objectives:

1. To obtain land use information from classified satellite imagery;

2. To evaluate the applicability and effectiveness of selected model in a river basin

using remote sensing data and GIS-derived inputs;

3. To simulate effect of land use to surface flow and water quality at the

downstream outlet;

4. To illustrate the importance of remote sensing data and GIS for watershed study.

CHAPTER II

LITERATURE REVIEW

Catchment Background

Topography

The Upper Bernam river basin has diverse relief. About 80% of the basin are steep mountainous country rising to a height of 1830 m in the northern and eastern direction.

The remaining area at the central and downstream part is hilly land with swamps. The topographic maps (1:50,000) of the catchment and adjoining areas were used in obtaining physiographic data.

General Geology and Soil

Generally, 8 types of soil have been identified in the study area. At the upper main range which consist of steep land, perennial granite predominantly coarse-grained mega crystic granite is found. In the upper reach area, the soils consist mainly of Rengam and

Munchong-Seremban series. These sedimentation soils content generally fine to coarse quartz sand set in a clay matrix. The lower areas consist of Serdang series which is more sandy and thicker. Local Alluvium-Colluvium and soils derived from riverine alluvium such as Akob and Merbau Patah series also can be found along the river’s bed. Most of

the soils mentioned above are well drained. Textural classes mostly lie between loam to clay with moderate to average soil moisture holding capacity.

Land Uses/Land Covers

The dominant vegetation in the river basin consists of tropical hill rainforest, or in

Syminton’s (1943) classification, Hill and Upper dipterocarp forest and lowland plantation, oil palm and rubber. Other land covers that can be found are grasslands, orchards, shrubs and a few small or medium size urbanized/built up areas especially along river beds or roadsides. Good aerial photographs and a vegetation/land use map for the entire catchment can be obtained from Agricultural Department of Malaysia.

Landsat Thematic Mapper (TM) data (Path 127, Row 57) of 1989, 1993, 1995 and 1998 covering the Bernam area were obtained for watershed characterization and parameter estimation.

Climate

Upper Bernam river basin is characterized by high temperatures and humidity with relatively small seasonal variations. The mean humidity is about 77% on average, while the minimum and maximum temperatures are 26oC and 32oC respectively. Rainfall records for period between 1992 and 1999 show that rainfall peaks during the Northeast

Monsoon (October to January) and Southwest Monsoon (May to September) with a larger peak in October/December. February and July/August which are inter-seasonal

periods have the least rainfall. Average annual rainfall ranges from 2000 mm to 3500 mm. Mean annual evaporation ranges from 1200 to 1650 mm and mean annual runoff ranges from 800 mm to 1850 mm. The annual average daily sunshine is 6.2 hours. The wind is calm for most of the time. The annual average daily wind run is 89 km/day.

The weather data of weather stations were obtained from the Drainage and Irrigation

Department (DID) and the Malaysian Meteorological Services (MMS).

Water Quality

Water samples were collected for the quality analysis from all the main tributaries (water quality station) within the study area. The samples were analyzed at the Ampang

Research Station of DID for parameter such as suspended sediment, pH, alkalinity, BOD,

COD, Calcium, phosphate and etc.

As the main purpose of this research was to simulate amount of sediment load from the river basin, only this item will be described here. The content of sediment load increases in proportion to the increase in the river runoff. It is 120 mg/lit when the runoff is 20 cu.m/s and 220 mg/lit when 30 cu.m/s. On the other hand, the content of sediment load is remarkably affected by tide. During the low tides, it is about 1,100 mg/lit and about

5,000 mg/lit during the high tides (DID, 1999).

Geographical Information System (GIS)

GIS in the general context of the environmental applications can be defined as a computer-assisted and integrated environment for geographical data creation, storage and retrieval, management, manipulation, analysis and display (Nordin Jaafar, 1997). It is able to provide effective and efficient functions for handling large spatial database and aids data inventory, management, problem solving and decision making.

Concept of GIS

According to Holdstock (1998), one of the world’s leading GIS software vendors organizes data in such a way that they can be envisioned as digital layers or coverages of information. Each coverage is registered to the same common map base; each has a distinct type of feature such as points, polylines or polygons. The GIS stores the spatial data and attribute data. A coverage represents a single theme, such as land-use

(polygon), soils type (polygon), rivers (line), roads (line), and buildings (point).

The Benefits and advantages of using GIS

Some of the advantages of using GIS based water management model are as follow;

• GIS provides high efficiency in data input, database management, analysis

and presentation;

• GIS can be an assistant in making a decision;

• Elimination of all redundant data through the creation of all readily

assessable database;

• Accessibility and availability of data through a line sharing of data via

remote terminals.

Figure 2.1: Spatial Data (Source: David A. Holdstock, GIS/GPS Director, Institute for Transportation Research and Education (ITRE), North Carolina State University)

ArcView - 3D Analyst and Spatial Analyst

The 3D Analyst and Spatial Analyst are two external extensions under ArcView-GIS software that are specifically used to analyze and visualize surface data, not only through the interface tools they provide, but through their extensions to the Avenue scripting language, which allows users to create powerful custom application.

The 3D analyst is also a very powerful tool to create surface plot or 3 dimensional map, provided with two types of surface models, grids and TINs. Grids can be made by importing DTET files, raw ASCII files, or one of many image formats. Grids can also be created from point data using interpolators such as IDW, spline, or linking. TINs can be created by triangulating features represented by points, lines, and polygons or by converted from grid map.

With these surfaces, both 3D Analyst and Spatial Analyst can create spot heights, profiles, contours, viewsheds, steepest paths, and more. This new information, produced from surface analyst functions, can be used on its own or fed back into the GIS to be modeled with other spatial data and operators.

Remote Sensing

A brief review of research in remote sensing of water resources indicates that there are many positive results, and some techniques have been applied operationally. Currently, remote sensing data are being used operationally in precipitation estimates, soil moisture measurements for irrigation scheduling, snow water equivalent and snow cover extent assessment, seasonal and short term snowmelt runoff forecasts, and surface water inventories.

In water resources predicting and monitoring, the potential of satellite imagery is widely recognized. Through the use of successive satellite imagery, ongoing forest resources information for a particular area can be obtained. With remotely-sensed data of satellite images, one can monitor and detect changes that occur either abruptly or gradually in surface conditions over extended periods of time. Similarly, major agricultural land uses for crops such as rubber, oil palm, cocoa and rice plantation can be mapped and distinguished from urban settlement area and forest land through the use of satellite images.

In 1994, Rango has presented a detailed description of the theory and the application of the remote sensing technology in hydrology and water resources. Generally, remote sensing involves the inference of properties of the subsurface, surface and atmosphere from the measurement of reflection or emission of electromagnetic radiation made primarily from satellites and aircraft. The remote sensing technology offers many

opportunities for obtaining frequent measurements of hydrologic data over a wide range of spatial and temporal scales. Gradually remote sensing is becoming an essential component of hydrological sciences since it allows examination of interactions of different terrestrial components (Duhon and Nicks, 1990). Because hydrologic processes, for example, modify the electromagnetic signal in some part of the spectrum, important variables can be measured directly by remote sensing instruments. For example, remote measurements of visible, near-infrared, and thermal infrared wavelengths obtained from remote sensors can be used to estimate hydrologic parameters including the presence or absence of vegetation cover, structure of vegetation

(e.g.. biomass, leaf density), stress in vegetation, moisture content of soils and plant leaves, soil texture, and soil organic matter.

Due to their Photogrammetric properties, high resolution, and stereo coverage of ground features, satellite imagery have been utilized in forestry and water resources mostly for the stratification of forest types and other land uses.

Advantages of Using Remote Sensing

Very often, forested and catchment areas are located on hilly sites, influenced by rugged condition and poor accessibility, study on these areas become difficult without the aid of remote sensing. Detection using satellite imagery provides several advantages compared to aerial photographs in terms of cost, time and updated information of the natural resources.

Landsat data, whether in digital or visual form, provide a synoptic view of very large areas under uniform illumination conditions. This strongly favours the study of regional environmental research such as water quality.

The process of transferring the boundaries of thematic classes from one image to the map base can usually be accompanied by more or less direct tracing from one document to the other (Baltaxes, 1980). Thus, transference of details from imagery to maps is much easier compared to aerial photographs.

Another great advantage of using satellite imageries involves the capability to monitor the areas where vegetation cover is subjected to alteration by either natural or men made activities.

Thematic Mapper (TM) image

The Thematic mapper (TM) scanner is a multispectral scanning system which records reflected/emitted electromagnetic energy from the visible, reflective-infrared, middle- infrared, and thermal-infrared regions of the spectrum. TM has higher spatial, spectral, and radiometric resolution than Multispectral Scanner (MSS). The Thematic Mapper

(TM) of Landsat 4 and 5 consists of seven bands (wavelengths ranging from 0.45 μm to

11.6 μm). For the Landsat images that were used in this project, the spatial resolution is

30 m x 30 m. The seven bands measure vegetation, soil, and atmospheric

electromagnetic reflection, with each object having a unique curve of spetral reflectance

(Figure 2.2). The radiometric resolution is 8-bit, meaning that each pixel has a possible range of data values from 0 to 255. The seven spectral bands of the TM, along with a brief summary of the intended principal applications are listed in table 2.1.

Figure 2.2: Relative reflectance of typical ground cover types as a function of wave lengths (After Engmen and Gurney, 1991).

Table 2.1: Characteristic of Landsat Thematic Mapper (TM) data (From Thomas and Ralph, 1994).

Watershed Models

Watershed models are the mathematical models that are designed and set up to simulate hydrologic and water quality processes in a watershed. These models can be classified as descriptive and conceptual models. The former have the greatest applications and are of particular interest to practising hydrologists, water resources engineers and planners, and erosion control professionals because they are designed to account for observed phenomena through empiricism and the use of basic fundamentals such as continuity assumptions. Conceptual models, on the other hand, rely heavily on theory to interpret the phenomena rather than to represent the physical process. Examples of such model are those based on probability theory.

Watershed models can also be divided into physical models and mathematical models.

Viessman et al. (1987) described physical models as those models that include analogue technologies and principles of similitude applied to small-scale models whereas mathematical models use the mathematical statements to represent the systems.

Watershed models are calibrated by comparing results of the simulation with existing records. Once the model is adjusted to fit the known period of data, additional period of stream flow as well as water quality can be generated.

Model Selection

Several widely used models have been chosen and studied in detail. These model are (1)

CREAMS (Chemicals Runoff and Erosion from Agricultural Management System,

Knisel, 1980); (2) AGNPS (Agricultural Non Point Source, Young, et al., 1989); (3)

BASINS (Better Assessment Science Integrating Point and Non Point Sources, Lahlou,

1998); (4) SWAT (Soil and Water Assessment Tool, Neitsch, 1993); (5) SWAT-

ArcView Interface (Neitsch, Diluzio, 1999). Based on the literature and knowledge on

these models, SWAT-ArcView Interface has been selected as the modeling tool in this

research. A comparison of the advantages and limitations for some of these models are

as shown below:

Table2.2: Advantages and limitations of models reviewed

Model CREAMS BASINS SWAT-ArcView Interface

Advantages 1. Use readily available data, 1. GIS interface, 1. Use readily available data, 2. Simple, 2. Basin scale, 2. GIS interface, 3. Daily time step. 3. Account for differences in 3. Basin scale, soils, land use, crops, 4. Daily time step, topography, etc. 5. Account for differences in soils, land use, crops, topography, etc.

Limitations 1. Field scale, 1. Hourly time step, 2. Account for relatively 2. Extensive data homogenous soils, requirements. 3. Account for single management practices, 4. Account for single crop.

Overview of SWAT

SWAT is a GIS interface physically based model, which operates in daily time step.

Input information for each watershed is grouped or organized into the following categories: climate; unique areas of land cover; soil; management factors, ponds/reservoirs; groundwater; and the main channel, or reach draining the watershed.

SWAT allows a number of different physical processes to be simulated in a watershed.

For modeling purposes, a watershed/subbasin may be partitioned into a number of hydrologic response units (HRU). The use of HRU in a simulation is particularly beneficial when different areas of the watershed are dominated by land uses or soils different enough in properties to impact hydrology. By partitioning the watershed into

HRU, different areas of the watershed is able to relate one another spatially.

Model Theory

Simulation of the hydrology of a watershed can be separated into two parts. The first part is the land phase of the hydrologic cycle, depicted in Figure 2.3. The land phase of the hydrologic cycle controls the amount of water, sediment, nutrient, and pesticide loading to the main channel in each watershed.

The second division is the water or routing phase of the hydrologic cycle that can be defined as the movement of water, sediments, etc. through the channel network of the watershed to the outlet.

Figure 2.3: Land phase of the hydrologic cycle (Source: SWAT User’s Manual, Version 99.2)

Land Phase of the Hydrologic Cycle

The hydrologic cycle as simulated by SWAT is based on the water balance equation:

t SWt = SW + ∑ (Ri – Qi – ETi – Pi – QRi ) i=1

where

SWt is the final soil water content (mm),

SW is the soil water content available for plant uptake, defined as the initial soil

water content minus the permanent wilting point water content (mm), t is the time (days),

Ri is the amount of precipitation (mm),

Qi is the amount of surface runoff (mm),

ETi is the amount of evapotranspiration (mm),

Pi is the amount of percolation (mm), and

QRi is the amount of return flow (mm).

The subdivision of the watershed enables the model to reflect differences in evapotranspiration for various crops and soils. Runoff is predicted separately for each watershed/subbasin and routed to obtain the total runoff for a river basin.

Climate

The climatic variables required by SWAT consist of daily precipitation, maximum/minimum air temperature, solar radiation, wind speed and relative humidity.

The model allows values for daily precipitation and maximum/minimum air temperatures to be input from records of observed data, whereas solar radiation, wind speed and relative humidity are always generated by the model.

GENERATED SOLAR RADIATION. Solar radiation are generated from a normal distribution. A continuity equation is incorporated into the generator to account for temperature and radiation variations caused by dry versus rainy conditions. Solar radiation is adjusted downward when simulating rainy conditions and adjusted upward when simulating dry conditions. The adjustment is made so that the long term generated values for average monthly solar radiation agree with the input averages.

GENERATED WIND SPEED. A model developed by Richardson and Wright (1984) is used to generate daily mean wind speed and direction given the mean monthly wind speed. This model is based on a modified exponential equation.

GENERATED RELATIVE HUMIDITY. The relative humidity model uses a triangular distribution to simulate the daily average relative humidity from the monthly average.

As with solar radiation, the mean daily relative humidity is adjusted to account for wet- and dry-day effects.

Hydrology

As precipitation descends, it may be intercepted and held in the vegetation canopy or fall to soil surface. Water on the soil surface will infiltrate into the soil profile or flow overland as runoff. Runoff moves relatively quickly toward a stream channel and contributes to short-term stream response. Infiltrated water may be held in the soil and later evapotranspired or it may slowly make its way to the surface-water system by underground paths.

CANOPY STORAGE. Canopy storage is the water intercepted by vegetative surfaces

(the canopy) where it is held and made available for evaporation. When using the curve number method to compute surface runoff, canopy storage is taken into account in the surface runoff calculation.

INFILTRATION. Because SWAT operates on a daily time step, it is unable to directly model infiltration. Thus, the amount of water entering the soil profile is calculated as the difference between the amount of rainfall and amount of surface runoff.

PERCOLATION. The percolation component of SWAT uses a storage routing technique to predict flow through each soil layer in the root zone. Downward flow occurs when field capacity of a soil layer is exceeded if the layer below is not saturated.

GROUNDWATER FLOW. Groundwater flow contribution to total stream flow is simulated by creating a shallow aquifer storage (Arnold et al, 1993). Percolate from the

bottom of the root zone is recharge to the shallow aquifer. A recession constant, derived from daily stream flow records, is used to lag flow from the aquifer to the stream.

EVAPOTRANSPIRATION. Evapotranspiration is a collective term for all processes by which water in the liquid phase at or near the earth’s surface becomes atmospheric vapor.

The model computes evaporation from soil and plant separately. Potential soil water evaporation is estimated as a function of potential evapotranspiration and leaf area index

(LAI). Actual soil evaporation is estimated by using exponential functions of soil depth and water content. Plant transpiration is simulated as a linear function of potential evapotranspiration and LAI. The model offers three options for estimating potential evapotranspiration: Penman-Monteith (1965), Priestley-Taylor (1972), and Hargreaves

(1985).

LATERAL SUBSURFACE FLOW. It is one of stream flow contributions which originates below the surface but above the zone where rocks are saturated with water. A kinematic storage model is used to predict lateral flow in each soil layer.

SURFACE RUNOFF. It is flow that occurs along a sloping surface. Using daily rainfall amounts, SWAT simulates surface runoff volumes using a modification of the

SCS curve number method (USDA Soil Conservation Service, 1972). The curve number varies non-linearly with the moisture content of the soil. The curve number drops as the soil approaches the wilting point and increases to near 100 as the soil approaches saturation.

(R – 0.2s)2 Q = ------, R > 0.2s R + 0.8 s

Q = 0.0, R ≤ 0.2s

Where

Q is the daily runoff

R is the daily rainfall

s is a retention parameter,

Peak runoff rate Predictions are based on a modification of the rational method. In the modified Rational Formula, the peak runoff rate is a function of the proportion of daily precipitation that fall in the watershed, the daily surface runoff volume, and the watershed time of concentration. The proportion of rainfall occurring in the watershed is estimated as a function of total daily rainfall using a stochastic technique. The watershed time of concentration is estimated using Manning’s Formula considering both overland and channel flow.

Erosion

Erosion and sediment yield are estimated for each watershed with the Modified

Universal Soil Loss Equation (MUSLE, William, 1975). While the USLE uses rainfall as an indicator for erosive energy, MUSLE uses the amount of runoff to simulate erosion and sediment yield.

0.56 Y = 11.8 (V qp) (K) (C) (PE) (LS)

Where

Y is the sediment yield from watershed (ton.ha-1),

V is the surface runoff volume (m3),

3 -1 qp is the peak flow rate for the watershed (m s ),

K is the soil erodibility factor,

C is the crop management factor,

PE is the erosion control practice factor,

LS is the slope length and steepness factor.

The hydrologic model supplies estimates of runoff volume and peak runoff rate. The crop management factor is evaluated as a function of above ground-biomass, crop residue on the surface, and the minimum C factor for the crop. Other factors of the erosion equation are evaluated as described by Wischmeier and Smith (1978).

Routing Phase of the Hydrologic Cycle

Once SWAT determines the loading of water, sediments, nutrients and pesticides to the main channel, the loading are routed through the stream network of the watershed using a command structure.

Flood Routing

As water flow down stream, a portion may be lost due to evaporation and transmission through the bed of the channel. Another potential loss is removal of water from the channel for agricultural or human use. Flow may be supplemented by the fall of rain directly on the channel and/or addition of water from point source discharges.

Flow is routed through the channel using a variable storage coefficient method (William,

1969).

SI = SI-1 + II – OI – TL – EV + dv + rt

SC = 48/(2 TT + 24)

OI = SC (II + SI-1)

Where

3 SI is the storage in the reach (m ),

3 SI-1 is the storage in the reach from the previous day (m ),

3 II is the inflow (m ),

3 OI is the outflow (m ),

TL Is the transmission loss (m3),

EV is the evaporation (m3), dv is the diversion (m3), rt is the return flow (m3)

SC is the storage coefficient,

TT is the travel time (h)

Sediment Routing

The transport of sediment in the channel is controlled by simultaneous operation of two processes, deposition and degradation. Deposition in the channel is based on sediment particle fall velocity calculated with Stroke’s law. Stream power is used to predict degradation in the routing reaches. This equation has been modified to make it no longer a function of water surface slope. Available stream power is used to reentrain loose and deposited material until all of the material is removed. Excess stream power causes bed degradation.

SEDOUT = SEDIN – DEP + DEG

DEG = (DEGR + DEGB) (1 – DR)

Where

SEDOUT is the sediment reaching the watershed/subwatershed outlet,

SEDIN is the sediment entering the reach,

DEGR is the sediment reentrain,

DEGB is the degradation of the bed material,

DR is the delivery ratio.

SWAT- ArcView Interface

The SWAT ArcView interface was created as an extension rather than a stand-alone program. This allows the interface to take advantage of special features available in other ArcView extension packages.

Once a SWAT project has been opened, the SWAT project window is displayed. The

SWAT Watershed View menu bar contains the same pull-down menus as the basic

ArcView window plus five menus unique to the SWAT ArcView interface. These are

Watershed, Input, Edit Input, Simulation, and Reports. Figure 2.4 shows the

ArcVIew View menu and tool bars while Figure 2.5 shows the SWAT Watershed View menu and tool bars.

Figure 2.4: ArcView View menu and tool bars.

Figure 2.5: SWAT Watershed View menu and tool bars.

The custom menus in the SWAT Watershed View contain commands needed to process the watershed data, generate input files for SWAT, run the model, and display model results. Detail discussion of the functions for these menus is found in the “ArcView

Interface for SWAT 99.2” User’s Guide.

CHAPTER III

METHODOLOGY

This chapter covers the methods used and techniques applied in the research. As to achieve the objective of this research systematically, the whole process was divided into

6 sections: (1) Description of study area, (2) Data acquisition, (3) Data processing and management, (4) SWAT project development, (5) Model calibration and application, and (6) Results analyses.

Description of Study Area

This study focuses on Upper Bernam Basin (UBB) that has been identified as the ultimate and largest source of water supply for the Bernam Valley especially for irrigation. It is located in South Perak and North Selangor, Malaysia with SKC Bridge river monitoring station as the downstream outlet. The SKC Bridge monitoring station is 17 Km upstream of Bernam River Headwork (BRH); 85 Km upstream of the Bagan

Terap Pumphouse and 147 Km upstream of the estuary of the Bernam River. A brief description of the river basin is summarized in Table 3.1. Figure 3.1 shows the location of the study area.

Table 3.1: Background of the river basin

Name of study area Upper Bernam Basin (UBB)

Catchment area 1090 km2 upstream of SKC Bridge monitoring station

Longitude/latitude 3036’ to 4008’N and 101019’ to 101040’E

Main tributaries Sg. Behrang, Sg. Inki, Sg. Trolak, and Sg. Slim

Discharge and river 3813411 – Sg Bernam at SKC bridge stage stations 3615412 – Sg Bernam at Tanjung Malim

Suspended sediment 3813511 – Sg Bernam at SKC Bridge Stations 3615512 – Sg Bernam at Tanjung Malim

Climatological 3615003 – Pekan Tanjung Malim stations 3714152 – Ladang Ketoyang, Tanjung Malim 3813158 – Ladang Trolak, Trolak 3057”N 101022”E – Felda Trolak Utara 3046”N 101031”E – Haiwan Behrang Ulu

4.6

4.4

4.2

3.8

Lat 3.6

3.4

3.2

2.8

100.6 100.8 101 101.2 101.4 101.6 101.8 102

Long

Data Acquisition

Data required for this project were the climatic, water quality, and soil data of the specific river basin. Topographic maps, land use maps and satellite imageries were also required. In general, the required data and their relational sources are as listed in Table

3.2 below:

Table 3.2: Sources of data required

Sources Types of data Department of Irrigation and Hydrological / Water quality data

Drainage (DID) -Daily rainfall -30 minit rainfall intensity -Daily mean flow/discharge -Daily suspended sediment (SS) Malaysian Meteorological Services Meteorological data

-Daily rainfall -Daily maximum and minimum temperature -Daily Sunshine Hour -Daily dew point temperature -Daily 0700 and 1300 ST relative humidity -Daily potential evapotranspiration -Daily potential surface evaporation -Daily cloud cover -Daily wind speed MACRES Satellite image

-Landsat TM images (Path 127, Row 57) Year: 89, 93, 95, 98 Survey and Mapping Department of Topographic maps (scale 1: 50 000) Malaysia

Agricultural Department, Malaysia Soil maps and soil properties & Landuse map Department of Soil Science, UPM

Data Processing and Management

The data collected were transformed into computer-readable form before writing the data to the GIS database which was ArcView version 3.1. These processes include format conversion, editing, merging, registration, interpolation, calculation, and image interpretation.

TOPOGRAPHIC AND SOIL MAPS. These maps were used to delineate the project location and used to produce all needed digital thematic map layers. These include a layer of boundary area, rivers/reach, soil types, urbanized areas, spot heights, contours, etc. A three-dimensional map-Triangulated irregular network (TIN) and a Digital Elevation Model (DEM) with a resolution of 30m x 30m also were produced base on the above topographic maps using the ArcView-3D Analyst extension.

LANDSAT THEMATIC MAPPER (TM) IMAGERIES. For the landsat TM imageries that were used, enhancement was needed to remove atmospheric effects such as cloud, haze and noise that appear in each image. The enhanced imageries were then used to derive the land use maps of the Upper Bernam Basin area using a combination of different classification strategies available in ERDAS Imagine 8.3.1. By comparing the resulted classified imageries, information on land use for the entire period of study was obtained. The schematic diagram in Figure 3.2 shows the sequence involves in image processing.

RAW DIGITAL MAP

Subset FEATURE EXTRACTION

1 - To reduce haze and noise 1 2 - To eliminate Cloud Radiometric Enhancement ENHANCEMENT Mask 2

LAND USE IDENTIFICATION

CLASSIFICATION

Band Combination

SIGNATURE TN SELECTION

Feature Space

SUPERVISED UNSUPERVISED CLASSIFICATION CLASSIFICATION

CLASSIFICATION OVERLAY

RECODE CALSSES MERGE CLASSES

NEIGHBOURHOOD + FILTER

Ground truth

ACCURACY ASSESSMENT (post processing operation)

LAND USE MAP

Figure 3.2: Sequence of Landsat TM image processing.

OBSERVED NUMERICAL DATA. These include the hydrological and meteorological data. As most of these daily time step data were not in correct or readable format required by the interface, they were the main data that need to be edited and sorted. Precise statistical calculations such as mean, probability, coefficient of variation, standard deviation and skew coefficient were also performed for the data from year 1992 to 1999 using the Microsoft Excel 2000. The processed data were then transformed into organized GIS database tables which were then be used by SWAT for modeling purposes or used to plot relevant graphs for analysis purposes. Another Watershed Data Management (WDM) software called WDMUtil was also used to import the processed data for permanent storage. Although this was not necessary in this research, it was thought very helpful for data management, retrieving and also for data presentation in the form of charts, graphs etc.

SWAT INPUT PARAMETER. This includes those data that need to be supplied to the SWAT database files in order to run the model. There are seven databases in SWAT available for editing as shown below. These databases can be accessed through ‘Edit SWAT Databases’ in SWAT-ArcView main interface window. Most of the necessary data have been added in their related databases. These data were either obtained from the relevant government department, reliable literatures or by calculations. Some reasonable estimation also have been made for data that is no place be found locally. This is further described in next chapter.

Figure 3.3: Data management using WDMUtil

Figure 3.4: SWAT databases

SWAT Project Development

In order to develop a SWAT project, the interface needs to access ArcView map themes and database files that provide certain types of information about the river basin. The necessary maps and database files need to be prepared prior to run the interface.

Reference was made to “Preparing ArcView Input file” in section 3, “ArcView Interface for SWAT 99.2” User’s Manual (Neitsch, Diluzio, 1999). To do these, the customized

Watershed menu in the SWAT –ArcView interface window was used. This menu contains all the commands required to import and process the ArcView grid themes and tables used for the project. Figure 3.5 below displays the Watershed menu.

Figure 3.5: Customized Watershed menu in ArcView interface used for SWAT

project development

By executing all the commands in the Watershed menu one by one following the sequence, All the necessary grid themes were loaded. The interface was then automatically overlaid those maps that were created with the same projection and resolution against each other, divided the river basin into subbasins based on topographic relief, further subdivided each subbasin into a number of HRUs based on land use and soil type, and finally extracted model inputs data from map layers and related databases for each HRUs to form a SWAT project.

Model Calibration and Simulation

The goal of model calibration was to select a set of parameter values that predict flows, storage and other hydrologic components of the river basin so as to achieve the best possible match of predicted outputs with the recorded output (Amjad Nabi, 1998). In this study, The calibration was carried out using data set of year 1998 due to completeness and better consistency. Actual recorded daily rainfall, maximum and minimum air temperatures, streamflow and sediment load were used for prediction and parameter optimization. This was achieved by running the model, comparing the predicted and observed stream flow hydrographs, their objective functions and error statistics, and then adjusting especially those sensitive parameters such as recession constant, Threshold depth for return to occur, and initial groundwater storage etc. The optimized values of parameters obtained after calibration are given in Appendic B.

The same calibrated model with parameters set was applied to the river basin for all the three SWAT projects that were developed except the groundwater storage was adjusted to obtain a perfect match of initial streamflow rate.

Due to rapid land use changes, each SWAT run was performed for only two months for daily results. The month before the image taken was considered the ‘warming up’ or equilibrium period for the model to calculate initial soil water content for each HRU. In another word, only the results for the next month starting from the date the image were taken into consideration for each SWAT run.

For the data set that represents year 1993, due to data incompleteness during the second month of model application, the model generated predicted rainfall was used in place of those missing observed rainfall data for model application.

Result Analyses

The purpose of results analyses is to provide the user with the information whether the model generated results are reliable and can be accepted as part of the support to decision making. One of the most frequently used method is to perform statistical calculation including the Theil’s technique (Thamir et al., 1982) and correlation analysis

(McCuen, 1998).

Theil’s Technique

This Technique was used to evaluate the performance of selected model by calculating the Theil’s Coefficient that showed the total deviation of the generated results at the end of model simulation.

The values of Theil’s coefficient will fall between zero and one ( 0 ≤ U ≤ 1 ). These values of theil’s coefficient are an indication on whether the modeled results are acceptable or not acceptable. The modeled results are considered perfect match with actual recorded events if the coefficient calculated is zero; and the modeled results are considered acceptable if it is near to zero ( e.g.: U < 0.5 ). On the other hand, the modeled results are not reliable if the coefficient approaches one ( e.g.: U > 0.5 ); and if the results produces a coefficient of one, it should be discarded.

By using the Theil’s technique, the main advantage is its ability to provide a clear picture on whether or not the results should be accepted.

Correlation Analysis

Correlation analysis was carried out to determine the relationship between the modeled results and the observed results (daily streamflow and sediment load). This involved the calculation of the correlation coefficient, ρ, which is a measure of goodness of fits. The

advantage of correlation coefficient is the ability to show the relationship of two variables directly.

The values of correlation coefficients lie in the interval of –1 and 1 ( -1 ≤ ρ ≤ 1 ).

Shahin et al. (1993) classified the correlation of variables into five categories as shown in tables 3.3.

Table 3.3: Categorization of model application results using correlation analysis

Results Category Correlation Coefficient

Good results ρ ≥ 0.80

Moderately strong 0.60 ≥ ρ > 0.80

Moderate 0.40 ≥ ρ > 0.60

Weak 0.20 ≥ ρ > 0.40

Poor results ρ < 0.20

The following schematic diagram illustrates the overall process of this study.

Remote Sensing Data ArvView 3.1

ERDAS Digitizing Imagine

Satellite Imagery Thematic Layer Database

Supervised Classification ArcView + 3D Analyst Land Use Map

Grid Maps ArcView Overlay ------SWAT-ArcView Interface + Spatial Analyst SWAT Project

Model Calibration

ELSE Model Application

Compare Observed Results Modeled results Statistical Analyses

Landuse, Runoff, Accept Modeled Sediment Load Results

Relationship

Figure 3.6: Schematic diagram showing overall procedure of the study.

CHAPTER IV

RESULTS AND DISCUSSION

GIS Outputs

As was discussed in chapter 3, GIS was used successively in this study, especially in preparing needed digital map layers and database tables. This section describes the preparation and results/outputs for some of those GIS data.

Digital Map

An ArcView digital map with more than 20 layers representing points, polylines and polygons has been produced. Most of the layers were needed to create grid theme required by SWAT model such as watershed boundary, Contour, Spot Height, Soil, land use etc. The rest gives locality information (urban, gauging stations, weather station and monitoring station). Some of the layers have been displayed in various figures above. A

3-dimentional surface call Triangulated Irregular network (TIN) also has been created for viewing purposes. It was created using a combination of point, polyline, and polygon themes.

Figure 4.1: Topographic relief of the Upper Bernam Basin

Figure 4.2: Top view of Triangulated Irregular Network (TIN) representing the project area.

Grid maps

Three grid maps representing a DEM, land use and soil types were produced prior to development of a SWAT project file. Since the selected model is a physically based model, these grid themes were extremely important in this study for data overlaying, so that various types of information in each pixel can be related spatially.

Digital Elevation Model (DEM)

DEM is actually a gridded surface with elevation information. The DEM layer that has been created was a floating-point grid theme with a resolution of 30 x 30m. This layer was used by the interface to delineate subbasins within the study area and estimates surfacial parameters such as area, slope, and length of flow path for each HRU.

Soil Grid

The soil grid layer was obtained by converted directly from digitized thematic soil layer using the same boundary and pixel size as DEM layer above. This layer had 8 classes represent 8 soil types series found in the project area as below.

Soil Classes Soil Name 1. AKB Akob series 2. BTM-DRN Assoc. Batu Anam-Durian Association 3. MPH-AKB Assoc. Merbau Patah-Akob Association 4. MUN Munchong 5. RGM Rengam 6. SDG Serdang 7. SDG-MUN Assoc. Serdang-Munchong Association 8. STP Steep Land

Figure 4.3: Digital Elevation Model (DEM) for Upper Bernam Basin.

Figure 4.4: Grid map representing soil types of UBB.

Land Use Grid

The land use grid layers that were derived from classified Landsat TM Images have 7 broad classes of land cover that is: Bare Land, Forest, Oil Palm, Rubber, Scrub,

Urban, and Water Body. A sample of these land use grid layers is given in Figure 4.5.

ArcView Database Tables

In addition to the above ArcView digitized thematic and grid layers, several ArcView

database tables were also prepared to provide information needed by SWAT. Those

database tables and functions are as listed in Table 4.1.

Figure 4.5: Classified land use grid map derived from Landsat TM imagery (date: 28/1/1998)

Table 4.1: ArcView database tables and their functions

DATABASE TYPES/NAMES DESCRIPTIONS / FUNCTIONS

1 Location Table Used to specify the location of :

- nutrient.dbf - water quality station - pcpubb.dbf - rainfall station - strflow.dbf - stream flow station - tmpubb.dbf - temperature station - wgnstations.dbf - weather station

2 Land Use Look Up Table Used to specify the SWAT land cover/plant code to be modeled for each category in the land use map grid. - luc.dbf

3 Soil Look Up Table Used to specify the type of soil to be modeled for each category in the soil map grid. - soilc.dbf

4 Precipitation Data Table Used to store the daily precipitation for a weather station:

- hbupcp.dbf - Station Behrang Haiwan Ulu - ftupcp.dbf - Station Felda Trolak Utara - lkypcp.dbf - Station Ladang Ketoyang - ltlpcp.dbf - Station Ladang Trolak - ptmpcp.dbf - Station Pekan Tanjung Malim

5 Temperature Data Table Used to store the daily maximum and minimum temperature for a weather station: - tmpftu.dbf - Station Felda Trolak Utara - tmphbu.dbf - Station Haiwan Behrang Ulu

6 Discharge & Water Quality Used to store daily mean discharge and daily sediment for a Data Table** station: - Sg. Bernam at Tanjung Malim - Sg. Slim at Slim River

** Not required by SWAT model. Used to compare with simulated data.

Figure 4.6: Samples of location tables and look up tables

Figure 4.7: Samples of data tables

Raster Outputs

The Landsat TM images (path 127 and row 57) for years 1989, 1993, 1995 and 1998 have been obtained from Malaysian Centre for Remote Sensing (MACRES) and processed using Earth Resources Data Analysis System (ERDAS) Version 8.3.1 to get the land use grid maps.

Feature Extraction

Since all the obtained images are either full scene or quarter scene imagery which cover a very big area, therefore, the study area was extracted using the subset function to avoid unnecessary surrounding area being processed. This could greatly save time and hard disk space for the whole process.

Enhancement

Radiometric enhancement has been performed to make the images more interpretable.

In this case, the Haze Reduction function was used to reduce atmospheric effect, the

Noise Reduction function was used to remove spikes caused by errant detectors and finally, the mask function available in ERDAS Imagine was used to remove cloud that appear in all the obtained images.

Figure 4.8: Landsat TM Image before enhancement

Figure 4.9: Landsat TM image after enhancement.

Land Use Identification

Land use identification was done by comparing the Landsat TM images with the hard copy land use map (1992, 1995 and 1997) that cover the study area. Comparison was also made among those images itself with special attention to area with colour changes.

By doing so, major land cover types available in the study area have been identified as shown in Figure 4.10.

Land Use Classification

The enhanced images were classified using both supervised and unsupervised classification. Due to the accuracy judged by aerial photographs, topographic maps and land use maps, the result of supervised classification was used.

Supervised Classification

The supervised classification was carried out started with signature selection. The signatures or training sites in this study were selected based on pixel digital number (DN) value, look up table (LUT) values and using different combination of layers. After many trials, the range of pixel DN and LUT values for each desired class were well observed.

It was also found that Bands 4, 5 and 3 were good for viewing purpose as well as for general crop classification, and bands 1 and 2 were the best for urban area and bare land

differentiation. Feature space image based on layers 2 and 5 of landsat TM images is

good for selecting signatures of water bodies such as rivers and ponds.

1 7

6 5 3

1 7

Where,

1 - Bare land 2 - Forest 3 - Oil Palm 4 - Rubber 5 - Scrub / Shrub 6 - Urban area /Housing area 7 - Water body ( river and pond )

Figure 4.10: Land use identification based on Landsat TM image bands 4,5 and 3.

For each of the Landsat TM image, around 30 signature were selected at the beginning stage, each signature represent a certain colour and pixel value as shown in Figure 4.11.

Figure 4.11: Signature selection.

Recode Classes

After a supervised classification was performed, each signature selected above could be assigned as an individual class. Therefore, a technique call Recode Classes was used to merge the thematic raster layer to assign a new class number to all classes so that the resulted raster layer would only have 7 major classes as shown in Figure 4.12.

Column “Class Name” represents the 4 digits land cover code used by SWAT model and column “Histogram” represents the total number of pixels being assigned to a certain class.

Where OILP – Oil Palm RUBB – Rubber FRSE – Evergreen Forest SCRB – Scrub / Shrub WATR – Water Body URBN – Urban / Housing area BARE – Bare Land

Figure 4.12: Recode classes

Classification Overlay

This technique was used to compare the original image with the individual classes of the classified thematic layer. If a certain class was found not classified correctly, then the supervised classification was repeated again by reselect signatures for that particular class until all the classes were satisfactorily classified.

Neighbourhood and Filter

The newly classified image would generally have a lot of small clumps (< 10 pixels). These small clumps were not meaningful for basin scale study and they would disturb the appearance of classified image if not removed. In this case, the

Neighbourhood and the Filter functions were used to eliminate them so that the final classified image was more interpretable.

Figure 4.13a: Classified land use map derived from Landsat image (Date: 26 February 1993)

Figure 4.13b: Classified land use map derived from Land sat image (Date: 14 October 1995)

Figure 4.13c: Classified land use map derived from Land sat image (Date: 28 January 1998)

Accuracy Assessment

This is used to compare certain pixels in the classified maps with previously tested maps, aerial photographs, or other data. In this study, the result was judged by topographic maps (scale 1: 50000), aerial photographs, and land use maps.

In order to assess the classification accuracy, 250 points were generated randomly throughout each image using the Add Random Point utility. A class value was then entered for each of these points, which would be the reference points used to compare to the class values of the classified image. As a result, an overall accuracy of 95% has been obtained. A sample of sampling points, reference values and accuracy assessment report are as shown in Figures 4.14 to 4.16.

Figure 4.14: Sampling points for classified raster image accuracy assessment

Figure 4.15: Reference values for generated sampling points

Figure 4.16: Classification accuracy assessment report.

SWAT Databases and Data Estimation

This section describes the result of SWAT Databases developed and data estimation being carried out. The SWAT databases that will be discussed here are (1) User Soils,

(2) User Weather Stations, and (3) Land Cover / Plant Growth.

SWAT Database – User Soils

The User Soils database contents all the necessary soil data in this project. The Soils information is mostly obtained from the Department of Agriculture of Malaysia and unpublished information obtained from Dr. Lim Jit Sai (Soil Management Division,

Department of Agriculture).

For area which is classified as Steep land in the basin (slope > 20%), no soil information is available (neither soil types, nor soil properties), it has to be estimated. According to

Dr. Lim Jit Sai, most of the Upper Main Range areas consist of granitic parent material, and the soils derived from granitic rock normally belongs to Rengam series, If we refer to geological soil maps, a big portion of sedimentation soil surrounded the Main Range area within the basin also consist of Rengam series. Therefore, it is reasonable to conclude that the soil series at those steep land areas also belongs to Rengam series.

Another soil property that needs estimation is the moist soil albedo. The ratio of amount of reflected solar radiation by a soil surface to amount incident upon it. Based on

literature of about 20 research papers especially from Malaysia (Tan and Rajaratnam,

1975; Shaharuddin, 1980; Ling and Robertson, 1978; Fritschen, 1965; McCaughey,

1987). A value of 0.20 should be taken for most of the soil types in Malaysia at field capacity (33 KPa).

Figure 4.17: SWAT database - User Soils

SWAT Database – User Weather Stations

This database contains all the statistical weather data needed by SWAT model. All the necessary weather data in this project were obtained from the Malaysian Mateorological

Services and the Drainage and Irrigation Department of Malaysia.

The maximum 0.5 hour and Maximum 6 hours rainfall ever recorded for the study area were obtained from the auto recording rainfall station at Pekan Tanjung Malim (Station

No: 3615003). A spread sheet software (Microsoft Excel 2000) was also used to calculate the following statistical data for various stations using daily data from year

1992 - 1999:

1. Average Maximum temperature in month (0C). 2. Average Minimum temperature in month (0C). 3. Coefficient of variation for average temperature in month. 4. Average daily solar radiation in month (Langleys). 5. Maximum 0.5 hour rainfall ever recorded for month (mm). 6. Probability of wet day following a dry day in month. 7. Probability of wet day following a wet day in month. 8. Average number days of precipitation in month (day). 9. Average daily precipitation in month (mm/day). 10. Standard deviation for daily precipitation in month. 11. Skew coefficient for daily precipitation in month. 12. Average dew point temperature in month (0C). 13. Average wind in month (m/s).

Figure 4.18: SWAT database - User Weather Stations

SWAT Database – Land Cover / Plant Growth

The model came together with a very complete database of land cover / plant growth data for various crops. However, data of oil palm and rubber needed in this research were not included. Therefore, these data were added directly as new records to the database. Necessary changes on other land use classes also have been made to fulfill the local conditions. Full description of the whole database required many parameters.

Only important parameters on how their values were obtained and/or assumed are discussed here.

For oil palm trees, most of the estimated parameter values were based on 8 to 9 years old trees. The optimal temperature for Oil Palm growth is 28.1 0C (Henry, 1958; Hong and

Corley, 1976), base temperature where growth completely prevented is 15 0C (Henry,

1958). Maximum leaf area index is 4.0 (Monteny et al., 1985; Corley et al., 1971).

Maximum stomatal conductance is 0.0077 m/s (Corley, 1973). Harvest index is 0.438

(Corley and Mok, 1972; Turner, 1976). Based on a research of Turner (1976), he found that water deficit in Malaysia is normally less than 100 mm/year, and a yield drop of

12% would seem to be an appropriate figure. Therefore, the lowest harvest index due to water stress was taken as 0.385.

For mature rubber trees, the estimated optimal temperature is 27 0C (Pushpadas and

Karthikakutty Amma, 1980), base temperature is 10 0C, (Pushpadas and Karthikakutty

Amma, 1980) and maximum leaf area index is 5.5, (templeton, 1968; kato, et al., 1978).

Maximum stomatal conductance is 0.0070 (Conceicao, 1985; Monteny et al., 1985;

Samsuddin and Impens, 1979). For harvest index, based on the definition in SWAT user’s manual, this is the plant seed yield divided by the above ground biomass.

Therefore, a minimum allowable value of 0.01 was assigned.

For USLE crop management factor, a C value of 0.1 was taken for both oil palm and rubber (After Wischmeier and Smith, 1978; Roose, 1977; Singh, Babu and Chandran,

1981; El- Swaify, Dangler and Armstroog, 1982; Hurni, 1987; Hashim and Wong, 1988;

Morgan and Finney, 1982)

For the other parameters in this crop cover / crop growth database, their existence were either unnecessary or can be avoided. For example, the fraction of nitrogen and phosphorus in plant and seed are not needed as they are only essential for nutrient modeling. The value threshold vapor pressure deficit (VPD) is only needed if the

Penman-Monteith method was used to estimate potential ET. In this case, it can be avoided by selecting the Hargreves or Priestley-Tayloy method.

Historical Data Analyses

Rainfall-Runoff Ratio and Basin Water Balance

Rainfall-runoff ratio and water balance calculations have been carried for the study area as shown in Figure 4.2. The annual runoff calculated was expressed in unit depth, milimeters. Average basin losses were obtained from the difference between rainfall and runoff. Figure 4.19 shows the trend of basin rainfall and runoff from year 1960 to

1999.

Table 4.2: Calculated rainfall-runoff ratio for SKC Bridge monitoring station

Year Rainfall Runoff runoff Watershed Year Rainfall Runoff runoff Watershed ratio loses ratio loses mm mm % mm mm mm % mm A B C = B/A D = A-B A B C = B/A D = A-B

1960 3000 1980 2510 1100 43.8 1410 1961 3000 2050 68.3 950 1981 2150 1150 53.5 1000 1962 2600 1850 71.2 750 1982 2850 1200 42.1 1650 1963 2850 1635 57.4 1215 1983 1970 830 42.1 1100 1964 2775 1620 58.4 1155 1984 2900 1550 53.4 1350 1965 2750 1720 62.5 1030 1985 2600 1350 51.9 1250 1966 3250 1850 56.9 1400 1986 2300 1150 50.0 1150 1967 2550 1720 67.5 830 1987 2630 1250 47.5 1380 1968 2995 1400 46.7 1595 1988 2600 1350 51.9 1250 1969 2800 1700 60.7 1100 1989 2250 1150 51.1 1100 Ave 2841 1727 61.1 1114 Ave 2476 1208 48.8 1268

1970 2850 1725 60.5 1125 1990 2200 900 40.9 1300 1971 2860 1750 61.2 1110 1991 2550 1085 42.5 1465 1972 2860 2000 69.9 860 1992 2125 750 35.3 1375 1973 3625 2490 68.7 1135 1993 2950 1559 52.8 1391 1974 2250 1490 66.2 760 1994 2865 1504 52.5 1361 1975 2490 1490 59.8 1000 1995 3407 1682 49.4 1725 1976 2485 1250 50.3 1235 1996 3006 1518 50.5 1488 1977 2100 1240 59.0 860 1997 2832 1377 48.6 1455 1978 2150 700 32.6 1450 1998 2415 889 36.8 1526 1979 2505 1110 44.3 1395 1999 2715 1409 51.9 1306 Ave 2618 1525 57.3 1093 Ave 2706.5 1267 46.1 1439

Trend of Basin Rainfall and Runoff of Bernam River

4000

3500

3000

2500

2000

1500

1000 Rainfall Rainfall / Runoff ( mm ) 500 Runoff 0 1960 1965 1970 1975 1980 1985 1990 1995 2000 Year

Figure 4.19: Average annual rainfall from representative stations and runoff for the Bernam River Basin upstream of SKC Bridge gauging station

From Table 4.2, it can be seen that there are several years with very low rainfall-runoff ratio ( < 40% ), such as 1978, 1992, and 1998. With reference to the National Oceanic and Atmospheric Administration (NOAA) of the United State of America web site, this is because of very strong El Niño effects happened on these years. El Niño is a disruption of ocean-atmosphere system in the tropical Pacific that caused a rise in sea surface temperature and drought in the west Pacific area including Malaysia, Indonesia and Australia. The long lasting drought period has caused higher infiltration rate and watershed loses when rainfall occurred, and the streamflow is low in most of the time.

Overall, the basin’s rainfall and runoff trend plotted above show a larger gap over the years. This means that the rainfall-runoff ratios for the basin are in decrease. Results

from calculations had shown that it is decreased from 61.1% and 57.3% in the sixties and seventies to 48.9% and 46.1% in the eighties and nineties. This can be explained from the view of very active land clearance activities since nineteen sixties for rubber and oil palm plantations. The ratios are in decrease because of some of those crops had become mature after many years.

Relationship Between Discharge and Sediment

Figure 4.20 shows the relationship between discharge and sediment load for the years with available daily data. From this figure, steeper curve slope means that sedimentation rate increase more rapidly resulting from higher erosion risk. Therefore, it is very clear that year 1998 has greater severity of erosion and sedimentation due to larger open areas.

Another two years that were found with high sedimentation rate are 1993 and 1996.

This might be because of heavy land development such as the Proton Mega City project and the Behrang 2020 project happened in during these years.

The other years that following these two years were generally have low sedimentation rate. A good reason for this could be the result of cultivation or revegetation of the cleared land and maturity of oil palm or rubber plantations, which provided the land cover for the open spaces. Another reason for this probably is because of the more erodible top soil has been washed away by rainfall leaving the compacted subsoils less conducive to erosion.

By rearranging the above data as monthly events, Figure 4.21 shows that the sedimentation rate for a certain month can be estimated. By comparing the slope of each curve, another interesting point is that sedimentation rate increases more rapidly for the month with less rainfall (July) than that of the month with higher rainfall (November).

This can be explained as for dryer months, the soil surface is dry, when small rainfall occurred, most of the rain water will be infiltrated or absorbed by soil and thus, no surface runoff and erosion will happen. As the rainfall increased, surface runoff will erode the loose surface soil more rapidly. Therefore, great care should be taken if any land clearing project is to be carried in those dryer months.

2800 1998 1993 2400 1996 1994 1997 2000 1999 1995 Best Fit 1998 1600 Best Fit 1993 2.4277 Best Fit 1996 y = 0.0988x y = 13.008x1.0554 R2 = 0.8441 2 y = 4.5686x1.2466 Best Fit 1994 R = 0.9961 1200 R2 = 0.9853 Best Fit 1997 y = 1.2458x1.542 Best Fit 1999 R2 = 1 800 Best Fit 1995 Sediment Loads (tonnes/day) y = 0.2284x1.9033 R2 = 0.8923

1.6153 400 y = 0.5168x y = 0.887x1.4632 R2 = 0.9417 R2 = 0.9293

0 10 20 30 40 50 60 70 80 90 Discharge (Cumecs)

Figure 4.20: Sedimentation rate at S.K.C. Bridge station plotted by year

2500 Jul Jun Apr y = 0.1997x1.9562 Oct 2 R = 0.9901 May 2000 Nov Feb Jan y = 1.3781x1.5103 Mar R2 = 0.7678 1500 Aug y = 0.2592x1.9374 Dec R2 = 0.9705 Sep Best Fit Jul

1.7137 Best Fit Jun y = 0.2932x1.9035 y = 0.6573x 1000 2 Best Fit Apr R2 = 0.9666 R = 0.9562 Best Fit Oct

Sediment Load (tonnes/day) Load Sediment 1.6115 y = 0.6902x Best Fit May y = 0.4417x1.8408 R2 = 0.8377 2 1.5352 Best Fit Nov 2.3812 R = 0.7132 y = 1.0749x y = 0.0644x 1.6999 y = 0.6146x 2 R2 = 0.6975 R = 0.7486 Best Fit Feb y = 0.1335x2.0933 R2 = 0.7554 500 Best Fit Jan R2 = 0.7132 y = 1.1913x1.5149 R2 = 0.8482 Best Fit Mar Best it Aug Best Fit Dec Best Fit Sep 0 0 20 40 60 80 100 120 Discharge (cumecs)

Figure 4.21: Sedimentation rate at SKC Bridge station plotted by month of the year.

SWAT Project

Three SWAT projects have been developed using different Lansat images taken on 26

February 1993, 14 October and 28 January 1998. This section shows some of the important outputs for the process undertaken to create those projects prior to perform

SWAT run.

Once a new SWAT project was opened, and the map projection was defined, the dialog box for Watershed, Subbasins and Stream Definition was automatically displayed.

This dialog box was used to load the DEM layer. A digitized ArcView stream shape file also was loaded to burn-in the stream network. This is necessary because of the downstream area is consist flat area with very little relief where the elevation map was not detailed enough to accurately predict the stream network. In the stream definition section, a value of 1000 was adopted as the stream threshold area, which is the drainage area in hectares required to define the beginning of a stream. This eliminated most of the unnecessary small streams. For modeling purposes, only 13 outlet were selected in this study for further processes.

Figure 4.22: Watershed, subbasins and stream definition

After the above information were properly defined, the ArcView interface were then started to calculate topographic parameters such as deriving slope, computing downstream and upstream flow length etc. These parameters were necessary for subbasin delineation. As a result, a total of 13 subbasins were finally delineated.

Tables 4.3: Subbasins delineated, coverage areas and main reach draining each subbasin

No. Main Reach Subbasin Area (Km2)

1 Sungai Slim 164.424 2 Sungai Bersih 44.209 3 Sungai Slim 126.966 4 Sungai Bil 70. 343 5 Sungai Trolak 114.117 6 Sungai Slim 79.163 7 Sungai Slim 8.602 8 Sungai Bernam 99.961 9 Sungai Bernam 1.805 10 Sungai Behrang 101.050 11 Sungai Bernam 74.698 12 Sungai Bernam 109.761 13 Sungai Inki 94.081 Total 1089.92

Figure 4.23: Subbasins delineation.

The Land Use–Soil Definition Command was then used to load the land use and soil grid maps, and their respective look up tables. The function of these look-up tables is to link the grid maps with their database prepared in standard ArcView tables. With these information, the interface allows multiple HRUs to be created for each subbasin by using user specify sensitivity of the interface to unique land use/soil combinations. For this project, an HRU was created for each land use greater than 1% over subbasin area or soil class greater than 10% over land use area. A report of HRUs distribution for the study area is as given in Appendix A.

Figure 4.24: Land use/soil combination used to create multiple HRUs for each subbasin.

Weather Stations is the last command in the Watershed menu that was executed to complete a SWAT project, where locations of rain gages station and climate stations were linked to their ArcView database table. Three point layers (Raingages, Weagages and Tempgages) showing the location of these stations were also added on the

Watershed project window by the interface.

Figure 4.25: Weather and river monitoring stations available in the project area.

Model Application

Results from the running SWAT produced several output files that are of great importance in this project. These are the HRU output files, the Reach (.rch) output files, and the Subbasin (.bsb) output file.

The HRU output files contain summary information such as coverage area, land use and soil type for each HRU in the river basin. Using these output files, land use information within the simulation periods have been calculated as given in Table 4.4.

The Main Channel Output File and the Subbasin Output File contain information of predicted result such as streamflow and sediment load at each outlet. These data were used to compare between observed and predicted data as shown in subsequent tables and figures below. To relate the stream flow and sediment load with rainfall, the graphs for daily rainfall during the simulation periods for each gauging station and the average areal rainfall calculated using the Theissen polygon method were also plotted.

For the missing data period in year 1993, as there was no comparison can be made, even though the model can generate predicted rainfall to obtain simulated streamflow and sediment load, there were leaf blank as shown in plotted graphs in Figure 4.26 and 4.27 below.

Table 4.4: Land use information derived from Landsat TM imagery

1993 1995 1998 Land Cover Area % Area % % Area % % (km2) (km2) Changes (km2) Changes

Bare 20.73 1.90 18.77 1.72 - 0.18 38.61 3.55 + 1.64

Urban 3.92 0.36 13.77 1.26 + 0.90 14.38 1.32 + 0.96

Oil palm 63.72 5.85 89.61 8.23 + 2.38 251.71 23.12 +17.26

Rubber 212.48 19.51 209.57 19.24 - 0.27 152.46 14.00 - 5.51

Scrub 182.90 16.79 157.72 14.48 - 2.31 36.66 3.37 -13.43

Forest 603.10 55.39 597.45 54.88 - 0.51 593.90 54.55 - 0.84

Water body 2.06 0.19 2.05 0.19 0.00 1.16 0.11 - 0.08

Total 1088.9 100 1088.9 100 1088.9 100

Table 4.5: Observed and Predicted results of the Upper Bernam Basin for year 1993

Streamflow ( m3/s ) Sediment Load ( tonnes/day ) Date Observed Predicted Difference Observed Predicted Difference O P (O-P)/O % O P (O-P)/O %

4-Mar 23.9 23.8 0.4 371 369 0.5 5-Mar 20.6 18.6 9.7 316 296 6.3 6-Mar 22.2 23.1 -3.9 343 333 2.9 7-Mar 21.4 19.9 6.9 328 322 1.8 8-Mar 27.1 21.5 20.6 425 344 19.1 9-Mar 28.9 24.3 15.8 455 181 16.3 20-Mar 25.9 24.6 5.0 405 365 9.9 21-Mar 26.6 22.7 14.8 416 320 23.1 22-Mar 22.3 23.6 -5.8 343 327 4.7 23-Mar 21.0 24.1 14.8 323 332 -2.8 24-Mar 20.2 19.6 3.0 310 291 6.1 25-Mar 17.5 15.5 11.3 264 233 11.7

35 Pekan Tanjung Malim 30 Ladang Ketotang Ladang Trolak 25 Felda Trolak Utara Haiwan Behrang Ulu 20 Average Areal Rainfall

15 Rainfall ( mm ) 10

0 3/4 3/8 3/12 3/16 3/20 3/24 Date

30 Observed Predicted 27 /s ) 3 24

21 Streamflow ( m Streamflow 18

15 3/4 3/8 3/12 3/16 3/20 3/24 Date

500 Observed 450 Predicted

400

350

300

250 Sediment Load ( tonnes/day )

200 3/4 3/8 3/12 3/16 3/20 3/24 Date

Figure 4.26: Comparison of rainfall, streamflow and sediment load of The Upper Bernam Basin for year 1993.

30 Observed

) 25 Predicted 6 x 10

3 20

Cumulative Flow ( m 5

0 3/4 3/8 3/12 3/16 3/20 3/24 Date

5000 Observed 4500 Predicted 4000 3500 3000 2500

( tonnes ) 2000 1500 1000 Cumulative Sediment Load 500 0 3/4 3/8 3/12 3/16 3/20 3/24 Date

Figure 4.27: Mass curve of streamflow and sediment load of The Upper Bernam Basin for year 1993.

Table 4.6: Observed and Predicted results of the Upper Bernam Basin for year 1995

Streamflow ( m3/s ) Sediment Load ( tonnes/day ) Date Observed Predicted Difference Observed Predicted Difference O P (O-P)/O % O P (O-P)/O %

14-Oct 22.9 22.8 0.4 81 75 7.4 15-Oct 28.6 24.0 16.1 114 87 23.7 16-Oct 29.7 33.2 -11.8 121 94 22.3 17-Oct 71.3 85.6 -20.1 448 520 -16.1 18-Oct 81.9 96.7 -18.1 544 774 -42.3 19-Oct 70.0 84.7 -21.0 431 600 -39.2 20-Oct 73.7 106.0 -43.8 466 707 -51.7 21-Oct 88.5 129.3 -46.1 609 920 -51.1 22-Oct 86.4 112.6 -30.3 588 774 -31.6 23-Oct 85.3 112.0 -31.3 577 693 -20.1 24-Oct 107.2 97.7 8.9 810 572 29.4 25-Oct 112.2 79.6 29.1 866 458 47.1 26-Oct 94.8 65.5 30.9 674 350 48.1 27-Oct 71.1 47.6 33.6 447 296 33.8 28-Oct 63.8 51.2 19.7 375 260 30.7 29-Oct 78.6 74.4 5.3 512 500 2.3 30-Oct 101.8 147.0 -44.4 750 1164 -55.2 31-Oct 120.7 164.0 -35.9 966 1333 -38.0 1-Nov 125.4 139.5 -11.2 1021 1070 -4.8 2-Nov 114.4 96.7 15.5 890 727 18.3 3-Nov 100.0 74.4 25.6 730 500 31.5 4-Nov 98.5 80.9 17.9 714 491 31.2 5-Nov 93.9 54.9 41.5 666 390 41.4 6-Nov 69.2 41.9 39.5 424 229 46.0 7-Nov 57.8 33.5 42.0 324 114 64.8 8-Nov 61.4 58.0 5.5 355 280 21.1 9-Nov 75.1 173.0 -130.4 479 640 -33.6 10-Nov 102.7 267.0 -160.0 761 1670 -119.4 11-Nov 117.9 183.0 -55.2 931 1167 -24.6 12-Nov 111.5 122.8 -10.1 857 880 12.6 13-Nov 105.9 85.6 19.2 794 646 18.6

80 Pekan Tanjung Malim 70 Ladang Ketoyang Ladang Trolak 60 Felda Trolak Utara Haiw an Behrang Ulu 50 A verage A real Rainfall

30 Rainfall ( mm ) Rainfall 20

0 10/14 10/18 10/22 10/26 10/30 11/3 11/7 11/11 Date

300

Observed 250 Predicted

/s ) 200 3

150

100 Streamflow ( m Streamflow

0 10/14 10/18 10/22 10/26 10/30 11/3 11/7 11/11 Date

1800

1600 Observed ) Predicted 1400

1200

1000

800

600

400 Sediment Load ( tonnes/day Sediment 200

0 10/14 10/18 10/22 10/26 10/30 11/3 11/7 11/11 Date

Figure 4.28: Comparison of rainfall, streamflow and sediment load of the Upper Bernam Basin for year 1995

20000

18000 Observed Predicted 16000 14000

12000

10000 8000 ( tonnes ) 6000

4000

Cumulative Sediment Load 2000 0 10/14 10/18 10/22 10/26 10/30 11/3 11/7 11/11 Date

300 Observed )

6 250 Predicted x 10 3 200

150

100

50 Cumulative Flow ( m

0 10/14 10/18 10/22 10/26 10/30 11/3 11/7 11/11 Date

Figure 4.29: Mass curve of streamflow and sediment load of the Upper Bernam Basin for year 1995.

Table 4.7: Observed and Predicted results of the Upper Bernam Basin for year 1998

Streamflow ( m3/s ) Sediment Load ( tonnes/day ) Date Observed Predicted Difference Observed Predicted Difference O P (O-P)/O % O P (O-P)/O %

28-Jan 18.9 18.0 4.8 124 80 35.5 29-Jan 26.2 18.6 29.0 274 99 63.9 30-Jan 22.8 16.0 29.8 196 66 66.3 31-Jan 21.7 13.5 37.8 173 46 73.5 1-Feb 19.5 12.5 35.9 134 42 68.6 2-Feb 17.6 11.5 34.7 104 41 60.7 3-Feb 16.6 11.3 31.9 91 40 55.8 4-Feb 16.0 11.0 31.2 83 39 52.9 5-Feb 19.9 13.0 34.7 141 71 49.5 6-Feb 22.8 31.8 -39.5 196 794 -305.9 7-Feb 18.2 21.0 -15.4 113 185 -63.4 8-Feb 16.0 14.0 12.5 83 73 11.8 9-Feb 15.0 12.2 18.7 71 51 28.0 10-Feb 14.1 11.5 18.4 61 47 22.8 11-Feb 13.6 11.0 19.1 56 42 24.7 12-Feb 13.6 11.0 19.1 56 40 28.3 13-Feb 14.1 11.7 17.0 61 53 13.0 14-Feb 15.8 16.0 -1.3 80 93 -15.8 15-Feb 18.8 29.1 -54.8 122 490 -300.1 16-Feb 39.1 53.1 -35.8 725 1560 -115.3 17-Feb 52.1 56.5 -8.4 1455 1824 -25.4 18-Feb 48.3 44.5 7.9 1210 1103 8.9 19-Feb 41.8 37.2 11.0 852 610 28.4 20-Feb 38.3 30.0 21.7 689 404 41.4 21-Feb 37.2 31.1 16.4 642 309 51.9 22-Feb 37.9 32.3 14.8 672 375 44.2 23-Feb 28.4 23.0 19.0 333 200 40.0 24-Feb 22.0 16.2 26.4 179 88 50.9 25-Feb 21.2 13.5 36.3 164 59 64.0 26-Feb 18.1 11.7 35.4 112 51 54.3 27-Feb 16.9 11.4 32.5 95 44 53.5

80 Pekan Tanjung Malim Ladang Ketoyang 70 Ladang Trolak Felda Trolak Utara 60 A verage A real Rainfall

30 Rainfall ( mm ) Rainfall

0 1 /2 8 2 /1 2 /5 2 /9 2 /1 3 2 /1 7 2 /2 1 2 /2 5 Date

70 Observed 60 Predicted

50 /s ) 3 40

Streamflow ( m 20

0 1 /2 8 2 /1 2 /5 2 /9 2 /1 3 2 /1 7 2 /2 1 2 /2 5 Date

2000

1800 Observed Predicted 1600

1400

1200

1000

800

600

400

Sedimant Load ( tonnnes/day ) 200

0 1/28 2/1 2/5 2/9 2/13 2/17 2/21 2/25 Date

Figure 4.30: Comparison of rainfall, streamflow and sediment load of the Upper Bernam Basin for year 1998.

70 Observed 60 Predicted ) 6 50 x 10 3 40

20 Cumulative flow ( m flow Cumulative

0 1/28 2/1 2/5 2/9 2/13 2/17 2/21 2/25 Date

10000

9000 Observed Predicted 8000

7000

6000

5000

4000 ( tonnes ) 3000

2000

Cumulative Sediment Load 1000

0 1/28 2/1 2/5 2/9 2/13 2/17 2/21 2/25 Date

Figure 4.31: Mass curve of streamflow and sediment load of the Upper Bernam Basin for year 1998.

From model outputs, it is clearly shown that the daily predicted streamflows and sediment load match closely to the observed data both in terms of magnitude and timing, except for some of the peak flows such as on 10 December 1995 with very heavy rainfall, the results was overestimated. This is a limitation typically happen in humid tropic watershed modeling. The results also show a good agreement between rainfall, streamflow and sediment load, in which all curves plotted following the same trend, this imply that there is a close relationship between them. However, there are some days with recorded rainfall, but not reflected with increase streamflows or sediment load, for example on 12 February 1998. This is because of the rainfall occurred after a few-days dry period and it happened at the upstream area far from SKC Bridge station.

Overall, The SWAT model appears to satisfactorily represent the hydrologic response of a river basin. From plotted graphs. It can be seen that the model generally underestimated for both the daily streamflow and sediment load during low flow and the opposite happen during peak flow, with an average gap of 22% and 34% respectively. these results is considered moderately strong for a basin scale study of this size.

Statistical Analyses

The statistical analyses for model application are as summarized in table 4.8. From these figures, the quality of model outputs, and the effectiveness of SWAT model in hydrologic and water quality modeling is clearly shown. This further shows that SWAT was successfully used for determining surface runoff and sediment load in this research.

100

Table 4.8: Statistical analyses of modeling results for the Upper Bernam Basin

Daily Stream Flow Daily Sediment Load Year Theil’s Correlation Theil’s Correlation (Interpretation) (Interpretation) 1993 0.06 0.68 0.07 0.81 (M. strong) (Good) 1995 0.22 0.62 0.20 0.72 (M. strong) (M. strong) 1998 0.12 0.90 0.24 0.84 (Good) (Good)

Effect of Land Use to Streamflow and Sediment Load

Using streamflow data and sediment data taken from the SKC Bridge station, and daily rainfall data calculated from 5 representative gauging stations using the Theissen polygon method. The effects of open areas to surface flow and sediment load for each year was calculated as shown in Table 4.9 and Table 4.10.

Table 4.9: Relationship between open areas and streamflow for the month the satellite imageries were taken.

Year Surface flow Rainfall SF/R Open Area SF R OA (mm/day) (mm/day) (mm / mm) (km2) 1993 1.835 8.38 0.22 20.73 1995 6.713 14.77 0.45 18.77 1998 1.900 6.28 0.30 38.61

101

Table 4.10: Relationship between open areas and sediment load for the month the satellite imageries were taken.

Sediment load Rainfall SL/R Open Area SL/(R x OA) Year SL R OA (tonnes/day) (mm/day) (tonnes/mm) (km2) Tonnes/(mm x km2) 1993 358.25 8.38 42.78 20.73 2.06 1995 591.13 14.77 40.04 18.77 2.13 1998 301.46 6.28 56.81 38.61 1.47

In Table 4.9, For year 1993 and 1998 with almost the same average daily rainfall amount within the simulation period, year 1998 have higher runoff rate due to larger open area.

This shows that runoff rate increased in relative with open areas. Year 1995 with the least open area, but the runoff rate is the highest among all because of the very high rainfall intensity during the simulation period. This however can be explained using the

“bucket model” concept, which assumed that all the subsequent rainfall would becomes surface runoff after the bucket is filled.

For the relationship between open area and sediment loading rate, it is clearly shown with reference to column 4 and 5 of table 2, in which sedimentation rate increased with an expansions of open area. The increasing rates for each year are as shown in column 6 ranges from 1.47 to 2.06 tonnes/(mm x km2), which means that for each kilimeters- square increased of open areas, the average increment of sediment loading rate ranges from 1.47 to 2.06 tonnes per millimeter of rainfall depend on some other factors such as

102

rainfall intensity, locations of the open area with respect to river, soil type, slope length, and slope steepness as stated in the (Modified) Universal Soil Loss Equation

(USLE/MUSLE).

It should be noted here that the results calculated above are only based on the three years

(1993, 1995, and 1998) with classified land use information. It is only applicable within the simulation period above. Therefore, more work is still needed before any generalization can be drawn.

103

CHAPTER V

CONCLUSIONS AND RECOMMENDATIONS

Summary

This study is a preliminary research that combined remote sensing data, GIS and an watershed model called SWAT for hydrologic and water quality modeling. The procedure developed in this research was applied and evaluated on the Upper Bernam river basin.

The data required for this research is the topographical, hydrometeorological, soil, and the land use data. All of them are integrated in a GIS in tabular, vector and grid formats.

The land used data used in this project were derived from Landsat TM images. These images were enhanced and classified using a combination of different classification strategies. The resulted land use maps compares reasonably well with the map showing broad vegetation types of the river basin.

SWAT model is a well establish GIS interface physically based model, which operate in daily time step. For modeling purposes, SWAT allows one or more HRUs to be created for each subbasin to reflect and relate areas spatially. SWAT simulates surface runoff volumes using a modification of the SCS curve number method and estimates sediment yield with the Modified Universal Soil Loss Equation.

104

Model application has been performed on a short-term basis. The actual measured data were used to compare with the modeled output to evaluate the effectiveness of SWAT model in predicting streamflow and sediment load. The modeled results generally show that the SWAT model appears to satisfactorily represent the hydrologic response of a river basin with predicted results match closely to the observed data.

The land use information derived from classified Landsat imageries were also used in conjunction with data obtained from representative rainfall gauging stations and river monitoring station at SKC bridge to simulate the effect of land use to surface runoff and sediment load. The results show that surface runoff and sediment load are changed in parallel with open area.

Conclusions

Based on the results obtained, the following conclusions can be drawn:

1. The Land use maps resulted from Landsat TM images classification compared

reasonably well with the topographic maps showing broad vegetation types of

the river basin. An average accuracy of 95% has been obtained from

classification accuracy assessment judged by topographic and land use maps.

The urban areas and significant open areas were also identified successfully in

the classified land use maps. This appears to be a quick and satisfactory way to

obtain ground cover information for a river basin of this size when sufficient

ground data are either not available or for any reasons are difficult to obtain.

105

Therefore, remote sensing imagery can be used successfully in providing high

resolution and up to date information for watershed study.

2. To use SWAT, only 3 GIS layers are needed, all the input data are prepared in

GIS environment. In addition, necessary data can be loaded and SWAT run can

be performed using customized menu in GIS project window easily. Statistical

analyses of SWAT results using The Theil’s coefficient and Correlation

coefficient ranges from 0.06 to 0.24 and 0.62 to 0.90 respectively. All of these

supported that SWAT was a user friendly, simple but effective physically based

watershed model for hydrologic and water quality modeling,

3. Results of surface runoff rate calculated with respect to land use show that in

1993 and 1998 with almost the same average daily rainfall amount, the average

runoff-rainfall ratio is 0.22 when the total open area is 20.73 km2, and it is 0.30

when the total area is 38.61 km2. The sediment loading rates calculated are

42.78, 40.04 and 56.81 tonnes/mm rainfall in 1993, 1995 and 1998 with a total

open area of 20.73, 18.77, and 38.61 km2 respectively. These suggested that a

close relationship exists between land use, surface runoff and water quality.

4. This research has shown that integration of remote sensing, GIS and hydrologic

model are useful in environmental management.

5. The importance of remote sensing and GIS for watershed study was also

successfully illustrated through land use classification, data overlaying, data

management, input preparation for model application, and output interpretation

and presentation.

106

Recommendations

Based on the experience of this research, the following actions are recommended for future studies either to improve the results or to further extend this research. The recommendations are:

1 One of the limitations of this project is that the Landsat imageries that were used

have a resolution of 30 x 30 m2. As a result, land use area smaller than this size

could not be identified. This might seriously affect the results obtained as many

of the open areas are smaller than this size. Therefore, it is suggested that better

resolution image such as Spot Panchromatic or Ikonos to be used to overcome

the problem.

2 From the results obtained suggest that more work should be carried out to

calibrate the model using more data set either in short term or long them basis.

This should be able help to get a more ideal parameters set and better results.

3 To apply the same model to other river basins for further evaluation of model’s

performance and effectiveness.

4 To extend the study for other water quality parameters such as dissolved oxygen

(DO) and ammoniacal nitrogen (AN).

5 To simulate the impact of land use changes to water quality that consider the

effects of other factors in the Universal Soil Loss Equation (USLE).

6 To develop an “add in” subprogram to the SWAT model so that the effect of any

land use changes in the watershed to the downstream outlet is directly reflected.

107

BIBLIOGRAPHY

Aber, J.D., and J.M Melillo. 1991. Terrestrial Ecosystems. Saunders College Publishing, Chicago. 429pp

Ahmad Nadzri, B.M. 1992. Classification of forest landuse and landcover using satellite landsat Thematic Mapper (TM) Data. Research Paper, UPM, Malaysia. 95pp.

Allen, R.G., Smith, M., Pereira, L.S., and Perrier, A. 1994. An update for the calculation of reference evapotranspiration. International Commission of Irrigation and Drainage, ICID Bulletin Vol. 43(2). 92pp.

Amjad Nabi. 1998. A hydrologic model for studying climate change impact on evapotraspiration and water yield in a humid tropical watershed. PhD Thesis, UPM, Malaysia.

Ben-gai, T., A. Bitan, A. Manes, P. Alpert and A. Isreali. 1998. Theor. Appl. Climatol. 61: 207-215.

Bicknell, B.R., J.C. Imhoff, J.L. Kittle, A.S. Donigian, R.C. Johanson. 1996. Hydrological Simulation Program – Fortran (HSPF) User’s Manual for release 11. Environmental Research Laboratory, Environmental Protection Agency of United States, Athens, Geprgia 30613.

Calder, I.R., I.R. Wright, and D. Murdiyaso. 1986. A study of evaporation from tropical rainforest – West Java. Journal of Hydrology 89: 13-31.

Cecillia Yeo, J.L. 1999. A water quality study of Sungai Selangor Basin using Geographical Information System (GIS). Project paper, UPM, Malaysia.

Cooper, J.P., N.M Tainton. 1968. Light and temperature requirements for the growth of tropical and temperate grasses. Herb. Abstr. Vol. 38: 167-176.

Corley, R.H.V. 1973. Midday clossure of stomatain the oil palm in Malaysia. MARDI Research Bulletin 1(2): 1-4.

Corley, R.H.V. 1976. Photosynthesis and productivity. Pages:55-75 In: Corley R.H.V., J.J. Hardon, B.J. Wood, eds., Development in crop science (1): Oil Palm Research. Elsevier Scientific Publishing Company, Amsterdam.

Corley, R.H.V., C.K. Mok. 1972. Effect of Nitrogen, Phosphorus, Potassium and magnesium on growth of the oil palm. Expl Agric. (1972), 8: 347-353. Department of Irrigation and Drainage Malaysia. 1995. Hydrological Data-Rainfall and Evaporation Records 1986 _ 1990. Ministry of Agriculture, KL, Malaysia. 231 – 245.

108

Department of Irrigation and Drainage Malaysia. 1995. Hydrological Data-Streamflow and River Suspended Sediment Records 1986 _ 1990. Ministry of Agriculture, KL, Malaysia. 67-74, 167-174.

Desch, H.E. 1941. Malayan Forest Records N0. 14: Dipterocarp Timbers of the Malay Peninsula. Caxton Press Limited, Kuala Lumpur, Malaysia. 62-65.

Dolman, A.J., J.H.C. Gash, J. Roberts, and W.J. Shuttleworth. 1991. Stomatal and surface conductance of tropical rainforest. Agricultural and Meteorology 54: 303-318

ERDAS® education services. 1997. 4th edition. ERDAS field guide – revised and expended. ERDAS®, Inc. Atlanta, Georgia. USA.

ERDAS® education services. 1997. ERDAS Imagine® V8.3.1 – Tour Guides. ERDAS®, Inc. Atlanta, Georgia. USA.

ERDAS® education services. 1997. Imagine AdvantageTM Training Reference Manual. ERDAS®, Inc. Atlanta, Georgia. USA.

ERDAS® education services. 1997. Imagine EssentialsTM Training Reference Manual. ERDAS®, Inc. Atlanta, Georgia. USA.

Food and Fertilizer Technology Center for the Asian and Pasific Region. 1995. Soil Conservation Handbook. Wenchow Street, Taipei, Taiwan, ROC. i-174.

Foxworthy, F.W. 1982. Malayan Forest Records N0. 15: Dipterocarpaceae of the Malay Peninsula. Printers Limited, Singapore. 58.

Fritschen, L.J. 1965. Net and solar radiation relations over irrigated field crops. Agricultural Meteorology 4(1967): 55-62.

GIS from ESRI. 1997. ArcView 3D Analyst – 3D surface creation, visualization, and analysis. USA

GIS from ESRI. 1997. ArcView GIS – Using ArcVIew. USA.

Grace, J., D.U.U. Okali, F.E. Fasehun. 1982. Stomatal conductance of two tropical trees the wet season in Nigeria. Journal of applied ecology 19: 659-670.

Graser, E.A., and C.H.M. Van Bavel. 1981. The effect of soil moisture upon soil albedo. Agricultural Meteorology 27(1982): 17-26.

109

Heatwole, C.D., A.B. Bottcher, K.L. Campbell. 1987. Basin scale water quality model for Coastal Plain Flatwoods. Transaction of ASAE Vol 30(4): 1023–1029.

Hong T.K., R.H.V. Corley. 1976. Leaf temperature and photosynthesis of a tropical C3 plant, Elaeis Guineensis. MARDI Research Bulletin. Vol. 4(1): 16-20.

Houl, P., B. Mayo. 1995. ArcView® GIS Exercise Book. Onward Press, 2530 Camino Entrada, Santa Fe, NM 87505-4835 USA.

Hutchinson, S., L .Daniel. 1995. Inside ArcView® GIS, 2nd Edition. Onward Press, 2530 Camino Entrada, Santa Fe, NM 87505-4835 USA.

Chong, S.F. 1997. The network of hydrological stations and data base of DID Malaysia. Hydrology Division, DID Malaysia.

Japan International Cooperation Agency (JICA). 1987. Feasibility study on the Tanjong Karang irrigation development and management project. Department of Irrigation and Drainage, Ministry of Agriculture, Government of Malaysia. 543pp.

Kato, R., Y.Tadaki, and H. Ogawa. 1978. Plant biomass and growth increment studies in Pasoh forest. Malaysian Nature Journal 30(2): 211-224.

Kirkby, M. J., R. P. C. Morgan. Soil Erosion. A Wiley-Interscience Publication. John Wilay & Son, New York. 39 – 45.

Lahlou, M., L. Shomaker, S. Choudhury, R. Elmer, A. Hu, H. Manguerra, A. Parker. Tetra Tech, Inc. Fairfax, Virginia.. 1998. Better Assessment Science Integrating Point and Nonpoint Sources (BASINS) Version 2.0 User’s Manual. United States Environment Protection Agency, Washington, DC 20460.

Lim, J.S., A.H. Basit and H.H. Yew. 1980. Detailed reconnaissance soil survey of the Kesedar region, Kelantan. Ministri of Agriculture, Malaysia. 159pp.

Ling, A.H., and G.W. Robertson. 1981. Reflection coefficients of some tropical vegetation covers. Agricultural Meteorology 27(1982): 141-144.

MacNamara, J.P., D.L. Kane, L.D. Hinzman. 1998. An analysis of streamflow hydrology in the Kuparuk River Basin, Arctic Alaska: a nested watershed approach. Journal of Hydrology 206: 39 – 57.

Malaysian Meteorological Services. 1989. Annual summary of meteorological observations 1989. Malaysia. i – xv, 49 – 52, 85 – 88, 139 – 177.

110

Matson, K.C., C. Stallings, G.D. Jennings, R.A. Mclaughlin, S.W. Coffey. 1999. Watershed scale hydrologic modelling in a GIS environment. North Carolina University Cooperative Extension Service, Raleigh, NC 27695. 1 – 11.

McCaughey, J.H. 1987. The albedo of a mature mixed forest and a clear-cut site at Petawawa, Ontario. Agricultural and Forest Meteorology 40(1987): 251-263.

Mohamed, A.N. 1992. Classification of forest landuse and landcover using satellite landsat Thematic Mapper (TM) data. Project paper, UPM, Malaysia.

Mokhtaruddin, A.M., J. Shamshuddin, H. Aminuddin, Chow. W. T. 1985. Advances in soil research in Malaysia. Proceedings of the technical workshop on soil science organized by UPM/DOA/Belgium, 8 – 10 Nov 1983. UPM, Serdang, Selangor, Malaysia. 5, 10, 15-19.

Shahin, M., H.J.L.Van Oorschot, S.J. De Lange. 1993. Statistical analysis in water resources engineering. Rotterdam: A. A. Balkema. 275-278.

Monteny, B.A., J.M. Barbier, and C.M. Bernos. 1985. Determination of the energy exchanges of a forest type culture: Hevea Brasiliensis. Pages: 211-233 In: Hutchigon, B.A., and B.B. Hicks, eds., The forest atmosphere interaction. Reidel Publ. Company, Dordrecht.

Morgan, R.P.C. 1997. Soil erosion and conservation, Second Edition. Longman. 32, 67, 73.

Mustafa K. Baharuddin and G. McGuire. 1995. Erosion risk map of Pulau Langkawi. Pages: 139-144 In: Soil Science Conference of Malaysia 1995. Langkawi, Malaysia.

Neitsch, S.L., J.G. Arnold, J.R. Williams. 1999. Soils And Water Assessment Tool User’s Manual, version 99.2. Blackland Research Center, Texas Agricultural Experiment Station, 808 East Blackland Road – Temple, Texas 76502.

Neitsch, S.L., M. Diluzio. 1999. ArcView Interface For SWAT 99.2 User’s Manual. Blackland Research Center, Texas Agricultural Experiment Station, 808 East Blackland Road – Temple, Texas 76502.

Paramananthan, S. 1978. Register of soils Peninsular Malaysia, Soils and analytical services bulletin No. 7. Division of Agriculture, Ministry of Agriculture, Malaysia.

Poiani, K.A., B.L. Bedford. 1995. GIS-based nonpoint source pollution modeling: Consideration for wetlands. Journal of soil and water conservation. 613 – 619.

111

Pushpadas, M.V., M. Karthikakutty. 1980. Agroecological requirements. Pages: 87-97. In: Bhaskaran, V. K, eds., Handbook of natural rubber product in India. The Rubber Institude of India, Rubber Board, Kottayam.

Quenzer, A.P. Melancon, N. Armstrong, and the USEPA. 1999. BASINS2: Using the NPSM Model. USA

Quenzer, A.,P. Melancon, N. Armstrong, and the USEPA. 1999. Introducrtion to Basins2. USA.

Ramachandran P., R. Narayanan, J.C.X. Knecht. 1973. A planting distance experiment on Dura palms. Pages: 72-74. In: Advances in oil palm cultivation, The proceedings of the International Oil Palm Conference. Kuala Lumpur. The Incorporated Society of Planters. Malaysia.

Rango, A. 1994. Application of remote sensing methods to hydrology and water resources. Journal of Hydrological Sciences, 39: 309 – 319.

Robert, J., O.M.R. Cabral, G. Fisch, L.C.B. Milion, C.J. Moore and W.J. Shuttleworth. 1993. Transpiration from an Amazonian rainforest calculated from stomatal conductance measurements. Agricultural and forest meteorology 65: 175-196.

Rosenthal, W.D., R. Srinivasan, J.G. Arnold. 1995. Alternative river management using a linked GIS-hydrology model. Transaction of ASAE Vol. 38(3): 783 – 790.

Roslinah Samad, Norizan Abdul Patah. Soil erosion and hydrological study of the Bakun Dam catchment area, Sarawak using remote sensing and Geographic Information System (GIS). Malaysian Centre for Remote Sensing (MACRES), KL, Malaysia.

Rowntree, P.R. 1991. Atmospheric parameterization schemes for evaporation over land: Basic Concepts and climate modelling. Pages: 5-30 In: Land surface evaporation measurement and parameterization. Springer-Verlag, New York.

Samsudin Musa. 1984. Suspended and dissolved sediment concentration of two disturbed forested watersheds in Air Hitam Forest Reserve. Project paper, UPM, Malaysia. 3 – 36.

Samsuddin, Z. and I. Impens. 1979. Photosynthesis and resistances to carbon dioxide in Hevea Brasiliensis muel. aGr. Clones. Oecologia. 37: 361-363.

Saugier, B., and N. Katerji. 1991. Some plant factor controlling evapotranspiration. Agricultural and meteorology 54: 263-277.

112

Schwab, G. O., R.K. Frevert. Soil and water conservation engineering, Third Eddition. John Wilay & Son, New York. 101-103.

Shaharuddin, B.A. 1980. The effect of soil properties on albedo. Ilmu Alam, N0.9(1980): 113-120.

Sieh, K.C., K. Sivapakianathan. Hydrological procedure no. 19: The determination of suspended sediment discharge. Bahagian Parit Dan Taliair, Kementerian Pertanian, Malaysia. 37pp.

Soo, S.W. 1970. Reconnaissance soil survey of Perak, Ministry of Agricultural and Land, Malaysia.

Tan, B.C., A. Rajaratnam. 1975. Preliminary result of reflection coefficient of some typical terrains for shortwave radiation. Malaysian Agricultural Research 4(1): 211-215.

Technical Bulletin, Faculty of Agricultural, UPM. 1983. Register of soil physical properties of Malaysia. Soil Science Department, UPM, Serdang Selangor, Malaysia.

Templeton, J.K. 1968. Growth in Hevea Brasiliensis 1: Growth analysis up to seven years after budgrafting. Journal of the Rubber Research Institute of Malaysia 20(3): 136-146.

Thamir Ahmed Mohammed, Salim Said, Mohd. Zohadie Bardaie. 1999. Smart modeling of flood occurrences in a river basin. 6pp. In: Proceedings of National Conference on Engineering Smart Farming for the next millenium, UPM 14-16 March 1999. Serdang: MSAE and UPM, Malaysia.

Tim, U.S. 1996. Emerging technologies for hydrologic and water quality modeling research. Transaction of ASAE Vol.39(2): 465 – 476.

Turner, P.D. 1976. The effects of drought on oil palm yields in South-east Asia and the South Pasific region. 21pp. In: Malaysian International Agricultural Oil Palm Conference 1976. Kuala Lumpur, Malaysia.

United States Environmental Protection Agency. 1999. WDMUtil Version 1.0 (BETA) User’s Manual- A tool for managing watershed modeling time-series data. Washington, DC 20460.

Wan Yusoff Wan Ahmad. 1998. Application of landsat/spot digital and visual analysis as a tool for forest classification and mapping in Lesong Forest Reserve, Peninsular Malaysia. Fellowship project report in remote sensing at the ASEAN Institute of Forest Management. Forestry Department Headquarters, KL, Malaysia.

113

Wong, I.F.T. 1970. Reconnaissance soil survey of Selangor. Ministry of Agricultural and Lands, Malaysia.

114

APPENDICES

115

APPENDIX A

HRU Output File for Year 1998

Tue Dec 05 06:56:29 2000 WATERSHED: Basin AREA: 1088.898 [Km²]

MULTIPLE HRUs LandUse/Soil OPTION THRESHOLDS : 2 / 20 [%] ------ELABORATED SUBBASINS COMPOSITION Number of HRUs: 90 ------HRU SUBBASIN NUMBER: 1 AREA: 164.424 [015.10%] LANDUSE: FRSE AREA: [091.15%] SOIL: STP AREA: [100.00%] 1 LANDUSE: OILP AREA: [008.85%] SOIL: STP AREA: [100.00%] 2

SUBBASIN NUMBER: 2 AREA: 44.2093 [004.06%] LANDUSE: FRSE AREA: [091.13%] SOIL: STP AREA: [100.00%] 3 LANDUSE: OILP AREA: [008.87%] SOIL: STP AREA: [077.92%] 4 SOIL: RGM AREA: [022.08%] 5

SUBBASIN NUMBER: 3 AREA: 126.966 [011.66%] LANDUSE: BARE AREA: [002.91%] SOIL: AKB AREA: [038.81%] 6 SOIL: RGM AREA: [061.19%] 7 LANDUSE: RUBB AREA: [010.59%] SOIL: SDG AREA: [026.45%] 8 SOIL: RGM AREA: [073.55%] 9 LANDUSE: SCRB AREA: [002.42%] SOIL: AKB AREA: [030.31%] 10 SOIL: RGM AREA: [069.69%] 11 LANDUSE: FRSE AREA: [060.68%] SOIL: STP AREA: [100.00%] 12 LANDUSE: OILP AREA: [023.41%] SOIL: STP AREA: [037.96%] 13 SOIL: SDG AREA: [025.99%] 14 SOIL: RGM AREA: [036.05%] 15

SUBBASIN NUMBER: 4 AREA: 70.3428 [006.46%] LANDUSE: RUBB AREA: [016.58%] SOIL: SDG AREA: [100.00%] 16 LANDUSE: FRSE AREA: [068.82%] SOIL: STP AREA: [100.00%] 17 LANDUSE: OILP AREA: [014.61%] SOIL: SDG AREA: [100.00%] 18

SUBBASIN NUMBER: 5 AREA: 114.117 [010.48%] LANDUSE: BARE AREA: [004.74%] SOIL: SDG AREA: [100.00%] 19 LANDUSE: RUBB AREA: [023.22%] SOIL: SDG AREA: [100.00%] 20 LANDUSE: URBN AREA: [002.54%] SOIL: SDG AREA: [100.00%] 21 LANDUSE: SCRB AREA: [017.94%] SOIL: SDG AREA: [100.00%] 22 LANDUSE: FRSE AREA: [022.37%] SOIL: STP AREA: [100.00%] 23 LANDUSE: OILP AREA: [029.19%]

116

SOIL: SDG AREA: [100.00%] 24

SUBBASIN NUMBER: 6 AREA: 79.1629 [007.27%] LANDUSE: BARE AREA: [003.81%] SOIL: AKB AREA: [032.28%] 25 SOIL: SDG AREA: [067.72%] 26 LANDUSE: RUBB AREA: [030.37%] SOIL: SDG AREA: [100.00%] 27 LANDUSE: URBN AREA: [002.98%] SOIL: AKB AREA: [052.96%] 28 SOIL: SDG AREA: [047.04%] 29 LANDUSE: SCRB AREA: [005.82%] SOIL: SDG AREA: [100.00%] 30 LANDUSE: FRSE AREA: [008.56%] SOIL: STP AREA: [100.00%] 31 LANDUSE: OILP AREA: [048.46%] SOIL: SDG AREA: [100.00%] 32

SUBBASIN NUMBER: 7 AREA: 8.6023 [000.79%] LANDUSE: BARE AREA: [003.33%] SOIL: SDG AREA: [100.00%] 33 LANDUSE: RUBB AREA: [041.01%] SOIL: AKB AREA: [039.16%] 34 SOIL: SDG AREA: [060.84%] 35 LANDUSE: SCRB AREA: [003.36%] SOIL: AKB AREA: [023.72%] 36 SOIL: SDG AREA: [076.28%] 37 LANDUSE: OILP AREA: [052.29%] SOIL: AKB AREA: [058.22%] 38 SOIL: SDG AREA: [041.78%] 39

SUBBASIN NUMBER: 8 AREA: 99.9608 [009.18%] LANDUSE: RUBB AREA: [026.37%] SOIL: SDG AREA: [100.00%] 40 LANDUSE: FRSE AREA: [015.30%] SOIL: STP AREA: [041.75%] 41 SOIL: SDG AREA: [026.23%] 42 SOIL: SDG-MUN Assoc. AREA: [032.02%] 43 LANDUSE: OILP AREA: [058.33%] SOIL: SDG AREA: [050.33%] 44 SOIL: SDG-MUN Assoc. AREA: [049.67%] 45

SUBBASIN NUMBER: 9 AREA: 1.8511 [000.17%] LANDUSE: RUBB AREA: [009.63%] SOIL: AKB AREA: [100.00%] 46 LANDUSE: WATR AREA: [006.56%] SOIL: AKB AREA: [043.75%] 47 SOIL: MPH-AKB Assoc. AREA: [056.25%] 48 LANDUSE: URBN AREA: [005.64%] SOIL: AKB AREA: [045.45%] 49 SOIL: SDG AREA: [054.55%] 50 LANDUSE: SCRB AREA: [003.23%] SOIL: AKB AREA: [042.86%] 51 SOIL: SDG AREA: [034.92%] 52 SOIL: SDG-MUN Assoc. AREA: [022.22%] 53 LANDUSE: OILP AREA: [074.95%] SOIL: AKB AREA: [061.13%] 54 SOIL: MPH-AKB Assoc. AREA: [038.87%] 55

SUBBASIN NUMBER: 10 AREA: 101.050 [009.28%] LANDUSE: BARE AREA: [011.44%] SOIL: SDG AREA: [100.00%] 56 LANDUSE: RUBB AREA: [009.50%] SOIL: AKB AREA: [021.11%] 57 SOIL: SDG AREA: [078.89%] 58 LANDUSE: FRSE AREA: [064.44%] SOIL: STP AREA: [078.17%] 59 SOIL: SDG AREA: [021.83%] 60

117

LANDUSE: OILP AREA: [014.62%] SOIL: SDG AREA: [100.00%] 61

SUBBASIN NUMBER: 11 AREA: 74.6984 [006.86%] LANDUSE: BARE AREA: [008.98%] SOIL: AKB AREA: [029.60%] 62 SOIL: SDG AREA: [070.40%] 63 LANDUSE: RUBB AREA: [026.70%] SOIL: SDG AREA: [071.16%] 64 SOIL: BTM-DRN Assoc. AREA: [028.84%] 65 LANDUSE: URBN AREA: [004.86%] SOIL: AKB AREA: [061.44%] 66 SOIL: SDG AREA: [038.56%] 67 LANDUSE: SCRB AREA: [004.49%] SOIL: AKB AREA: [031.22%] 68 SOIL: SDG AREA: [040.60%] 69 SOIL: BTM-DRN Assoc. AREA: [028.18%] 70 LANDUSE: FRSE AREA: [021.19%] SOIL: STP AREA: [057.41%] 71 SOIL: SDG AREA: [042.59%] 72 LANDUSE: OILP AREA: [033.78%] SOIL: AKB AREA: [035.85%] 73 SOIL: SDG AREA: [035.19%] 74 SOIL: SDG-MUN Assoc. AREA: [028.96%] 75

SUBBASIN NUMBER: 12 AREA: 109.761 [010.08%] LANDUSE: RUBB AREA: [005.75%] SOIL: AKB AREA: [032.21%] 76 SOIL: SDG AREA: [067.79%] 77 LANDUSE: FRSE AREA: [090.56%] SOIL: STP AREA: [100.00%] 78 LANDUSE: OILP AREA: [003.69%] SOIL: AKB AREA: [056.80%] 79 SOIL: SDG AREA: [043.20%] 80

SUBBASIN NUMBER: 13 AREA: 94.0808 [008.64%] LANDUSE: BARE AREA: [004.96%] SOIL: AKB AREA: [046.23%] 81 SOIL: MUN AREA: [053.77%] 82 LANDUSE: RUBB AREA: [013.59%] SOIL: AKB AREA: [024.99%] 83 SOIL: MUN AREA: [075.01%] 84 LANDUSE: URBN AREA: [005.27%] SOIL: AKB AREA: [062.37%] 85 SOIL: MUN AREA: [037.63%] 86 LANDUSE: FRSE AREA: [058.44%] SOIL: STP AREA: [100.00%] 87 LANDUSE: OILP AREA: [017.74%] SOIL: AKB AREA: [025.91%] 88 SOIL: BTM-DRN Assoc. AREA: [025.06%] 89 SOIL: MUN AREA: [049.03%] 90

118

APPENDIX B

Model Input and Calibrated Parameters Set

Daily Rainfall and Temperature from Each Recorded Station

119

Statistical Data for Selected Weather Stations – Felda Trolak Utara

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec TMPMX 35.05 35.21 35.38 35.49 35.79 35.15 35.15 34.74 34.63 34.5 35.02 34.58 TMPMN 19.21 19.38 20.22 21.44 21.51 21.27 20.89 20.63 20.80 20.96 20.81 20.05 TMPCVT 0.017 0.023 0.021 0.021 0.018 0.014 0.015 0.010 0.013 0.014 0.013 0.034 SOL 813.5 845.7 876.0 887.2 825.4 831.4 815.4 834.2 891.0 852.8 844.6 797.4 HHMX 44.3 66.9 43.2 55.2 53.4 40.4 30.3 40.0 33.0 62.1 50.1 31.2 PRB_D 0.215 0.208 0.204 0.150 0.167 0.256 0.151 0.161 0.156 0.151 0.128 0.151 PRB_W 0.269 0.333 0.382 0.594 0.419 0.228 0.194 0.296 0.344 0.559 0.661 0.495 PCPD 14 15 18 22 18 15 11 14 15 22 24 20 ADP 5.16 7.94 8.30 11.23 7.37 6.09 6.01 5.43 5.13 11.00 12.89 8.12 SDP 12.11 15.73 15.10 18.27 13.80 12.53 14.88 13.50 11.28 19.08 20.07 17.96 COEF 3.54 3.23 2.85 2.51 3.16 2.93 3.48 3.66 3.20 2.62 2.37 3.21 DEWPT 23.27 23.52 23.87 24.60 24.76 24.28 23.84 23.63 23.74 23.84 23.98 23.57 WIND_AV 1.19 1.20 1.33 1.29 1.21 1.37 1.36 1.33 1.30 1.28 1.15 1.18

Statistical Data for Selected Weather Stations – Haiwan Behrang Ulu

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec TMPMX 34.4 34.8 34.9 34.8 34.9 34.7 34.3 34.6 34.4 34.0 34.0 33.8 TMPMN 19.5 19.6 19.8 20.2 20.8 20.8 20.5 20.2 20.4 20.9 20.9 20.0 TMPCV 0.006 0.003 0.010 0.009 0.005 0.007 0.006 0.009 0.009 0.012 0.011 0.016 SOL 811.7 844.6 875.8 887.7 826.6 832.9 816.9 835.2 890.8 851.8 854.9 795.2 HHMX 44.3 66.9 43.2 55.2 53.4 40.4 30.3 40.0 33.0 62.1 50.1 31.2 PRB_D 0.183 0.202 0.172 0.172 0.161 0.222 0.156 0.188 0.189 0.140 0.139 0.129 PRB_W 0.258 0.280 0.414 0.528 0.425 0.294 0.215 0.274 0.317 0.570 0.633 0.554 PCPD 14 14 18 21 18 16 11 14 15 22 23 21 ADP 4.84 7.64 9.88 11.73 7.91 8.44 5.94 5.57 7.36 10.16 10.88 10.24 SDP 10.1 13.1 19.4 18.5 16.3 17.2 12.8 14.8 13.5 16.2 17.4 17.0 COEF 3.14 2.16 2.99 2.25 3.5 2.88 2.55 2.97 2.26 3.13 2.64 2.64 DEWPT 23.08 23.32 23.75 24.25 24.49 24.15 23.70 23.74 23.76 23.91 24.06 23.66 WIND_AV 1.19 1.20 1.33 1.29 1.21 1.37 1.36 1.33 1.30 1.29 1.15 1.18

120

Input Data for Soil Types

Soil Name HYDGRP ALB USLE-K BD AWC CLAY SILT SAND Layer Depth (cm) AKB Z1 = 10 C 0.2 0.4 1.1 0.43 27.0 53.0 20.0 Z2 = 30 1.25 0.50 46.0 47.0 7.0 Z3 = 100 1.26 0.50 50.0 40.0 10.0 MUN Z1 = 50 B 0.2 0.2 1.14 0.39 85.4 4.4 10.2 Z2 = 100 1.11 0.38 84.4 4.4 11.2 RGM Z1 = 50 B 0.2 1.33 0.41 50.0 2.8 47.2 Z2 = 100 0.2 1.42 0.40 50.9 4.3 44.8 SDG Z1 = 50 B 0.2 0.2 1.58 0.47 19.2 10.7 70.1 Z2 = 100 1.60 0.46 38.9 7.6 53.5 STP Z1 = 50 B 0.2 0.2 1.33 0.41 50.0 2.8 47.2 Z2 = 100 1.42 0.40 50.9 4.3 44.8 BTM-DRN Assoc Z1 = 50 B 0.2 0.3 1.58 0.45 46.0 34.6 19.4 Z2 = 100 1.48 0.62 36.6 53.5 9.9 MPH-AKB Assoc Z1 = 10 C 0.2 0.4 1.1 0.43 36.5 50.0 13.5 Z2 = 50 1.26 0.56 43.0 38.5 18.5 Z3 = 100 1.24 0.50 57.0 29.0 14.0 SDG-MUN Assoc Z1 = 50 B 0.2 0.2 1.36 0.43 52.2 7.6 40.2 Z2 = 100 1.36 0.42 61.6 6.0 32.4

121

Sample of Input Parameters for HRUs

Sample of Input Parameters for Main Channels / Reach

122

BIODATA OF THE AUTHOR

Lai Sai Hin was born in Kuching, Malaysia on 17 April, 1972. He received his primary education at Sekolah Rendah Chung Hua Batu 10 and continued his secondary education at S.M.K. Penrissen No.1 and later changed to Kolej Tun Abdul Razak for higher education in Kuching, the capital of Sarawak. After optaining STPM, He continued to pursue the Bachelor of Engineering (Biological and Agricultural) at Universiti Putra

Malaysia in 1995. He then further his study at the same University as a full time masters student in 1999. His research interests are in Geographical Information System (GIS),

Ground Positioning System (GPS), and remote sensing.