Erosion and Water Resources Assessment in the Upper Inabanga Watershed, Philippines: Application of Wepp and Gis Tools

EROSION AND WATER RESOURCES ASSESSMENT IN THE UPPER INABANGA WATERSHED, PHILIPPINES: APPLICATION OF WEPP AND GIS TOOLS

IMELIDA C. GENSON (BSc Ag Eng)

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE (HONOURS)

WATER RESEARCH LABORATORY SCHOOL OF NATURAL SCIENCE, UNIVERSITY OF WESTERN SYDNEY HAWKESBURY CAMPUS, RICHMOND, NEW SOUTH WALES AUSTRALIA

JULY 2006

I wish to express my deepest gratitude to individuals and groups for taking part in the success of this undertaking:

To the Australian Center for International Agricultural Research (ACIAR) through the John Allwright Fellowship for the financial support extended;

To my government through the Bureau of Soils and Water Management (BSWM) for the official leave of absence permitted;

To my supervisors, Associate Professor John Bavor, Dr. Anthony Haigh and Dr. Berthold Rembertus Hennecke for their support and guidance;

Special thanks and recognition to the efforts of Dr. Willie Joshua for his help and the fruitful discussions during the writing of this thesis;

Sincere appreciation to Bronwyn Davies for taking time to edit this report;

And to the field staff of the Watershed Project Bohol, Philippines for providing the data.

DEDICATION

This piece of work is dedicated to my family: Papa & Mama and to my eight brothers and three sisters.

Also, I wish to dedicate this work to my friend Prasan Sharp for her unselfish support and encouragement. And, to the Carty Family (Michael, Vicky and Phillip) who adopted me into their family. Thanks for providing a warmth environment of being homed away from home.

For all these blessings, to God be the glory.

STATEMENT OF AUTHENTICATION

This thesis contains no material which has been accepted for the award of any other degree or diploma in any university or institution and, to the best of the author’s knowledge and belief, contains no material previously written or published by another author except when due reference is made in the text.

______

Imelida C. Genson

Table of Contents

LIST OF TABLES ……………….. iv

LIST OF FIGURES ……………….. vi

LIST OF ACRONYMS AND ABBREVIATIONS ……………….. viii

ABSTRACT ……………….. ix

CHAPTER 1. INTRODUCTION ……………….. 1 1.1 Introduction ……………….. 1 1.2 The Philippine uplands ……………….. 1 1.3 The Bohol Watershed Project ……………….. 2 1.4 Aims ……………….. 5 CHAPTER 2. LITERATURE REVIEW 2.1 The integrated watershed approach of resources management ……………….. 6 2.2 Erosion processes ……………….. 8 2.3 Factors of erosion modeling ……………….. 9 2.3.1 Rainfall ……………….. 10 2.3.2 Soil properties ……………….. 11 2.3.3 Surface cover and cropping practices ……………….. 12 2.3.4 Topography ……………….. 13 2.4 Erosion modeling ……………….. 13 2.4.1 USLE model ……………….. 15 2.4.2 WEPP model ……………….. 16 2.4.2.1 Hillslope erosion component ……………….. 17 2.4.2.2 Hillslope surface hydrology ……………….. 19 2.4.2.3 Water balance and percolation ……………….. 19 2.4.2.4 Subsurface hydrology ……………….. 19 2.4.2.5 Soil component ……………….. 19 2.4.2.6 Plant growth component ……………….. 19 2.4.2.7 Climate component ……………….. 20 2.4.2.8 Residue decomposition ……………….. 20 2.5 Estimates of erosion at a plot and watershed scale ……………….. 20 2.6 Erosion studies in the Philippines ……………….. 21 2.6.1 Plot scale ……………….. 21 2.6.2 Watershed scale ……………….. 23 2.7 Erosion models and GIS ……………….. 24

i 2.7 GIS application in watershed management ……………….. 26

CHAPTER 3. MATERIALS AND METHODS 3.1 The study area ……………….. 28 3.2 On-site assessment of soil loss and runoff ……………….. 29 3.2.1 The runoff plots ……………….. 30 3.2.1.1 Agroforestry ……………….. 32 3.2.1.2 Cassava/corn ……………….. 32 3.2.1.3 Forest ……………….. 32 3.2.1.4 Grassland ……………….. 33 3.2.1.5 Oil palm ……………….. 33 3.2.2 Watershed measurement of flow ……………….. 33 3.2.3 Climate ……………….. 34 3.2.3.1 Rainfall erosivity ……………….. 35 3.3 Computer simulations ……………….. 35 3.3.1 Inputs for WEPP and GeoWEPP ……………….. 35 3.3.1.1 Climate file ……………….. 35 3.3.1.2 Slope file ……………….. 36 3.3.1.3 Soils input file and soil map ……………….. 37 3.3.1.4 Crop management file and land cover map ……………….. 37 3.3.2 Scenarios ……………….. 38 3.3.2.1 Hillslope application ……………….. 38 3.3.2.2 Watershed application ……………….. 40

CHAPTER 4. RESULTS AND DISCUSSION 4.1 On-site measurements ……………….. 42 4.1.1 Experimental runoff plots and rain gauges ……………….. 42 4.1.1.1 Weekly rainfall from the local rain gauges ……………….. 42 4.1.1.2 Weekly rainfall >60 mm ……………….. 44 Weekly rainfall-runoff from the experimental 4.1.1.3 runoff plots ……………….. 44 Weekly rainfall-soil loss from the experimental 4.1.1.4 runoff plots ……………….. 47 4.1.1.5 Totall rainfall, runoff and soil loss ……………….. 49 4.1.1.6 High erosion week ……………….. 51 4.1.2 Rainfall erosivity analysis ……………….. 52 4.1.2.1 Rainfall erosivity and soil loss ……………….. 54 4.1.3 Watershed measurement of flow ……………….. 56 4.1.3.1 Land cover and slope ……………….. 57 4.1.3.2 Rainfall and discharge curves ……………….. 58

ii 4.1.3.3 Suspended sediment concentration ……………….. 59 4.1.4 Summary of the on-site measurement of erosion ……………….. 61 4.2 Application of WEPP and GeoWEPP erosion models ……………….. 61 4.2.1 Erosion assessment at farm level ……………….. 62 4.2.1.1 Increasing slope ……………….. 63 4.2.1.2 Effects of terracing ……………….. 63 4.2.1.3 Additional terrace ……………….. 63 4.2.1.4 Grass strips ……………….. 64 4.2.1.5 Soil loss graph and deposition points ……………….. 64 4.2.1.6 Multiple flow simulations ……………….. 68 4.2.2 Erosion hazard assessment in the Bugsok Subwatershed ……………….. 70 4.2.2.1 Existing land use conditions ……………….. 72 4.2.2.2 Agriculture ≤ 18% slope> forest ……………….. 73 4.2.2.3 Application to land use planning ……………….. 75 CHAPTER 5. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 5.1 Summary ……………….. 76 5.2 Overall Conclusions ……………….. 81 5.3 Recommendations ……………….. 82

REFERENCES ……………….. 84

APPENDICES ……………….. 90

iii

LIST OF TABLES Some erosion rates at plot level extracted from Table 2.1 ………….. 22 different studies conducted in the Philippines

Selected watersheds and sediment yield in the island Table 2.2 ………….. 23 of Mindoro, 1984 (David 1988)

Landcover types and the area of the runoff plots used in the Table 3.1 field monitoring of soil loss and runoff. These land cover ………….. 30 types represent the major vegetation in the area.

Crop parameters and management files taken from WEPP Table 3.2 and GeoWEPP databases (USDA-ARS, 1995) and used in ………….. 38 simulations

Scenarios for the hillslope application of WEPP on steep Table 3.3 ………….. 39 slopes cropped with corn.

Land use scenarios for WEPP watershed application using Table 3.4 ………….. 41 the Bugsok AWaS catchment area.

Percentage contribution of soil loss above 60-mm weekly Table 4.1 rainfall to total soil loss from each experimental runoff ………….. 44 plot.

Week with highest soil loss contribution to the total soil Table 4.2 loss collected within 98 weeks including the respective ………….. 51 runoff and rainfall amounts.

Land cover distribution of the Bugsok and Pamacsalan Table 4.3 ………….. 57 Subwatersheds based on March 2002 Landsat-7 ETM+.

Slope distribution within Bugsok and Pamacsalan Table 4.4 Subwatersheds derived using ArcGIS 9 from a 30-m DEM ………….. 58 and classified according to BSWM cclassification criteria

WEPP simulation of soil, sediment and runoff for non- Table 4.5 ………….. 65 terraced and terraced conditions under different slopes

Percent decrease in soil loss, sediment yield and runoff Table 4.6 resulting from the use of one and two 1-m terraces relative ………….. 66 to a no-terrace condition

Percent decrease in soil loss, sediment yield and runoff Table 4.7 when terraces were replaced with grass strips relative to ………….. 67 no-terrace conditions.

Location of starting points of deposition and soil loss from Table 4.8 ………….. 69 the topmost part of the hillslope as determined by WEPP

iv simulation

On-site effects of land use change predicted by WEPP- Table 4.9 GeoWEPP and presented as percentage distribution of soil ………….. 70 loss under different land cover scenarios

Simulation results of Scenario A representing the area (in percent) occupied by each land cover type classified under Table 4.10 ………….. 72 tolerable and non-tolerable soil loss rates and further classified using the 18% slope criteria.

Off-site effects of land use changes predicted by WEPP- Table 4.11 ………….. 74 GeoWEPP model under each land use scenario

LIST OF FIGURES Location of the Bohol Watershed Project. Inset is the Figure 1.1 ………….. 3 map of the Philippines and the location of Bohol Island

The conceptual components of the Water Erosion Figure 2.1 ………….. 17 Prediction (WEPP) erosion model

Figure 3.1 Geographic location of the Upper Inabanga Watershed ………….. 29

Location of seven runoff plots, two Automatic Water Samplers (AWaS), two Automatic Weather Stations Figure 3.2 ………….. 30 (AWeS) and three rain gauges in the Upper Inabanga Watershed.

Field set up of runoff plots for monitoring soil loss and runoff under the forest land cover type. The galvanized Figure 3.3 ………….. 31 iron sheeting at the bottom part of the plots is secured on concrete lining.

Location of an Automatic Water Sampler (AWaS) along a stream. The pipeline provides a guide and an anchorage for the bubbler tube that was placed at the Figure 3.4 ………….. 34 bottom of the stream. The instrument was placed inside a metal casing and secured in one place on a concrete foundation.

Modification of hillslope topography to effect terracing: Figure 3.5 (a) no terrace (b) one terrace at the bottom of hillslope, ………….. 39 and (c) two terraces

Schematic diagram of overland flow elements for the Figure 3.6 ………….. 40 grass strips simulation

Weekly rainfall from the three localized rain gauges Figure 4.1 ………….. 43 during the 98-week data collection

Weekly rainfall and runoff from the agroforest, cassava/corn, forest, grassland and oil palm Figure 4.2 experimental runoff plots. Right-hand charts show data ………….. 45 points for rainfall below 60 mm and runoff below 0.5 mm.

Comparison of weekly rainfall and soil loss from the agroforest, cassava/corn, forest, grassland and oil palm plots. Right-hand chart shows data points at weekly Figure 4.3 ………….. 48 rainfall below 60 mm and weekly soil loss below 0.005 t·ha-1

Total rainfall, runoff and soil loss accumulated during the 98-week on-site monitoring. The values were Figure 4.4 ………….. 50 computed from weekly data.

vi Weekly rainfall and erosivity relationship computed Figure 4.5 from the 5-min data of Bugsok AWeS. ………….. 53

Weekly rainfall and erosivity relationship computed Figure 4.6 ………….. 53 from the 5-min rainfall data Pamacsalan AWeS

Weekly soil loss and rainfall erosivity for agroforest, Figure 4.7 ………….. 55 cassava/corn, forest, grassland and oil palm plots.

Catchment areas of Bugsok AWaS and Pamacsalan Figure 4.8 AWaS delineated using a DEM and their respective ………….. 57 coordinates.

Average daily discharge records from automatic water Figure 4.9 samplers and daily rainfall data from weather stations at ………….. 59 Bugsok and Pamacsalan.

Rainfall-discharge curves on December 2005 from Figure 4.10 ………….. 61 Bugsok and Pamacsalan monitoring sites.

Rainfall, discharge and sediment yield from Bugsok Figure 4.11 ………….. 62 AWaS recorded on July 22, 2004.

Rainfall, discharge rates and sediment yield from Figure 4.12 ………….. 63 Pamacsalan AWaS recorded on July 15-16, 2004.

Soil loss graph of a 50% slope (A) with 1-m width Figure 4.13 ………….. 67 grass strip at the bottom of the slope and (B) no terrace.

Trend of soil loss and deposition along a hillslope for three scenarios where the 2-grass strip condition has Figure 4.14 four OFEs; 1 grass strip condition has 2 OFEs and the ………….. 69 no grass strip condition has one OFE as determined in WEPP simulations.

Erosion maps of the Bugsok Subwatershed with six land use scenarios showing the on-site effects as predicted using GeoWEPP. White areas within the Figure 4.15 ………….. 71 subwatershed are the channels identified by GeoWEPP but were excluded in erosion simulation with the flowpath method.

vii LIST OF ACRONYMS AND ABBREVIATION

ACIAR Australian Center for International Agricultural Research

AWaS Automatic Water Sampler

AWeS Automatic Weather Station

BSWM Bureau of Soils and Water Management

DEM Digital Elevation Model

DENR Department of Environment and Natural Resources

FAO Food and Agriculture Organization

GIS Geographic Information System

ICRAF International Center for Research in Agroforestry

OFE Overland Flow Element

Philippine Council for Agriculture, Forestry and Natural Resources Research PCCARD and Development

SMU Soil Management Unit

TOPAZ Topographic Parameterization Program

UNCCD United Nations Convention to Combat Desertification

USDA- United States Department of Agriculture –Agricultural Research Service ARS

USLE Universal Soil Loss Equation

WEPP Water Erosion Prediction Project

viii ABSTRACT

To complement the Inabanga Watershed Project (BSWM, 2005), the study reported here was conducted to assess erosion and water resources degradation focused on the Upper Inabanga Watershed using the Water Erosion Prediction Project (WEPP) erosion model and geographic information system (GIS) tools. The study was divided into two sections. The first section was an assessment of the impact of land uses and farm management practices using five runoff experimental plots and two subwatersheds. A 98-week data set from the experimental plots was used to analyze runoff and soil loss linked to weekly rainfall data from localized rain gauges. Discharge rate data from water samplers and rain- intensity data from weather stations was used to characterize the subwatersheds in terms of runoff and sediment yield. The second section of the study was an application of the WEPP and GeoWEPP erosion models. Except for most of the crop management parameters, local climate, soil and topographic parameters were determined and used as inputs to run the model. The results of the study showed that cultivated areas for cassava/corn cropping generated high soil loss (42.5 t·ha-1·yr-1) compared to grassland, oil palm, agroforest and forest which were determined to have soil losses of 6.4 t·ha-1·yr-1, 2.6 t·ha-1·yr-1, 2.3 t·ha-1·yr-1 and 0.2 t·ha-1·yr-1, respectively. The major cause of high soil loss was attributed to farm soil management and cropping operations, which disturbed and exposed the soil surface to the impact of rainfall. The sediment concentrations from the Bugsok Subwatershed and Pamacsalan Subwatershed were as high as 201 mg·L-1 and 782 mg·L-1, respectively, during high rainfall events. These values indicated significant erosion taking place within the subwatersheds. Model simulations of the use of terraces on steep hillslopes (50-70% slopes) predicted reduced erosion by an average of 24% compared to a no-terrace hillslope. A further decrease in erosion was estimated when the terraces were replaced with grass strips. The WEPP-GeoWEPP watershed simulations predicted that any increase in agricultural areas increased on-site soil loss and sediment yield from the watershed.

ix Although constrained by limited input data especially with respect to crop parameters, the application of the model provided an initial step towards understanding erosion processes in the Upper Inabanga Watershed.

CHAPTER 1 INTRODUCTION

1.1 Introduction

Soil erosion is a most important environmental problem in the developing world (Ananda and Herath, 2003). Tropical soils are under particular threat as these soils are less stable than those in temperate climates because of their properties and climatic conditions (Steiner, 1996). Sloping uplands in Asia are particularly threatened by the serious problems of soil degradation (Lapar and Pandey, 1999) since much subsistence farming is carried out on these lands without soil conservation measures. At a global scale, land degradation may not pose a threat to food security but it does pose critical problems in areas where soils are fragile, property rights are insecure, and farmers have limited access to information and markets (Marcoux, 1996).

1.2 The Philippine uplands

Agricultural land degradation in the Philippines is a major environmental and development issue (Cramb, 2000a). FAO (2000), using the GLASOD (Global Assessment of Soil Degradation) database, estimate that 79 % of Philippine lands are threatened by severe degradation. Water erosion is a major driving force of land degradation in the Philippines. The process of land degradation is advancing at an alarming rate due to deforestation and inappropriate agricultural activities. In combination with fragile and highly sensitive mountainous environments, cyclones and frequent thunderstorms result in very high sediment yield rates throughout the country (White, 1995). Erosion rates in Philippines have been estimated at values exceeding tolerable soil loss rates. Soil loss rates of 10 tonnes per hectare per year was considered tolerable by Paningbatan (Paningbatan, 1987 as cited by PCARRD, 1991). In their

1 Philippine sites, the Management of Soil Erosion Consortium (MSEC) research program recorded soil loss of up to 54 tonnes per hectare per year (IWMI, 2002), David (1988) also reported that sediment discharges of Philippine rivers, in which catchments are subject to uncontrolled manipulation, exceed 30 tonnes per hectare per year. For instance, sheet erosion loadings to Magat Dam, in the Northern Philippines, was estimated to be in the order of 88 tonnes per hectare per year (Cruz et al., 1988). Adverse effects of soil erosion from upper watershed regions can be translated in terms of decreasing productivity and income of farmers in downstream areas. A specific case is presented by Lantican et al. (2003). In their study, soil erosion in the Upper Manupali Watershed in Northern Mindanao, Philippines, caused heavy siltation of irrigation canals consequently reducing rice yields of the affected areas by 27%. Excessive siltation decreases the volume of water for delivery to the service areas. As a result, water delivery schedules change, pressing affected farmers to change cropping patterns and/or shift from rice-based to vegetable farming. In order to effectively convey irrigation water to the service areas, silt deposits have to be removed. Removal of silt deposits incurs additional cost on top of regular irrigation management operation and maintenance costs. Several soil-water conservation technologies have been developed and tested (e.g. contour cropping, hedgerows) to address land degradation issues in the Philippines. Contour farming, for instance, has proven appropriate for Philippine uplands (Cramb, 2000b) based on a socio-economic evaluation of soil conservation technologies. Conservation-farming projects, such as ISFP (Integrated Social Forestry Program), have been implemented however with little success. Factors limiting adoption have been identified as the attributes of the technology itself and a range of social, economic and institutional environments where the technologies are promoted (Cramb, 2000b). Promising initiatives such as the Landcare Programs undertaken by ICRAF (International Center for Research in Agroforestry) and other groups are hoped to be more successful (ACIAR, 2004)

1.3 The Bohol Watershed Project

Bohol Island is the 10th largest island in the Philippines situated in the Central Visayas Region. Bohol Island is geographically located within 123o40’- 124o40’

2 longitudes and 9o30’ – 10o17’ latitudes, approximately 625 km southeast of Manila and 75 km east of Cebu (Figure 1.1).

Figure 1. 1. Location of the Bohol Watershed Project. Inset is the map of the Philippines and the location of Bohol Island.

The island has a land area of 411,700 hectares and is home to 1.14 million people. The population is growing at a high rate of 2.95% per annum compared to the national average of 2.38% based on 1995-2000 census data (NSO, 2002). Bohol is an agricultural province where 45% of the land area is cultivated in agricultural pursuits. The island is considered the leading food granary of the Central Visayas Region. Farming is the main source of income followed by fishing with seaweed farming (AusAid, 2001). PCARRD (1984), however, reported that more than half of the island land area is already eroded. Statistics also show that there is a higher incidence of poverty in the island reported at 47.3% compared to the national and regional poverty incidence of 28.4% and 32.3%, respectively (NSO, 2002). Similarly, an Australian

3 AID report (AusAID, 2001) identified widespread poverty especially in the upper catchments and small island coastal zones. Thus, assistance has been recommended to focus on these areas. In response to the needs identified by the Philippine Bureau of Soils and Water Management (BSWM), the University of Western Sydney (UWS) and the Australian Center for International Agricultural Research (ACIAR), a Watershed Project (ACIAR Project LWR1/2001/2003) was established to address the agricultural opportunities and natural resources problems in Bohol Island. The project’s major task was to inventory resources and develop strategies for protecting the environmentally and economically sensitive soil and water resources of the Inabanga River Watershed while maintaining agricultural productivity within the watershed. An integrated watershed approach of resource management was adopted in the project. The approach considered the elements within the watershed including social, economic, political, and environmental aspects in the sustainable management of land and water resources while producing goods and services for human consumption (Cruz, 1999). The approach is founded on the concept that watersheds are formed by natural landmasses and that water flows into a common point. In effect, pollutants and sediments carried by water end up in water bodies. Dealing with these problems as a whole is considered efficient in terms of data collection, monitoring and management rather than doing it in a piecemeal fashion (DeBarry, 2004). The ACIAR Watershed Project has taken advantage of GIS (Geographic Information Systems) techniques to develop a database of the current situation of natural resources and in identifying potential problem areas within the watershed. Field erosion plots provided an estimate of the extent of soil erosion under different land uses while actual surface water monitoring supplied an estimate of runoff generation and sediment yield from selected drainage areas. The project also involved the stakeholders and conducted socio-economic surveys to identify constraints and policy issues affecting soil and water resources use within the watershed. Based on the findings of the project, a number of management options have been recommended. In this project, a range of these options are evaluated using process based erosion model, WEPP (Water Erosion Prediction Project), and GIS

4 techniques. This report is on the process of assessing scenarios of land use and water resources management using an erosion model and GIS techniques. Soil erosion models and GIS are indispensable tools in erosion studies. Erosion models are predictive tools for evaluating the effectiveness of different management methods for conservation planning, project planning, and soil erosion inventories and for regulation (Nearing et al., 1994). Specifically, a process based erosion model predicts where and when erosion is occurring thus helping conservation planners target efforts to reduce erosion. GIS facilitates the collection, storage of data, and construction of model inputs as well as provision of display and analysis tools and presentation of the model outputs. Together with GIS techniques, erosion models provide better means of understanding erosion processes. The application of a process-based erosion model had not previously been carried out in the Upper Inabanga Watershed, and was considered a valuable planning tool for planners, thus the research was conducted.

1.4 Aims

The aims of the project are: 1. To describe the impact of land use management practices in terms of soil loss, runoff and sediment yield in both runoff plots and watershed basis 2. To apply the WEPP erosion model and its geo-spatial interface GeoWEPP a. To predict and simulate soil loss, runoff and sediment yield from agricultural hillslopes incorporating soil conservation measures b. To simulate and predict the effect of land cover change in a selected catchment in terms of runoff and sediment yield

The study is an initial step into understanding erosion processes within the Upper Inabanga Watershed by using an erosion model and GIS. The Upper Inabanga Watershed is an important catchment in Bohol Island since it is the drainage area of the major irrigation facility that supports rice-based farming of the downstream areas.

CHAPTER 2 LITERATURE REVIEW

The following key components of the research program are reviewed:

• Integrated watershed approach of soil and water resources management.

• Processes of soil erosion • Plot-scale and watershed-scale approaches to erosion assessment • Erosion models and GIS techniques • Application of erosion models and GIS A clear understanding of the factors affecting the erosion processes in the Upper Inabanga Watershed was considered as critical for development of relevant input into the modelling process in order to provide a viable tool for resource management and conservation planning.

2.1 The integrated watershed approach of resources management

Major causes of land degradation in Asia include deforestation, shortage of land due to increased populations, poor land use, insecure land tenure, inappropriate land management practices and poverty (FAO, 1995). The integrated watershed approach of managing resources is envisaged as an effective approach and is currently applied in Bohol Watershed Project. A review of the fundamental concepts of this approach is presented. The watershed approach of land and water resources management is based on the understanding that quality and quantity at a point on a stream reflects the characteristics of the upslope area (Davenport, 2003). A watershed is a landscape wherein rainwater is collected and drained at one point called the watershed outlet (Cruz, 1999). It is a dynamic system separated from other watersheds by a high ground perimeter that forms the boundary or watershed divide (Pereira, 1989).

6 The watershed is a major source of nutrients and pollutants (Davenport, 2003; Cruz et al., 1999). As water moves over the land surface, nutrients and sediments are carried with runoff and are deposited in lakes, coastal areas, lowland plains and rivers. These foreign materials can, in many circumstances, have more adverse effects than positive inputs to the new location. For instance, sedimentation of reservoirs is detrimental to the efficiency of the system while enrichment of a water body with nutrients causes eutrophication (Morgan, 2005;Davenport, 2003). Rates of land degradation are strongly influenced by the decisions of upland farmers (Coxhead and Shively, 2005). Disturbances in the upper areas of a watershed are translated into the downstream areas through the hydrologic process (Pereira, 1989). The adverse effects of the upper watershed disturbance directly affect the lives and property of the downstream community, as in the case presented by Lantican et al. (2003). The effect of floods, drought and sedimentation is a concern not just for the upper areas but also of the whole watershed (Pereira, 1989). The ecological linkage between the upstream land uses and the downstream water condition is a strong justification for an integrated watershed-based approach to resource management (Francisco, 2002) A policy note by Francisco (2002) pointed out three important points of undertaking a watershed as a planning unit for soil and water resources management in the Philippines. First, the watershed approach makes it possible to identify various sources of stressors at a point where the watershed drains. Second, stakeholders who have common concerns about the watershed are easily identified and organized. Lastly, various interventions are better implemented and monitored in a well-defined ecological unit such as the watershed. A range of on-site and off-site economic, social and environmental benefits can be derived from a sustainable watershed management as given in Cruz (1999). The economic benefits include production of agricultural crops, forest products for timber and water supply for domestic or industrial uses among others. Improved watershed management also provides social benefits such as reduction of risk to life and property brought about by natural disasters. The environmental benefits of sustainable watershed management include preservation of biodiversity, water and soil conservation and microclimate amelioration. In contrast, failure in watershed management could further exacerbate watershed degradation (Cruz, 1999), hence, reducing economic, social and environmental benefits.

7 Davenport (2003) recommended a watershed management model. This model consists of four phases: assessment phase, planning phase, implementation phase and evaluation phase. The assessment phase consists of a careful analysis of the land and water resources and a definition of issues, problem sources and critical areas within the study area. The assessment phase is the focus of this current review. There are a number of causes of watershed degradation (Cruz, 1999) and soil erosion process is one of the important contributors. To understand these processes, the following section reviews on the basic principles of soil erosion by water.

2.2 Erosion processes

Erosion is a natural process of soil formation. Erosion may either be geologically or human-induced, or a mix of both. Geologic erosion occurs without human interference while human-induced erosion is an accelerated erosion caused by land disturbance for crop production (Lal, 1994). Soil erosion is one form of land degradation characterized by the change in quality of soil, water and other characteristics that reduce the ability of land to produce goods and services that are valued by humans (UNCCD, 1994). The process of erosion by water can be described in three stages: detachment, transport and deposition (Hudson, 1995; Merritt et al., 2003; Morgan, 2005). In the first stage, soil particles are detached by the impact of raindrops or shear forces of flowing water. During the second stage or the sediment transport stage, sediments are moved downslope, which is caused by the splash action of raindrops and runoff. The third stage occurs when runoff velocity is reduced and load carrying capacity decreases causing some or all the sediments to be deposited. There are limiting conditions for erosion process (Hudson, 1995). The erosion process can be described as transport-limited erosion and detachment-limited erosion. Transport-limited erosion occurs when there is not enough runoff to carry away the soil particles detached by rainfall impact while detachment-limited erosion occurs when there is enough runoff to carry more soil than is actually being detached by rainfall impact (Hudson, 1995; Morgan, 2005). There are different types of erosion by water: sheet erosion, rill erosion, interill erosion, gully erosion and stream channel erosion (Schwab et al., 1992). Sheet erosion is the removal of a thin sheet of soil by overland flow (Schwab et al.,

8 1992) on hillsides. Overland flow occurs either when soil moisture capacity or infiltration rate of the soil is exceeded (Morgan, 2005). Rill erosion occurs when flows start to concentrate and create defined flow paths (Schwab et al., 1992). The detachment rate from the rills is a function of hydraulic shear stress of the flowing water, and the rill erodibility and critical shear of the soil, expressed as:

⎛ Qs ⎞ Dr = K r τ −τ c ⎜1− ⎟ ()⎜ T ⎟ ⎝ c ⎠ -2 -1 where: Dr is rill detachment rate in kg·m ·s -1 Kr is rill erodibility resulting from shear stress in s·m

c is critical shear stress below which no erosion occurs in Pa -1 -1 Qs is rate of sediment flow in the rill in kg·m ·s -1 -1 Tc is sediment transport capacity of rill in kg·m ·s is hydraulic shear of flowing water in Pa Interrill erosion is a combination of splash and sheet erosion (Schwab et al., 1992). It is a function of soil properties, rainfall intensity and slope expressed as:

2 Di = i SiK f -2 -1 where: Di is interill erosion rate in kg·m ·s -4 Ki is interrill erodibility of soil in kg·s·m i rainfall intensity in m·s-1

Sf slope factor Gully erosion is a result of too much surface flow either due to climatic conditions or change in land use (Morgan, 2005) . The process is often triggered or accelerated by a combination of inappropriate land use and extreme rainfall events (Valentin et al., 2005). Stream erosion, on the other hand, occurs along stream channels and river banks caused by the shear forces of stream flow.

2.3 Factors of soil erosion processes

Soil erosion hazards are dependent on soil erodibility, rainfall erosivity, slope factor and the type of ground cover (Morgan, 2005; Lal, 1985; Wischmeier and Smith, 1978). These factors are quantified in the Universal Soil Loss Equation

9 (USLE) (Wischmeier and Smith, 1978) to estimate average annual soil loss from agricultural and non-agricultural land. The USLE is expressed using the equation, A= RKLCSP where: A is the estimated soil loss per unit area (t·ha-1·yr-1), R is the rainfall erosivity factor (MJ·mm·ha-1·h-1·yr-1), K is the soil erodibility factor (t·ha·h·ha-1·MJ-1·mm-1), L is the slope length factor, C is the cover management factor, S is the slope steepness factor, and P is the conservation practice factor.

2.3.1 Rainfall

Rainfall impact is the major detachment agent of erosion (Morgan, 1995). The detachment force of raindrops expressed in terms of its kinetic energy is

2 associated with the intensity, KE = mv 2 , where KE is the kinetic energy of rain in Joules, m is the mass in kg and v is the velocity in m·s-1 (Morgan, 1995). A number of experiments have established relationships between the kinetic energy of rain per unit area and the intensity of rain in depth per unit time as presented in a review by Van Dijk et al. (2002). Rainfall erosivity is defined as the potential power of raindrops to detach soil particles from soil mass. In the USLE and revised-USLE erosion models, the R- factor accounts for the erosive power of rainfall and runoff. The R-factor was derived by Wischmeier and Smith (Renard et al., 1996) from research data which indicated that when factors other than rainfall are held constant, soil losses from cultivated fields are proportional to a rainstorm parameter, EI30. This parameter is expressed as the product of total storm kinetic energy (E) and the maximum 30- minute intensity (I30) of a storm. The kinetic energy of rainfall is an indicator of the potential ability of rain to detach soil particles (Salles et al., 2002). While meteorological stations do not measure this parameter, empirical relationships were established to relate kinetic energy and other available rain characteristics such as rainfall intensity. Wischmeier and Smith (1978) derived storm kinetic energy (E) as a logarithmic function of -2 rainfall intensity (I) expressed as, E = 9.11 + 7.8 log I where E is in J·m ·mm and I

10 is in mm·h-1. Brown and Foster (Renard et al., 1996) later recommended an exponential form of this relationship given by, E = 29[1− 72.0 exp()− 05.0 I ]. A critical review of literature on the relationship between rainfall intensity, drop size distribution and kinetic energy, conducted by Van Dijk et al. (2002), resulted in a new general predictive equation proposed as, E = [13.28 − 52.0 exp()− .0 042I ]. The predictive relationship was developed from a range of locations for which good datasets were available. In the tropical areas, Hudson (as cited in Hudson, 1995) found a threshold value of I that above which rainfall is erosive. The Hudson’s erosivity index, KE>25, consisted of the total kinetic energy of all the rain falling at intensity greater than 25 mm·hr-1and provided an excellent correlation with soil loss (Hudson, 1995). E was -2 -1 computed using the relationship, E = 30 −125 I where E is in J·m ·mm and I in -1 mm·hr . Unlike the EI30 of RUSLE, the KE>25 of a storm event is the sum of the product of E and the rainfall amount for each increment where I is greater than 25 mm·hr-1. There are two types of rainfall events related to erosion (Morgan, 2005): the “short duration high intensity rainfall event” where the infiltration capacity of the soil is exceeded and the “long duration low intensity rainfall event” which saturates the soil. The study by Kinnell (1983) on bare plots found that energy of raindrops was less effective in generating soil loss when runoff is absent. The antecedent conditions of the soil may also affect the response of soil to rainfall impact as shown in a number of studies cited in Morgan (2005).

2.3.2 Soil properties

The soil property that is considered in erosion processes is termed soil erodibility. Soil erodibility is define as the soil’s resistance to both detachment and transport (Morgan, 2005). The soil-erodibility factor (K) relates the integrated effect of rainfall, runoff and infiltration on soil loss (Wischmeier and Smith, 1978). The soil properties affecting erodibility are soil texture, aggregate stability, shear strength, infiltration capacity and organic matter and chemical content (Morgan, 2005).

11 In relating erosion with soil properties for tropical conditions, Lal (1985) proposed a simple rating method to assess tillage requirements for different soil conditions. The rating method was developed based on soil and climatic factors such as erosivity, erodibility, soil loss tolerance, compaction, soil temperature regime, available water holding capacity, cation exchange capacity and soil organic matter content based on soil conditions. Each factor was assigned a numerical value and the cumulative rating values for all factors ranged from 14 to 70. Below a cumulative rating of 30, no-till was recommended while for values exceeding 45, a conventional tillage system of plowing and harrowing was suggested. Minimum tillage was suggested for soils with intermediate cumulative ratings according to the study. The erodibility of topsoil however, is also affected by root density as shown in a study of De Baets et al. (2006). The study investigated the impacts of root density and root lengths in natural vegetative grass on soil resistance to erosion by concentrated flow. The study found that soil detachment rates decreased with increasing root density.

2.3.3 Surface cover and cropping practices

Vegetation of all kinds is nature's protective soil cover. Vegetation cover functions in two ways (Morgan, 2005): the above-ground components shields the soil surface from the impact of raindrops while below-ground components, the root system, increase the mechanical strength of the soil. In addition to vegetation, other surface cover such as mulch maintains or enhances the soil's infiltration capacity and retards velocity of runoff. The study of Paningbatan et al. (1995) associated reduced sediment concentration to higher surface cover which effectively protected soil from both rainfall detachment and runoff entrainment. The application of mulching with plant residues and the presence of densely planted hedgerows contributed to a reduction in overland flow that consequently reduced sediment export concentrations. In vegetable-based production systems in Manupali Watershed in the Northern Philippines, a comparison of soil erosion under three different crops showed that under tomato, erosion was high. The high erosion under tomato was attributed to sparse canopy cover and tillage operations.

2.3.4 Topography

The rates of erosion increase with increasing slope steepness and slope length as a result of increasing velocity and volume of runoff (Morgan, 2005). The L-S factors of RUSLE accounted for the effects of topography on soil loss (Renard et al., 1996). The slope length factor (L) can be computed using the equations:

m L l where L is the slope length factor, l is the slope length in metre and m is a = ()22 dimensionless exponent. The exponent m can be computed using: sin s θ , where −1 ⎛ ⎞ , in degrees and s is m = 8.0 θ = tan ⎜ ⎟ sinθ + .0 269()sinθ + 05.0 ⎝100 ⎠ the field slope in percent. Smith and Wischmeier (1978) defined slope length as the horizontal distance from the origin of the overland flow to the point where either the slope gradient decreased enough that deposition occured or runoff became a defined channel. The slope steepness factor (S) accounts for the effect of slope gradient on erosion and can be computed using the following relationships, as presented in Schwab et al. 1992:

8.0 S = 0.3 ()sinθ + 56.0 , for slope length≤ 4 m, S = 8.10 sinθ + 03.0 , for slope length > 4 m and s< 9% S = 8.16 sinθ − 5.0 , for slope length > 4 m and s ≥ 9% The stream power theory was used by Moore and Burch (1986) to derive a physically based length-slope factor in describing the slopes of a landscape. The

4.0 3.1 derived LS factor is expressed as LS = (l 13.22 ) (s .0 0896) Z where Z is a rilling factor that modifies the length-slope factor. Although the derived length-slope factor is equivalent to the USLE L-S factor, hydrological processes are accounted as they affect runoff and erosion. The L-S factor can also be a measure of the sediment transport capacity of runoff from a landscape.

2.4 Erosion Modeling

A comprehensive review of existing erosion and sediment transport models was conducted by Merritt et al. (2003). Recently, an updated review was carried out

13 by Aksoy and Kavvas (2005). Existing models differ in terms of complexity, inputs, processes represented and the manner in which these processes are represented, scale of intended use and the outputs or information provided (Merritt et al., 2003; van Noordwijk et al., 2004). There are also valuable comprehensive agroforestry models, such as WaNuLCAS, developed by van Noordwijk (2004), which take a wide whole- system approach and incorporate extensive soil, plant, cropping, climate and soil management parameters but do not focus as specifically on erosion as do the more mechanistic models. Existing erosion and sediment transport models can be categorized into three types based on physical processes simulated, model algorithms that described these processes and data dependence. These are: empirical, conceptual and physics-based models. The empirical models are considered as the simplest form of mathematical model (Merritt et al., 2003). These are generated by statistical analysis of the relationships between identified factors or variables from a considerable amount of data. A typical example of this model is the relationship between sediment discharge

b (Qs) and water discharge (Q) expressed asQs = aQ (as discussed in Morgan, 2005). The constants a and b could vary widely e.g. according to season and the type of storm events. While this type of model does require less data, a major disadvantage is that there is no indication why erosion takes place (Morgan, 2005). The models make no inferences as to the processes at work, instead relying on the observed or stochastic relationships between the causal variable and the modeled output (Merritt et al., 2003). Conceptual models are based on the representation of a series of internal storages (Merritt et al., 2003). This type of model provides a transfer component for sediment and runoff from one storage system to another system. Within each storage system, a characterization of its dynamic behavior is required. This type of model however, represents a more generalized description of the processes within a watershed while important interactions between these processes may be not included. Using such models, qualitative and quantitative indicators of landuse change can be generated with less input data. Physics-based models are based on the solution of fundamental physical equations (Merritt et al., 2003) such as the laws of conservation of mass and energy. The parameters used in the model are measurable.

14 Models also differ on the way the processes within a modeled area are being represented (Merritt et al., 2003): either lumped or distributed. Within a given area of interest, a distributed model can provide a detailed description of the characteristics of the modeled area by dividing the area into smaller subareas or cells, while a lumped model can only estimate a generalized condition of the area. A distributed model is capable of estimating the spatial variation of soil loss while lumped model can only estimate sediment yield at the outlet. The temporal variation of model predictions is another important consideration in erosion modeling (Merritt et al., 2003). An event-based model is capable of describing the response of a modeled area to high-intensity short-duration storm events. In terms of exploring trends over time to landuse change or management practices or rainfall patterns, the larger temporal resolutions model can be of an advantage. A continuous time-step type model, usually applying a daily time-step, can best simulate catchment behavior such as recession time of floods. Across all models reviewed by Merritt et al. (2003), the concepts behind each model can be, or have the potential to be incorporated into catchment scale approaches. Since there is no best model for all applications, the most appropriate model in a given situation depends on the intended use and the characteristics of the catchment under consideration (Merritt et al., 2003). Descriptions of some commonly used models are provided below.

2.4.1 USLE model

The Universal Soil Loss Equation (USLE) (Wischmeier and Smith, 1978) is the most widely accepted and utilized method of estimating soil loss. The average annual soil loss estimated by USLE is dependent on factors such as annual rainfall, estimate of soil erodibility, land cover and topographic information, as presented previously. The USLE can estimate long-term annual soil loss and guide conservationists in proper cropping, management, and conservation practices; however it cannot be applied to a specific year or a specific storm (Wischmeier and Smith, 1978). Due to several limitations of the method, the USLE has undergone modifications (i.e. modified-USLE and revised-USLE) to accommodate other parameters in the soil erosion process.

2.4.2 WEPP Model

The Water Erosion Prediction Project (WEPP) model has been considered to represent a major improvement to the previously most widely accepted method of estimating sediment loss, the USLE (Universal Soil Loss Equation) (Flanagan and Nearing, 1995). The model represents a new erosion prediction technology based on fundamentals of stochastic weather generation, infiltration theory, hydrology, soil physics, plant science, hydraulics, and erosion mechanics (Flanagan et al., 1995). The model has two applications: the hillslope and watershed applications. As applied to hillslope, the model is subdivided into nine conceptual components, namely: climate generation, winter processes, irrigation, hydrology, soils, plant growth, residue decomposition, hydraulics of overland flow and erosion (Flanagan et al., 1995). The hillslope model include erosion processes such as rill and interrill erosion, sediment transport and deposition, infiltration, soil consolidation, residue and canopy effects on soil detachment and infiltration, surface sealing, rill hydraulics, surface runoff, plant growth, residue decomposition, percolation, evaporation, transpiration, snow melt, frozen soil effects on infiltration and erodibility, climate, tillage effects on soil properties, effects of soil random roughness, and contour effects including potential overtopping of contour ridges. The model accommodates the spatial and temporal variability in topography, surface roughness, soil properties, crops, and land use conditions on hillslopes. The watershed application integrates erosion at hillslope, channels and impoundment (Ascough II et al., 1995). This application was developed to predict erosion effects from agricultural management practices within small agricultural watersheds. The development of the WEPP watershed model erosion component is based from the fact that watershed sediment yield is a result of detachment, transport, and deposition of sediment on overland (rill and interrill) flow areas and channel flow areas, that is, erosion from both hillslope areas and concentrated flow channels were desired to be simulated by the watershed version. The documentation for WEPP erosion model is presented in Flanagan and Nearing (1995). Figure 2.1 shows components of WEPP model. A brief summary is given in the following sections.

Erosion processes Climate generation

Surface and subsurface Soils hydrology component Conceptual components

Water balance and percolation Plant growth component Residue decomposition and management

Figure 2.1. The conceptual components of the Water Erosion Prediction (WEPP) erosion model.

2.4.2.1 Hillslope erosion component The hillslope erosion component of WEPP is presented by Foster et al. (1995). The WEPP erosion model computes soil loss along the hillslope and sediment yield at the end of the hillslope. The erosion processes considered in the WEPP erosion model are rill and interrill erosion processes. The movement of sediments is described by a steady-state sediment continuity equation, ∂G = D f + Di ∂x -1 -1 where G is the sediment load (kg·s m ), x is the distance downslope (m), Df is rill -1 -2 -1 -2 erosion rate (kg·s m ) and Di is interrill sediment delivery to the rill (kg·s m ). Rill erosion is a function of the flow’s ability to detach sediment, sediment transport capacity and the existing sediment load in the flow, expressed as,

⎛ G ⎞ D f = Dc ⎜1− ⎟ ⎜ T ⎟ ⎝ c ⎠ -1 -2 where Dc is detachment capacity (kg·s m ) and Tc is sediment transport capacity in the rill (kg·s-1m-1). Net soil detachment in rills occur when the hydraulic shear stress

17 of the flow exceeds the critical shear stress of the soil and when sediment load is less than sediment transport capacity. Dc is expressed as,

D K c = r (τ f −τ c )

-1 where Kr (s·m ) is a rill erodibility parameter, τf (Pa) is flow shear stress acting on the soil particles, and τc (Pa) is the critical shear stress of the soil. Rill detachment is considered to be zero when flow shear stress is less than the critical shear stress of the soil. Deposition in rills occurs when the sediment load, G, is greater than the sediment transport capacity, Tc, and is computed using the equation,

βV f D f = Tc − G q ()

-1 where Vf is effective fall velocity for the sediment (m·s ), q is flow discharge per 2 -1 unit width (m ·s ) and β is a raindrop-induced turbulence coefficient. Interrill erosion is a process of soil detachment by rainfall impact, transport by shallow sheet flow, and sediment delivery to rill channels, expressed as:

⎛ Rs ⎞ Di = K iadj I eσ ir SDRRR Fnozzle ⎜ ⎟ ⎝ w ⎠ -1 where Kiadj is adjusted interrill erodibility, Ie is effective rainfall intensity (m·s ), ir -1 is the interrill runoff rate (m·s ), SDRRR is a sediment delivery ratio which is a function of the random roughness, the row side-slope and the interrill sediment particle size distribution, Fnozzle is an adjustment factor to account for sprinkler irrigation nozzle impact energy variation, Rs is the spacing of the rills (m), and w is the rill width (m). The sediment delivery to rills is assumed as proportional to the product of rainfall intensity and interrill runoff rate.

2.4.2.2 Hillslope surface hydrology The surface hydrology component (Stone et al., 1995) provides the erosion component with the duration of rainfall excess, the rainfall intensity during the period of rainfall excess, the runoff volume, and the peak discharge rate. This component also estimates the amount of water, which infiltrates into the soil for the

18 water balance and crop growth/residue decomposition calculations which are in turn used to update the infiltration, runoff routing, and erosion parameters.

2.4.2.3 Water balance and percolation The water balance and percolation components (Savabi and Williams, 1995) estimate soil water content in the root zone and evapotranspiration losses throughout the simulation period using the input from the climate, infiltration, and crop growth components. Evapotranspiration and percolation are predicted over 24 hour period.

2.4.2.4 Subsurface hydrology The subsurface hydrology component of WEPP model predicts the effects of soil-water content in the generation of runoff. The distribution of soil water in the root zone also affects the interaction between soil water and plant growth and in residue decomposition.

2.4.2.5 Soil component The soil component describes the soil and soil-related variables that are predicted in WEPP. These variables include random roughness (associated with tillage), oriented roughness (height of ridges), bulk density (total pore volume of soil to predict infiltration parameters), effective hydraulic conductivity (for predicting infiltration and runoff), interrill erodibility (soil resistance to rainfall detachment), rill erodibility (soil’s resistance to detachment by concentrated flow), and critical shear stress (limit for shear forces to cause detachment).

2.4.2.6 Plant growth component The plant growth component of WEPP model (Arnold et al., 1995) simulates the impact of plant growth parameters on the hydrologic and erosion processes. The parameters simulated include canopy, cover and height, root growth and leaf area index.

2.4.2.7 Climate component The climate component of WEPP is a stochastic generation of climate parameters such as mean daily precipitation, daily maximum and minimum temperature, mean daily solar radiation and mean daily wind direction and speed

19 from historical data. A two-state Markov chain model is used to generate the number and distribution of precipitation events. A disaggregation model is also included to generate time-rainfall intensity from daily rainfall amounts which is needed to compute rainfall excess rates or runoff.

2.4.2.8 Residue decomposition and management The residue decomposition component of WEPP simulates decomposition of flat residues, standing material, submerged material and root mass. Harvesting and management options such as tillage, burning and shredding are also simulated.

2.5 Estimates of erosion at a plot and watershed scale

There are a number of valuable possible uses of runoff plots (FAO, 1993). Runoff plots can be used to a) demonstrate to farmers known facts such high erosion under bare plot compared to good vegetation cover; b) test and demonstrate the effect of conservation measures, such as mulching, on runoff and soil erosion; and c) collect data to construct and /or validate a model for predicting runoff and soil loss. Field plots range in sizes: from small, USLE plots and the unit-source watersheds (Mutchler et al., 1994). According to Mutchler et al., (1994) small plots, usually about 1 m2 area, are used in studying the basic aspects of soil erosion in detail as well as in developing or verifying basic operating equations that govern the physical processes of soil erosion. What are termed “standard USLE plots” (22.1m x 1.9 m on 9% slope) are used to study the combined effects of rill and interill erosion. Plots of 20 m long and 5 m wide are used for trials of cropping practices, cover effects, rotation and other practices (FAO, 1993). The unit-source watersheds are plots with at least one channel and which contain a single crop (Mutchler et al., 1994). The effects of all erosion processes and conservation measures are combined in one measurement (Mutchler et al., 1994) in a unit-source watershed. Watershed studies (FAO, 1993) take into account the real situation of erosion processes within a watershed. A sediment delivery ratio (SDR) (Walling, 1994) is used to calculate gross erosion rates within a watershed, computed as: SDR = sediment yield/gross erosion SDR is unique to every watershed and is influenced by a wide range of geomorphological and environmental factors including the nature, extent and

20 location of the sediment sources; relief and slope characteristics; drainage patterns and channel conditions; vegetation cover; land use; and soil texture (Walling, 1994). Sediment yield is the quantity of sediments leaving a watershed along a river over time (Morgan, 2005) and is commonly determined by monitoring suspended solids over a period of time (Walling, 1994).

2.6 Erosion studies in the Philippines

2.6.1 Plot scale Across the country, several erosion studies have been carried out to measure soil loss from different cropping systems and farmer practices. Results of a few studies are presented in Table 2.1. The studies were conducted to estimate the impact of management practices in terms of soil loss as compared to the traditional or conventional farming practice especially in the sloping uplands in the country. Paningbatan et al (1995) conducted a field experiment on a hillslope in UPLB Experimental Farm, Laguna to test and compare three soil conservation-oriented alley farming cropping treatments with farmer’s practice in terms of their effects on soil erosion and runoff. Twelve runoff plots with a dimension of 6-m width and 12- m length were laid out for the four treatments, which were replicated thrice. The four treatments were: T1) farmer’s practice which involves up-and-down slope tillage operations and clean or weed-free cultivation, T2) alley cropping, contour tillage and residue/weed free, T3) alley cropping, contour tillage and mulching (hedgerow trimmings and crop residues), and T4) alley cropping and zero tillage with mulching. Results showed a very large difference in soil loss from the farmer’s practice as compared to the alley cropping system (Table 2.1). This large difference has been attributed to the lack of protective surface cover especially during heavy rains. This trend of soil loss was also confirmed in the study of Poudel et al (1999) in the Manupali Watershed, Southern Philippines. The study of Poudel et al (1999) tested the effectiveness of soil conservation measures under sloping intensified vegetable system in reducing soil erosion. Soil conservation practice such as contouring, strip cropping, and high-value contour hedgerows were compared with the farmer’s practice of up-and-down cultivation. From 1995-1998, soil loss measurements on 42% slope showed that the up-and-down system of farming generated the highest soil loss as compared to the other conservation practice as shown in Table 2.1.

21 In a similar study conducted on the hillslopes of Visayas State College of Agriculture in Leyte, Philippines, Presbitero et al (1995), measured runoff and soil loss from a range of runoff plots including cultivated bare soil, common agricultural practices, and soil conserving practices such as intercropping and hedgerows. Results of the measurement revealed the highest soil loss and runoff from the bare plot as shown in Table 2.1. The three studies showed the negative impact of traditional practice of up-and-down farming in terms of soil erosion and the effect of conservation measures such as contouring and hedgerows in reducing soil loss on sloping agricultural lands.

Table 2.1. Some erosion rates at plot level extracted from different studies conducted in the Philippines

Soil loss Soil Location Vegetation/cropping system Slope (t·ha-1·yr- Source type 1) Los Banos, Farmer’s practice (up-and- 15-29 150 Luzon down slope tillage) % Paningbatan Clayey (conca et al (1995) Alley cropping 5 ve) ViSCA, Bare plot (cultivated and kept 68.6 Visayas bare) Maize/sweet potato rotation 38.1 Ipil-ipil hedges (upslope and 18.6 Presbitero downslope ends) 50 % Clayey et al (1995) Ipil-ipil hedges (upslope and downslope ends) 2.7 intercropped with peanut/main crop Up-and-down tillage system 65.3 Manupali Contouring 37.8 Poudel et al Watershed, Strip cropping 42% Clayey 43.7 (1999) Mindanao High-value contour hedgerows 45.4

22 2.6.2 Watershed scale Sediment yield is a measure of the annual sediment discharge at a given monitoring point across a stream. The point of observation defines the boundary and basin area of the watershed. David (1988) reported on a few estimates of sediment yield from selected watersheds in the island of Mindoro, Philippines. Table 2.2 shows the percentage land cover type and sediment discharges for the Bucayao and Bugsuanga Watersheds in Mindoro Island. Sediment discharge in the Bucayao Watershed is far small as compared to the Bugsuanga Watershed. The presence of forest areas, nearly half of the Bucayao Watershed, resulted to low sediment discharge rates. David (1988) indicated that sediment discharge from watersheds with primary forest is low averaging less than 0.5 t·ha-1·yr-1. Good forest cover and cover litter protect the soil against raindrop energy, intercept rainfall, improve soil structure, aggregation and infiltration, and increase the soil-surface resistance to overland flow. In the Bugsuanga Watershed grassland areas cover nearly half of the watershed. The grassland areas are open pasture or rangelands, which are usually overgrazed (David, 1988). Overgrazing disturbs and reduces vegetation cover exposing the soil to the impact of raindrop energy and runoff that could result to higher erosion rates.

Table 2.2 Selected watersheds and sediment yield in the island of Mindoro, 1984 (David 1988) Percentage Sediment Discharge Watershed Land cover type (%) (t·ha-1·yr-1) Forest 45 Bucayao Cultivated area 33 Watershed Grassland (good stand) 4 4.0 Area = 384 Savannah (cogon and talahib km2 grasses with shrubs and 18 brushes Forest 16 Bugsuanga Cultivated area 18 Watershed Grassland (good stand) 45 233.3 Area = 438 Savannah (cogon and talahib km2 grasses with shrubs and 18 brushes

23 2.7 Erosion models and GIS

The spatial dimension of environmental modeling, e.g. erosion and watershed models, provides an opportunity for GIS coupling for both input data preparation as well as for display and further analysis of model results (Fedra, 1993). GIS facilitates a fast and efficient means of generating the input data required for the model (Cox and Madramootoo, 1998). GIS handles and provides large amounts of detailed input data reducing uncertainty due to spatial averaging (Lenzi and Di Luzio, 1997). The graphical displays of the model results have proven to be a very effective and efficient way of interpreting the results and of decision making during model calibrations (Lenzi and Di Luzio, 1997). GIS allows for an easy assessment of erosion hazard over a watershed and under different land use management options (Cox and Madramootoo, 1998). The GIS linkage facilitates the use of readily available geo-spatial data of highly variable precision and accuracy, and allows for communicating with a diverse spectrum of users with different levels of expertise (Renschler, 2003). Cochrane (1999) developed two methods of incorporating GIS and DEM in the application of the WEPP watershed model. These methods are hillslope and flowpath methods. Cochrane (1999) emphasized that the flowpath method does not have a channel flow routing component and can only be applied to a watershed whose channels are not considered with a depositional or erosional mode. A GIS interface of WEPP was developed to automate the identification of hillslope and channel profiles (Cochrane, 1999) for use by the WEPP model. This procedure replaced the time-consuming manual identification of hillslopes and channels, which can vary from user to user. A digital elevation model (DEM), raster representation of topography, is used to derive the hillslope and channel topographic characteristics in the GIS environment. Cochrane and Flanagan (1999) described and evaluated three approaches of using GIS tools and data to facilitate the application of WEPP in predicting erosion in small watersheds. The three approaches are the manual method, hillslope method and flowpath method. These approaches are summarized in here. In the manual method, watershed components such as channels and hillslopes are set up and described using ArcView GIS from ESRI (Environmental Systems Research Institute). Channel locations are identified either by on-screen digitization or by

24 automatic extraction from a DEM. Hillslopes are defined by digitizing the hillslope boundaries using on-screen digitization. A line segment is drawn over each hillslope to indicate location and define hillslope profile. The line segment is overlaid on the DEM to define actual elevations. The width of the hillslope is computed using the area of the hillslope and the length of the profile. Soils and crop management inputs for each hillslope are taken from the soils and crop management maps. In the hillslope method, watershed components are automatically defined from a DEM by computer algorithms such as TOPAZ (Topographic ParametiZation) program. TOPAZ, a software package for analysis of digital landscape (Garbrecht and Martz, 1999), is used to identify channels and hillslopes and other landscape parameters. A critical source area is required to define the location of channel initiation by concentrated flows. Using a DEM, hillslopes represent a group of cells draining from the top, left and right of the channels. Each hillslope is composed of one or more flowpaths. The flowpaths define the route of water from one cell to another and end at the channel. The representative profile of the hillslope is computed by weighting all possible flowpaths for each hillslope. WEPP is applied based on the computed hillslope profile. Unlike the hillslope method, in the flowpath method, WEPP is applied to all possible flowpaths within the watershed using the individual flowpath profile. The width used by WEPP for each flowpath is the total area of all the flowpaths that contribute to a cell along a channel divided by the length of a flowpath. Soil loss and runoff values are computed for each flowpath. Deposition and detachment values of a cell along interacting flowpaths are weighted based on the length and drainage area of the flowpaths. Sediment yield and runoff values of each cell along a channel where flowpaths discharge is computed based on weighted average contributions from each flowpath. The WEPP watershed model was applied in selected upland watersheds in the Philippines in the work of Ella (2002). His work found major limitations of the model (April 2002 version) such as inability of the model to handle more than 30 interconnected channels and inability to handle numerous breakpoint rainfall data. His work was also constrained to a large extent by the unavailability of recent land cover data at the site. To attain a high degree of accuracy using the WEPP model, the study emphasized the importance of improving data collection methodologies and instrumentation such as the use of automatic weather stations at the site. In addition,

25 the availability of updated and accurate geo-referenced land cover data could further improved the model. While the study was constrained to some extent by data availability and model limitations, initial modeling results showed that land cover disturbances had a very pronounced effect on sediment yield.

2.8 GIS application watershed management

One of the approaches in addressing heterogeneity of physical properties and land use patterns over a watershed is by using simulation models (Cox and Madramootoo, 1998). Physical process based models and GIS are useful tools in developing watershed management strategies (Behera and Panda, 2006). Cox and Madramootoo (1998) presented preliminary results from a study on the application of a GIS and erosion model in developing watershed conservation strategies for St. Lucia Watershed in Eastern Caribbean. The main objective of the study was to develop GIS-based procedures in evaluating land management practices in terms of soil loss potential. The study used the revised-USLE (RUSLE) in estimating annual soil loss and employed GIS in preparing the input data, executing the model computations and displaying the results. The approach was applied to two agricultural watersheds and four management scenarios were developed for soil loss prediction. In the study, three primary GIS-based data layers were required to run the RUSLE model, namely, DEM, soil type layer and land use layers. The slope length (L) and slope steepness (S) factors were derived from the DEM. The soil erodibility factor (K) was derived from the soil type layer. The crop management (C) and conservation practices (P) factors were estimated based on the land use layer. The rainfall erosivity (R) factor was derived from an isoerodent map. In the study of Cox and Madramootoo (1998), management scenarios were developed based on the current and recommended land use conditions. Under those two conditions, simulations were run with and without conservation support practices. The recommended land uses were based on a land capability classification considering slope steepness and soil stability. The results of the simulation showed a decrease in average annual soil loss from the current land use to the recommended land use. Further decrease in soil loss was determined when support conservation practices were applied to the recommended land use and compared to a no conservation practice scenario.

26 In India, a process based watershed scale model, soil and water assessment tool (SWAT), and GIS were used to identify critical areas and develop best management strategies for a small watershed in Midnapore district of West Bengal (Behera and Panda, 2006). The SWAT model was calibrated and evaluated using observed hydrologic and water quality data from the monitored watershed. The SWAT satisfactorily predicted daily runoff, sediment yield and nutrient concentration in runoff according to standards statistics. Using the model, critical areas were identified based on average annual sediment yield and nutrient losses. Management scenarios for rice-based cropping were then simulated. Based on the simulations, conservation tillage, with a mixing efficiency 0.25, was found to be effective in terms of reducing runoff and nutrient losses compared to the existing conventional tillage with a mixing efficiency of 0.5. The study recommended a fertilizer application rate of 80:60 kg·ha-1 of N:P. At this application rate, rice yield was comparable with a fertilizer application rate of 120:80 kg·ha-1 of N:P and surface water pollution caused by NO3-N and P was insignificant based on the results of the study. In the Philippines, Ella (2005) applied the Water Erosion Prediction Project (WEPP) erosion model to Maagnao Watershed in Mindanao Island. Management scenarios varying the percentages of cropped areas were developed. Results of the simulations showed that increasing the percentages of cropped area also increased the sediment yield from the watershed. When the watershed was entirely uncultivated for crop or only shrubs and small trees were present, sediment yield was estimated at 1.9 t·ha-1·yr-1. On the other hand, when the entire watershed was subject to cultivation, sediment yield was predicted at 48 t·ha-1·yr-1. Although constrained by the available input data, the study by Ella (2005) provided an initial step into the application of a process based erosion model in the Philippines. Initial results of the study indicated the potential impact of land use change in the study watershed. This result could be applied to other watersheds in the Philippines. The study also unveiled the need for updated land cover data, agricultural crop management practices, soil properties, climate data to name a few, in order to proceed with erosion estimation as an input into watershed planning and management in the Philippines. Recommendations and approaches from the study were used in formulating and undertaking the study reported in this current investigation.

CHAPTER 3 MATERIALS AND METHODS

This chapter describes the study area and research methods used in conducting the study. The chapter is divided into three sections. The first section describes the study area, the second section describes the on-site data collection and the third section describes computer simulations using WEPP and GeoWEPP models.

3.1 The study area

The Upper Inabanga Watershed is the drainage basin of the Malinao Dam Reservoir (Figure 3.1). It covers the two upstream sub-watersheds of the biggest watershed in the island, the Inabanga Watershed. The Upper Inabanga Watershed lies across the boundaries of five major municipalities, namely: Sierra Bullones, Pilar, Garcia Hernandez, Jagna, and Duero. The Malinao Dam Reservoir is the convergence point of the two major tributaries. The two major tributaries are the Pamacsalan River in the eastern part and the Wahig River in the southwestern side. The dam was designed to serve about 5000 hectares of adjoining agricultural land since 1996 and has a catchment area of about 13,800 hectares including a 140-ha reservoir. The reservoir is situated at an altitude of about 140 m above mean sea level while the highest elevation of the catchment is at 861 m. The Upper Inabanga Watershed is an agricultural area. The majority of the population derives its income from cultivating agricultural crops. The major crops include rice, corn, cassava, rootcrops, coconut, vegetables and other economic crops as well as the newly introduced oil palm trees. Backyard animal raising supplements crop production. Usually farmers raise cattle, swine and poultry at backyard scale, and water buffalo as draft animals. The watershed has been the focus of a number of conservation programs by government and non-government organizations such as ICRAF and Soil and Water

28 Conservation Foundation (SWCF). These programs address growing concerns for land and water degradation within the watershed and target farmers and organized communities as media for forest restoration and resource conservation activities and as catalysts for the poverty reduction strategies of the government.

Figure 3. 1. Geographic location of the Upper Inabanga Watershed.

3.2 On-site assessment of soil loss and runoff

The on-site assessment of soil erosion, runoff and sediment yield were carried out on a plot and catchment scale. Seven major land cover types were identified for use in the plot scale and two subwatersheds receiving streams allow upscaling of model simulations. Climate monitoring was also included in the on-site activities. Figure 3.2 shows the location of the instruments used in the on-site monitoring activities.

29 Table 3.1. Landcover types and the area of the runoff plots used in the field monitoring of soil loss and runoff. These land cover types represent the major vegetation in the area. Land cover types Area (m2) Slope (%) Agroforestry 101.24 30 Cassava/corn area 76.24 10 Forest 101.24 48 Grassland 98.74 10 Oil palm area 151.95 10

3.2.1. The runoff plots

The field measurement of soil loss and runoff at the plot scale were carried out using runoff plots. The runoff plots were set up under five major land cover types. The land cover types and the dimension and average slopes of each plot are presented in Table 3.1. The field installation and set up of these plots is shown in Figure 3.2.

Figure 3.2. Location of five runoff plots, two Automatic Water Samplers (AWaS), two Automatic Weather Stations (AWeS) and three rain gauges in the Upper Inabanga Watershed.

30 20 m

5 m

(top view) Collecting tank

(side view)

Mechanical runoff recorder

Figure 3.3. Field set up of runoff plots for monitoring soil loss and runoff under the forest land cover type. The galvanized iron sheeting at the bottom part of the plots is secured on concrete lining.

Soil loss and runoff monitoring, was carried out on plots set up along the slopes with lengths varying from 15-20 metres and width of 5 metres, as illustrated in Figure 3.2. The plots were replicated three times. The plots were enclosed by 30 cm high galvanized iron sheeting buried to a 20 cm ground depth. The sheeting barrier prevented surface flow from entering or leaving the plot except through the collecting tank placed at the lower end of the plot. A collecting tank, approximately 100 L, was placed at the bottom end of each plot to measure the volume of runoff. The collecting tank was constructed with five discharge holes of uniform diameter. One of these holes was connected to a mechanical runoff recorder with a tipping bucket inside to measure the flow. This means that 1/5 of runoff flowing the collecting tank passed through the mechanical runoff recorder. The mechanical runoff recorder had a triangular tipping bucket with a capacity of 5 litres each tip. BSWM Soil Conservation Section fabricated the runoff

monitoring equipment in-house. Total runoff volume (VT) was computed as

VT = VTank + 5⋅Vtippingbucket ⋅ N where Vtank is volume of the tank, N is the number of

tips, and Vtipping bucket is the capacity of the tipping bucket, which was 5 litres.

31 Runoff samples were collected on a weekly basis. Sampling was done by mixing thoroughly the water inside the tank and then immediately collecting the sample. Three samples were collected from each plot using pre-washed 500 ml plastic bottle and stored in an ice chest. The refrigerated samples were submitted to the laboratory for total-N, total P and total suspended sediment (TSS) analysis.

3.2.1.1 Agroforests The agroforestry site was situated on a terraced plot of land with an average slope of 30%. Along the 20-m slope length, there were eight flat terraces. Several years prior to this study, the site was planted with upland rice (Oryza sativa L.). However, it was then planted with banana (Musa ssp.) and ipil-ipil trees (Leucaena leucocephala Lam.). The conversion from rice to other crops was the result of the decreasing soil moisture and, ultimately, an insufficient water supply for rice. During the period of data collection, the area was covered with cogon grass (Imperata cylindrical L.) and gabi (Colocasia esculentum L.) at the bottom of the plots in addition to the existing banana and ipil-ipil trees.

3.2.1.2 Cassava/corn The plots under cassava (Manihot esculenta Crantz) and corn (Zea mays L.) area were the only tilled area. During the start of the monitoring, the area was planted to cassava. After cassava, it was planted to corn. This cropping system was a typical practice not only in the study area but also for the whole island. The practice was also common to smallholder farmer. The cassava crops were grown for more than a year while corn crops were planted for 5-6 months to provide basic sustenance, as well as immediate cash crop for the common Boholano farmers. During the course of the monitoring activity, farming practices included ploughing, weeding, fertilizer application and harvesting. These practices were documented in the datasets.

3.2.1.3 Forest The forestry site was part of the local government’s reforestation project. The area was generally covered with mahogany trees (Swietinia macrophylla King.) and gemelina trees (Gmelina arborea Roxb.). Forest litter covered the ground during the period of data collection. The trees were estimated to be more than 30 years old.

32 Burning occurred in the area in the 1970s so that some trees were considered regeneration.

3.2.1.4 Grassland The grassland area was generally covered with carabao grass (Paspalum conjugatum P.J. Bergius), which was under pasture prior to putting up of the erosion plots. The area was oftentimes a passageway by people so the soil was compacted. The average slope was about 10%.

3.2.1.5 Oil palm This plot was planted to a 3-year oil palm trees (Elaeis guineensis Jacq.) Under the oil palm trees was carabao grass (P. conjugatum). The area was also planted to corn (Z. mays ) during the early start of the monitoring period.

3.2.2 Watershed measurement of flow Automatic water samplers, ISCO 6712 Portable Sampler, with integrated flow monitoring modules, were installed in selected subwatershed streams to monitor surface flow and collect water samples. Figure 3.4 shows a typical layout of the equipment. A topographic survey was conducted to determine the cross-sectional area of the stream, establish a stage-discharge relationship and locate the instrument. Water level was continuously monitored from May 2004 to December 2005. The water level was determined by a differential pressure transducer and bubbler mechanism through an ISCO 730 Bubbler Module that was installed with the sampler. Discharge was computed based on the discharge-level relationship that was established for the stream. The water sampler was programmed to collect samples at specified time intervals once a threshold water level was reached. Samples were drawn during storm events. Collected samples were sent to laboratory for nutrient and sediment analysis.

Automatic water sampler

Pipeline

Figure 3.4. Location of an Automatic Water Sampler (AWaS) along a stream. The pipeline provides a guide and an anchorage for the bubbler tube that was placed at the bottom of the stream. The instrument was placed inside a metal casing and secured in one place on a concrete foundation.

3.2.3 Climate Two automatic weather stations (AWeS) were used to monitor the climate in the Upper Inabanga Watershed: one in Bugsok, Sierra Bullones, and one in Pamacsalan, Pilar. The Watchdog Weather Station 900ET model measures rainfall, temperature, relative humidity, solar radiation, and wind direction/speed at 5-min intervals. In addition to the weather stations, there were three cumulative rain gauges installed near the runoff plots, as shown in Figure 3.1. A 49-year daily rainfall data was also available from an existing Dagohoy Rain Gauge Station, located approximately 20 km downstream of the subwatershed.

34 3.2.3.1 Rainfall erosivity Rainfall erosivity was computed using Hudson’s method (Hudson, 1995), KE>25mm/hr. During a rainfall event, incremental rainfall intensities greater than 25mm/hr were identified. The kinetic energy of rainfall at this intensity was computed using the equation, KE = ( 8.29 − 270 I)/ R , where KE is the kinetic energy of rainfall (J·m-2), I is rainfall intensity (mm·h-1) and R is rainfall depth. The erosivity (E), in J·m-2, of a rainfall event is the summation of the kinetic energy (KE) of all incremental rain having intensity greater than 25mm/hr. Erosive rain was computed as the proportion of the total rainfall depth having intensity greater than 25mm/hr.

3.3 Computer simulations

The model used in the study is the WEPP erosion model (Version Jan. 14, 2005) and GeoWEPP (ArcX 2004-3) interface, developed by USDA-ARS, NSERL and Purdue University, and were downloaded from USDA-ARS website. The model simulations conducted in this study were undertaken to explore the application of the model and to assess the potential impact of land use changes and farming practices. Hypothetical scenarios were created and the model was applied to predict soil loss, runoff and sediment yield under varying conditions.

3.3.1 Inputs for WEPP and GeoWEPP models To be able to run WEPP model, a number of data inputs were required. There are two applications of the WEPP model: hillslope and watershed application discussed in Chapter 2 Section 2.4. For the WEPP hillslope application, the inputs were grouped into four data files: climate, soils, crop management, and slope files. In the WEPP watershed application implemented using GeoWEPP (ArcX 2004-3), additional inputs are needed such as soils map, land cover map and a DEM.

3.3.1.1 Climate file In this study, the BPCDG program was used to generate a climate input file using one-year climate record. BPCDG was preferred in order to utilize the actual climate data from automatic weather stations (AWeS) and to gain experience in the

35 process of creating a climate input file in a format acceptable by WEPP. There were four input files needed to run BPCDG (Zeleke et al., 1999), namely:

• xxyyyyPL.CSV file containing the storm/rainfall characteristics such as day and month of the year, beginning and ending time of the storm, and the intensity of the storm on each day;

• xxyyyyCS.CSV file which contains the date, daily minimum and maximum temperatures, and wind velocity and direction at 8 and 18 hours;

• xxyyyyCL.DAT file which stores the Julian day, radiation (monthly or annual), dew point temperature (monthly or annual), and conversion tables for wind velocity and direction; and

• xxyyyyST.DAT file which contains the station name, location, elevation, and years of record information. The “xxyyyy” is a code for station name (first two letters) and year of data compilation (last four letters). In this study, two climate files were generated: BU2005 for Bugsok AWeS and PA2005 for Pamacsalan AWeS using 2005 climate records. The procedure for using BPCDG is given in Zeleke et al. (1999). Pre- processing of the data was carried out in Excel.

3.3.1.2 Slope file The slope input file requires slope segment, slope gradient (in percent), slope length, and aspect values. In the WEPP hillslope application interface window, these inputs were manually encoded. In the watershed application using GeoWEPP, these values were automatically extracted from a DEM and saved in individual file. The DEM was the source of slope and other topographic characteristics needed in running the WEPP model. In GeoWEPP, a topographic parameterization algorithm, TOPAZ (Garbrecht and Martz, 1999), was employed to identify hillslopes and channels from a DEM. A DEM is one of the core inputs in running the geo- spatial interface of WEPP and was not available for the study site. A DEM was created from digitized contours lines and streamlines in ArcGIS 9. The procedure is presented in Appendix A.

36 3.3.1.3 Soils input file and soil map The soil input file of WEPP requires values on percentages of sand, clay, organic matter, and cation exchange capacity (CEC) for each layer profile. In the WEPP window, these values were entered manually. Soil data is shown in Appendix B-1. Other soil parameters such as interill and rill erodibilities, critical shear and effective hydraulic conductivity were computed using WEPP default equations while the initial saturation level was assumed at 70% as recommended in WEPP (USDA- ARS, 1995). To create a soil map, the Soil Series classification was used as the mapping unit. The procedure by Minkowski (2005) was followed in creating the soilsmap.ascii and soilsmap.txt files.

3.3.1.4 Crop management file and land cover map The crop management file consists of a management editor, cropland initial condition database and plant database. Under the management editor, the type of operation and date for each operation were specified. The cropland initial conditions set the conditions that exist on January first of the modeling period. The plant database stores the parameters required for the plant growth model. The land cover information of a March 2002 image, from Landsat-7 ETM+ acquired on 29 March 2002, was assumed as similar to the current land cover and was used in this study. The Landsat-7 ETM+ image was procured from Geoimage Company. The image was classified using NDVI (normalized difference vegetative index) by the BSWM GIS Group. There were eight land cover types identified: forestry, grassland, ricefields, agricultural areas, shrubland, bare soil, built-up areas, and water (reservoir). Most of the crop management input data required by WEPP was not available from the study area. In this case, the crop management files were taken from WEPP and GeoWEPP databases, that accompany with the models, for crops that were more or less similar in nature to those in the study area. The land cover types and their crop management files are presented in Table 3.2. In the case of ricefields, the forest crop management file was selected. This was based on the recommendations of a number of studies, which pointed out that paddy fields were a way of restoring erosion and sedimentation functions of forest (i.e. Agus et al., 2004). Additionally, studies using the GLEAMS-PADDY model

37 (Chung et al., 2003) ignored erosion and sedimentation in paddy fields on the argument that since paddy plots were surrounded by berms, sediment transport was much smaller than that of upland fields.

Table 3.2. Crop parameters and management files taken from WEPP and GeoWEPP databases (USDA-ARS, 1995) and used in simulations. Land cover type Assumed equivalent crop management file Agricultural areas WEPP/Agriculture/corn-fall-moldboard Bare soil GeoWEPP/18%cover short grass prairie Built-up areas GeoWEPP/fallow* Forestry GeoWEPP/tree-20 yr old forest Grassland GeoWEPP/grass Ricefields GeoWEPP/tree-20 yr old forest Shrubland GeoWEPP/Mountain Big Sagebrush *Several parameters were modified to represent no cropping and compacted soils in order to be consistent with the rural built- up areas of Bohol.

To create a landcover map, the land cover of March 2002 was used and converted to ASCII format. A landuse text file describing the values of the land cover map was created. Procedures of creating the landcov.asc and landuse.txt files are given by Minkowski (2005).

3.3.2 Scenarios A range of land use scenarios was created and simulated using the WEPP model. The purpose of the simulations was to determine the effect of land use and management practices in terms of soil loss, runoff and sediment yield. Simulations were carried out at hillslope and watershed levels.

3.3.2.1 Hillslope application The hillslope application of the WEPP model was applied in simulating the effect of terracing and the use of grass strips on sloping land planted to corn. Simulations were carried out using the conditions set in the Table 3.3. The slope and slope length selection was based on the erosion study by Presbitero (2003) on steep slopes of humid-tropic in the Philippines. The slope length was set at 12-m. The climate file used was taken from Bugsok AWeS 2005 records. Soil type was assumed to be of Ubay Series, which was the dominant soil series in the study area.

38 The continuous corn-fall-moldboard management file and grass management file from WEPP and GeoWEPP databases were used. Simulation was run for a year. Results were compared with the non-terraced, terraced and the use of grass strips in terms of soil loss, runoff and sediment yield.

Table 3.3. Scenarios for the hillslope application of WEPP on steep slopes cropped with corn. Terrace/grass strip Number of terraces/grass strips Slope conditions (%) width (m) across the slope 10 1.0 0,1,2 50 1.0 0,1,2 60 1.0 0,1,2 70 1.0 0,1,2

Terracing In this simulation, the hillslope topography was altered to effect terracing. There were two sub-scenarios assessed. One scenario was when a 1-m terrace was located at the bottom of the hillslope while the other scenario was when there were two 1-m terraces, one at the bottom and the other at the middle of the hillslope. The terrace was set at 1% slope and was cropped with corn. The hillslope condition is illustrated in Figure 3.5.

a b c

Figure 3.5. Modification of hillslope topography to effect terracing: (a) no terrace (b) one terrace at the bottom of hillslope, and (c) two terraces. Slope lengths are shown in Figure 3.6.

Use of grass strips on steep slopes In this simulation, the terraces, instead of being cropped to corn, were replaced with grass strips while the slopes were cropped with corn. This scenario was based on the common practice of farmers cultivating very steep lands in Bohol Island. Farmers used the natural vegetative strips (NVS) as a soil conservation measure on sloping cropland.

Top 5 m Corn crops

11 m 1 m Corn crops Corn crops Grass strips 12 m

Corn crops 5 m

Bottom 1 m 1 m Grass strips Grass strips

1OFE 2OFE 4OFE

Figure 3.6. Schematic diagram of overland flow elements for the grass strips simulation

In the WEPP model, simulation of different crop management on a hillslope is considered as a multiple flow simulation. Multiple flow simulation involves more than one overland flow element (OFE). An OFE represents homogeneous crop management and soil conditions. Here in this simulation, the hillslope was divided into overland flow elements (OFE) where each element had a homogenous crop condition, as shown in Figure 3.4.

3.3.2.2 Watershed application The watershed simulations were carried out to assess the spatial variation of soil loss, sediment yield and runoff over a catchment. The two methods in the watershed application were used: one is the hillslope method and the other one is the flowpath method. Since the current version of GeoWEPP is limited to 1000 hillslopes, a smaller catchment area was selected. In this case, the catchment area of Bugsok AWaS was selected. Scenarios simulated are presented in Table 3.4. The

40 Bugsok AWeS 2005 climate file was used. The soil type and topographic parameters were based on the soil map and DEM.

Table 3.4. Land use scenarios for WEPP watershed application using the Bugsok AWaS catchment area. Scenario Description A Existing land cover (March 2002) B Slope >18% - all forested, while remaining areas are the same as Scenario A C Slope 0 – 18% cropped with corn, remaining areas are the same as Scenario A D Combination of B and C E All cropped with corn F All forest

Due to unavailability of current land cover information, the land cover data of March 2002 was assumed to be similar to the current land cover of the selected catchment and is named as Scenario A. This land cover data was used as the basis for comparison of the other scenarios. The creation of land cover maps for the scenarios was done ArcGIS 9.1 following the conditions given in Table 3.4.

Off-site effects The off-site outputs of WEPP model were presented as sediment yield and runoff discharge from hillslopes and channels. In the simulations carried out, erosion from the channels were not considered since existing channels in the study area were perrenial streams and were not represented in the WEPP channel erosion component. Off-site effects of erosion in terms of sediment yield and discharge runoff were computed as the sum from all the hillslopes.

On-site effects Results of the WEPP simulation using the flowpath method represent the on- site effects of erosion over a catchment. A soil loss tolerance value of 10 t·ha-1·yr-1 was used for mapping soil loss over the watershed. The value is within the permissible range used given in Morgan (1995). Soil loss tolerance is defined as the maximum permissible rate of erosion at which soil fertility can be maintained over 20-25 years (Morgan, 1995).

CHAPTER 4 RESULTS AND DISCUSSION

This chapter presents the results of on-site monitoring of soil loss, runoff and sediment yield, rainfall and stream discharges from the Watershed Project and the computer simulations carried out in the study.

4.1 On-site measurements The on-site data collection ran from March 2004 to February 2006. Rainfall from localized rain gauges, and runoff and soil loss data from the erosion plots were recorded on a weekly basis. These data are presented in Appendix Tables 4.1-4.5. Records were collected from the automatic weather stations (AWeS) and automatic water sampler (AWaS). However, due to instrument problem, there were periods with no data.

4.1.1 Experimental runoff plots and rain gauges

4.1.1.1 Weekly rainfall from the local rain gauges Figure 4.1 shows the weekly rainfall from the three local rain gauges. The three rain gauges represented rainfall from the monitored land-use plots: agroforest, cassava/corn, forest, grassland, and oil palm land usage. From Figure 4.1, variation of weekly rainfall amount from the three rain gauges was evident. The variation of rainfall from the three rain gauges was due to their spatial location. The highest weekly rainfall recorded at agroforest and forest, cassava/corn and grassland, and oil palm rain gauge stations was 355 mm (Week 78), 461 mm (Week 91) and 703 mm (Week 96), respectively. During the 98 weeks of the data collection there were 66, 74, and 86 weekly rainfall records from the oil palm, agroforest and forest, and cassava/corn and grassland rain gauges, respectively. The weekly rainfall data from the local rain gauges were used to correlate with the weekly runoff and soil loss from the plots where these rain gauges were located.

42 Agroforest-Forest

700

600

500

400

300

200

10 0

0 12 3456 789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98

Cassava/ corn

700

600

500

400

300

200

10 0

0 1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98

Oil Palm

700

600

500

400

300

200

10 0

Figure 4.1. Weekly rainfall from the three localized rain gauges during the 98-week data collection.

43 4.1.1.2 Weekly rainfall > 60 mm Weeks with high erosion from all the plots are few in number. This is because extremely high rainfall events causing high erosion were found to be rare during the study period. Similar findings as in the study of Poudel et al., (1999) where only three rain events contributed 47% of total soil loss. Table 4.1 shows the contribution of soil loss produced from weekly rainfall above 60 mm to the total soil loss during the data collection. About 95% of the total soil loss from the agroforest and cassava/corn plots was generated during the weeks with weekly rainfall above 60 mm. In the grassland, oil palm and forest plots, 89.6%, 82% and 81.4% of their total soil loss was monitored during the 60-mm weekly rainfall events. The 60-mm weekly rainfall periods represented 16-23% of the whole period of data collection. Data presented in Table 4.1 suggested that weekly rainfall below 60-mm do not cause much erosion. However, in order to understand erosion processes under each type of land use, the following sections present the weekly rainfall, runoff and soil loss for each plot.

Table 4.1 Percentage contribution of soil loss above 60-mm weekly rainfall to total soil loss from each experimental runoff plot. Number of weeks with rain % contribution to total Plot >60mm soil loss Agroforest 17 95.4 Cassava/corn 23 95.0 Forest 17 81.4 Grassland 23 89.6 Oil palm 16 82.0

4.1.1.3 Weekly rainfall-runoff from the experimental runoff plots Figure 4.2 shows the pattern of weekly rainfall and runoff from the five experimental runoff plots. The maximum weekly runoff from each plot varied widely. The maximum weekly runoff is in the order of 344 mm, 296 mm, 149 mm, 68.4 mm and 9.5 mm for cassava/corn, grassland, oil palm, agroforest and forest plots, respectively. Weekly runoff from the forest plot remained at a very low level over higher weekly rainfall periods, except for one particular week. That particular week (Week 18) produced the highest runoff as presented in the previous section. The weekly runoff in the oil palm plot was erratic compared to the weekly runoff in the cassava/corn and grassland plots where a trend of increasing runoff with increasing rainfall was observed.

Agroforest Agroforest 160 0.5

140 0.4 120

100 0.3 80

60 0.2 Runoff (mm)Runoff (mm)Runoff

40 0.1 20

0 0.0 0 100 200 300 400 0 102030405060 Rainfall (mm) Rainfall (mm)

Forest Forest 10 0.5 9 8 0.4 7 6 0.3 5 4 0.2 Runoff (mm) Runoff (mm) 3 2 0.1 1 0 0.0 0 100 200 300 400 0 102030405060 Rainfall (mm) Rainfall (mm)

Cassava/corn Cassava/corn 400 0.5

350 0.4 300

250 0.3 200

150 0.2 Runoff (mm) Runoff (mm) 100 0.1 50

0 0.0 0 100 200 300 400 500 0 102030405060 Rainfall (mm) Rainf all (mm)

Figure 4.2. Weekly rainfall and runoff from the agroforest, cassava/corn, forest, grassland and oil palm experimental runoff plots. Right-hand charts show data points for rainfall below 60 mm and runoff below 0.5 mm.

Grassland Grassland 350 0.5

300 0.4 250

200 0.3

150 0.2 Runoff (mm) Runoff (mm) 100 0.1 50

0 0.0 0 100 200 300 400 500 0 102030405060 Rainfall (mm) Rainfall (mm)

Oil palm Oil palm 70 0.5

60 0.4 50

40 0.3

30 0.2 Runoff (mm) Runoff (mm) 20 0.1 10

0 0.0 0 100 200 300 400 500 600 700 0 102030405060 Rainf all (mm) Rainf all (mm)

Continuation of Figure 4.2

From Figure 4.2, most of the weekly rainfall data occurred below 100 mm. When the data were analyzed, data points concentrated below 60 mm of rainfall and 0.5 mm of runoff. These data points were magnified and are presented by similar charts on the right-hand side in Figure 4.3. The right-hand chart of the cassava/corn plot shows that at 10-mm rainfall, runoff was already higher compared to the runoff from other land use plots at this rainfall amount. Also for the cassava/corn plot, data points beyond 40-mm have runoff greater than 0.5 mm so are not shown in the magnified chart.

46 Under these low rainfall events, runoff from agroforest site was smaller than in forest plot. This was probably due to the presence of terraces in the agroforest plot. From the five land use plots, runoff increased with rainfall but the increase was much higher in the cassava/corn plot.

4.1.1.4 Weekly rainfall-soil loss from the experimental runoff plots Figure 4.3 shows the weekly rainfall and soil loss from the five experimental runoff plots (three replicate plots for each land use). Soil loss from the forest plot was determined to be the lowest of all the land uses. Soil loss in the agroforest and grassland plots increased after 100 mm of rainfall. The maximum weekly soil loss is in the order of 0.36 t·ha-1, 0.92 t·ha-1, 3.75 t·ha-1, 4.94 t·ha-1and 21.00 t·ha-1 from the forest, agroforest, grassland, oil palm and cassava/corn plots, respectively. From Figure 4.3, most of the data points can be found at weekly rainfall levels below 100 mm. These data points were magnified and were found to be below 60-mm rainfall and 0.005 t·ha-1of soil losses. The right-hand side charts of Figure 4.3 show a clear pattern of rainfall and soil loss. Soil losses from the forest plot were higher compared to soil losses from the agroforest plot. In the cassava/corn plot, soil loss was already higher at 10 mm rainfall compared to the soil loss of the rest of the plots. The high soil loss in the cassava/corn plot was probably due to cultivation and poor canopy cover. In Indonesia, the clean-weeded system of cassava cropping induces erosion and the elevated canopy of cassava crops creates throughfall drops that are more erosive than incident rainfall at low rainfall intensities (Van Dijk and Bruijnzeel, 2004). Data points below weekly rainfall of 60 mm, runoff of 0.5 mm and soil loss of 0.005 t·ha-1covered 53 weeks and 57 weeks for the agroforest and forest plots, respectively. There were 4 data points from the agroforest plot with rainfall below 60-mm that were not included because the runoff values were very high (considered to be unreliable outliers) compared to the rest of the data points. Although runoff and soil losses below 60-mm rainfall were very small, there was a different trend in the agroforest and forest plots, that was revealed: that, runoff and soil loss from forest plot were higher compared to runoff and soil loss from the agroforest plot, as shown in right-hand side charts of Figure 4.2 and Figure 4.3. For the given rainfall, runoff and soil loss limits, the total runoff and soil loss from the agroforest was 1.23 mm and 0.0076 t·ha-1 while in the forest plot the total runoff and

47 Agroforest Agroforest 1.0 0.005 0.9 0.8 0.004 )

0.7 ) -1 -1 0.6 0.003 0.5 0.4 0.002 Soil loss (tha 0.3 Soil loss (tha 0.2 0.001 0.1 0.0 0.000 0 100 200 300 400 0 102030405060 Rainf all ( mm) Rainf all ( mm)

Cassava/corn Cassava/corn 25 0.005

20 0.004 ) ) -1 -1 15 0.003

10 0.002 Soil loss (t ha Soil loss (t ha

5 0.001

0 0.000 0 100 200 300 400 500 0 102030405060 Rainfall (mm) Rainfall (mm)

Forest Forest 0.40 0.005

0.004 0.30 ) ) -1 -1 0.003 0.20 0.002 il lossil (t ha lossil (t ha So So 0.10 0.001

0.00 0.000 0 100 200 300 400 0 102030405060 Rainfall (mm) Rainfall (mm)

Figure 4.3. Comparison of weekly rainfall and soil loss from the agroforest, cassava/corn, forest, grassland and oil palm plots. Right-hand chart shows data points at weekly rainfall below 60 mm and weekly soil loss below 0.005 t·ha-1.

48 Grassland Grassland 4.0 0.005

3.5 0.004 3.0 ) ) -1 -1 2.5 0.003 2.0

1.5 0.002 Soilloss (t ha Soilloss (t ha 1.0 0.001 0.5

0.0 0.000 0 100 200 300 400 500 0 102030405060 Rainfall (mm) Rainfall (mm)

Oil Palm Oil Palm 1.4 0.005

1.2 0.004 1.0 ) ) -1 -1 0.8 0.003

0.6 0.002 Soil loss (t ha 0.4 Soil loss (t ha 0.001 0.2

0.0 0.000 0 100 200 300 400 500 600 700 0 102030405060 Rainf all ( mm) Rainf all ( mm) continuation of Figure 4.3 soil loss was 1.34 mm and 0.0381 t·ha-1. The presence of terraces is suggested to have reduced runoff and soil loss in the agroforest plot for weekly rainfall beloww 60 mm. Van Dijk and Bruijnzeel (2004) indicated that as the rate of erosion decreased, the percentage of fine particles in runoff increased. Fine particles will not settle easily however, level terrace beds (Renard et al., 1996) provide enough time for fine sediments to settle down.

4.1.1.5 Total rainfall, runoff and soil loss The accumulated 98-week values of rainfall, runoff and soil loss are presented in Figure 4.4. From Figure 4.4, the highest values for runoff and soil loss were recorded from the cassava/corn plot, while the lowest values of these parameters were recorded from the forest plot. The total rainfall measured from the three localized rain gauges differed, ranging from 3850 to 5044 mm. The highest rainfall was measured in the oil palm area.

49 From Figure 4.4, the runoff per unit rainfall (1 unit rainfall = 1 mm) from cassava/corn, grassland, agroforest, oil palm and forest plots is in the order of 29%, 20.8%, 18%, 5.2% and 0.3%, respectively. In terms of soil loss per mm of runoff, the cassava/corn plot generated 0.0609 t·ha-1·mm-1 followed by forest (0.0284 t·ha- 1·mm-1), oil palm (0.0186 t·ha-1·mm-1), grassland (0.0127 t·ha-1·mm-1) and agroforest (0.0062 t·ha-1·mm-1). The high soil loss per unit runoff in cassava/corn plot is suggested to be linked to cultivation and cropping activities during the data collection. Additionally, at different periods during the data collection period, soil was bare and exposed to the detaching impact of rainfall. The agroforest plot has the lowest amount of sediment per unit runoff. The low concentration of sediment in the agroforest plot was most likely due to flat terraces allowing sediments to settle down after transit within the plot.

6000 90

80 5000 70

4000 60 ) -1 50 3000 40

2000 30 (tloss ha Soil Runoff and rainfall (mm) rainfall and Runoff

20 1000 10

0 0 Agro-forest Cassava/Corn Forest Grassland Oil Palm

Rainfall 3850 4515 3850 4515 5044 Runoff 694 1311 13 952 265 Soil loss 4.3 79.9 0.4 12.1 4.9

Figure 4.4. Total rainfall, runoff and soil loss accumulated during the 98-week on-site monitoring. The values were computed from weekly data.

During the 98 weeks or 1.9 years of data collection, the highest soil loss rate was 42.5 t·ha-1·yr-1 (computed as 79.9 t·ha-1 divided by 1.9 years) from cassava/corn plot. The rest of the plots showed erosion rates below the limit of 10 t·ha-1·yr-1: the erosion rates were 6.4 t·ha-1·yr-1, 2.6 t·ha-1·yr-1, 2.3 t·ha-1·yr-1, and 0.2 t·ha-1·yr-1 for grassland, oil palm, agroforest and forest, respectively. The cassava/corn plot generated more than 6 times the erosion of the next higher land cover type, which

50 was grassland. The cassava/corn plot was cultivated during the data collection and this is suggested as main reason the rate of soil erosion was relatively high in cassava/corn plot compared to other plots. Similar very high erosion rates were reported for various cultivated crops in the Philippines, particularly when cultivation occurred up and down the slope (Paningbatan et al., 1995; Craswell et al., 1998; Poudel et al., 1999). In a cassava monoculture study in Sri Racha, Thailand, Putthacharoen et al. (1998) associated high erosion rates to wide spacing and slow canopy development. The erosion rates presented above indicated rates over the period of data collection and do not necessarily represent the long-term erosion rates and are likely to explain erosion rates from different types of vegetation and land usage.

4.1.1.6 High erosion week From the weekly data provided by the Watershed Project, there was a particular week noted for its substantial contribution to the total soil loss from each respective plot, as presented in Table 4.2.

Table 4.2. Week with highest soil loss contribution to the total soil loss collected within 98 weeks including the respective runoff and rainfall amounts. Soil % of % of % of Runoff loss total soil Runoff total Rainfall total /rainfall Week/ (t·ha-1) loss (mm) runoff (mm) rainfall (%) Plots

(A) (B) (C) (D) (E) (F) (G) Week 12 Oil palm 1.27 25.9 68.4 25.8 163 3.2 42.0 Week 18 Agroforest 0.92 21.4 149.0 21.5 263 6.8 56.7 Forest 0.27 75.0 9.5 74.2 263 6.8 3.6 Week 91 Cassava/corn 20.97 26.2 344.1 26.2 461 10.2 74.6 Grassland 3.75 31.0 296.0 31.1 461 10.2 64.2 Note: Values for Column A, C and E were taken from Appendix Tables C 1-5. Column B is computed as Column A divided by the total soil loss from the corresponding plot. See Figure 4.4 for total soil loss. Column D is computed as Column C divided by the total runoff from the corresponding plot. See Figure 4.4 for total runoff. Column G is computed as column C divided by column E multiplied by 100.

From Table 4.2, most notable was the percentage contribution of soil loss on Week 18 to the total soil loss under the forest plot. Of the total soil loss under the forest plot, 75% was generated in Week 18. Notice also that only 3.6% of rainfall was runoff under this plot for this particular week compared to the other plots where

51 more than 40% of rainfall was runoff. Although erosion rates under the forest plot were very small compared to the erosion rates of other plots, Table 4.2 suggested that there may have been isolated rainfall events that caused high erosion rates for this type of vegetation. Such events most probably have happened within Week 18. Highest soil loss was also noted in the agroforest plot during that week. In Week 91, three-fourths of the total rainfall from the cassava/corn plot was translated into runoff causing very high soil loss of 20.97 t·ha-1. This amount of soil loss was 26.2% of the total soil loss from the cassava/corn plot. High erosion was also recorded in the grassland areas where 31% of the total soil loss was generated in Week 91. Apart from the soil cover factor in the cassava/corn plot, high erosion most likely was also due to the rainfall characteristics during the week. In the case of the oil palm plot, a quarter of the total soil loss was generated during Week 12. Rainfall for this week was only 3.2% of the total rainfall but the runoff from this rainfall was 25.8% of the total runoff from this plot. Similar to the cassava/corn plot, the high runoff and soil loss in the oil palm plot are hypothesized to be due to soil cover conditions and rainfall characteristics.

4.1.2 Rainfall erosivity analysis The 5-min rainfall data from the weather stations were used to compute daily and then weekly erosivity values. The days added to obtain the weekly values were based on dates from soil loss data. This was done in order to relate weekly soil loss and rainfall erosivity. Data from Bugsok AWeS has been applied to agroforest, forest and oil palm plots while Pamacslan AWeS data was applied to cassava/corn and grassland plots. During the 98-week period, there were 44 weeks in Bugsok AWeS and 27 weeks in Pamacsalan AWeS with erosive rainfall. The weekly rainfall erosivity was computed as the summation of all the erosive rainfall during the week. Erosive rainfall was considered to have occurred in events having component intensity greater than 25 mm·hr-1, as indicated by the experiments of Hudson (1965) (cited in Hudson, 1995). A relationship was established between weekly total rain and weekly erosivity. Figures 4.5 and 4.6 show the rainfall-erosivity relationship from the two AWeS and the regression equation representing the relationship.

Bugsok AWeS 1800 1600 1400 ) -2 1200 1000 800 600 Erosivity (J m 400 y = -0.014x2 + 13.145x - 25.853 2 200 R = 0.671 0 0 20406080100120140 Rainfall (mm)

Figure 4.5. Weekly rainfall and erosivity relationship computed from the 5-min data of Bugsok AWeS.

Pamacsalan AWeS 3000 y = 4.3739x1.1484 2500 R2 = 0.5483 ) -2 2000

1500

1000 Erosivity (J m 500

0 0 20 40 60 80 100 120 140 160 180 Rainfall (mm)

Figure 4.6. Weekly rainfall and erosivity relationship computed from the 5-min rainfall data Pamacsalan AWeS

A second-degree polynomial line represented the erosivity-rainfall relationship in Bugsok AWeS while a power function provided as best fit for the Pamacsalan AWeS data. The relationships were used to compute the erosivity values of rainfall measured by the rain gauges. Rainfall erosivity and soil loss is discussed in the following section.

53 4.1.2.1 Rainfall erosivity and soil loss The weekly rainfall - erosivity relationships derived using the data from the weather stations were used to compute the erosivity of weekly rainfall data measured by the local rain gauges. The relationship derived using the Bugsok AWeS data was used to compute weekly rainfall erosivity in the agroforest, forest and oil palm plots while the relationship established using the Pamacsalan AWeS data was applied to cassava/corn and grassland plots. The computed weekly rainfall erosivity was plotted against the weekly soil loss for each of the experimental runoff plots, as shown in Figure 4.7. From Figure 4.7, soil losses from the forest plot were at very low values even at higher erosivity compared to the rest of the plots. In general, the erosivity - soil loss patterns in Figure 4.7 were similar to the rainfall-soil loss patterns in Figure 4.3. The highest weekly erosivity from the three localized rain gauges varied: 5015 J·m-2 (Week 91), 3022 J·m-2 (Week 92) and 2877 J·m-2 (Week 78) in the cassava/corn and grassland rain gauge, oil palm rain gauge, and agroforest and forest rain gauge, respectively. Except for the cassava/corn and grassland areas, the most erosive week for the agroforest, forest and oil palm areas did not coincided with the highest soil loss week. In the agroforest and forest areas, the week with the highest soil loss had an erosivity value of 2,465 J·m-2. In the oil palm area, the week with the highest soil loss had an erosivity value of 1,743 J·m-2 which was about a third of the erosivity value of the most erosive week for the area. In the oil palm plot, the most erosive week did not generate the highest soil loss. Runoff during this week was only 19.6 mm from a rain of 390.9 mm or 5% of the weekly rainfall. In Week 12 where the highest soil loss occurred, runoff was 68.4 mm out of 162.8 mm rain or 42% of the total weekly rain. These differences can be attributed to the cover conditions during these periods. The canopy of the oil palm trees had increased from Week 12 to Week 91. The low soil loss over high rainfall erosivity in the oil palm plot was hypothesized to be due to increase in canopy cover of the oil palm trees. In the agroforest and forest plots, the erosivity of the week with the highest soil loss did not differ much from the most erosive week. In the cassava/corn plot, the very high rainfall erosivity in combination with poor cover conditions contributed to higher runoff and soil loss. The grassland plot also generated the highest soil under the same erosivity value.

1.00 0.30 0.90 Agroforest Forest 0.80 0.25 ) ) -1

0.70 -1 0.20 0.60 0.50 0.15 0.40 0.30 0.10 Soilloss (t ha 0.20 Soil loss (t ha 0.05 0.10 0.00 0.00 0 1,000 2,000 3,000 4,000 0 1,000 2,000 3,000 4,000 Erosivity (J m-2) Er os iv ity ( J m-2)

25 4.0 Cassava/corn 3.5 Grassland 20

) 3.0 ) -1 -1 2.5 15 2.0 10 1.5 Soil loss (t ha (t loss Soil

Soil loss(t ha 1.0 5 0.5

0 0.0 0 1,000 2,000 3,000 4,000 5,000 0 1,000 2,000 3,000 4,000 5,000 -2 Erosivity (J m-2) Er os iv ity ( J m )

1.4 1.2 Oil palm )

-1 1.0 0.8 0.6 0.4 Soil loss (t ha (t loss Soil 0.2 0.0 0 1,000 2,000 3,000 4,000 Erosivity (J m-2)

Figure 4.7. Weekly soil loss and rainfall erosivity for agroforest, cassava/corn, forest, grassland and oil palm plots.

The weekly scale of rainfall and erosivity values resulted in a low correlation between rainfall and erosivity. A finer scale such as daily values (used in Davison et al., 2005; Yu and Rosewell, 1996) and longer data records could possible increase the correlation between rainfall and erosivity in the Upper Inabanga Watershed.

55 The experiments of Hudson (1965) as cited in Hudson (1995) found that there was good correlation between soil loss and rainfall erosivity. Soil loss should be directly proportional to rainfall erosivity if all other factors of the erosion process are held constant (Wischmeier and Smith, 1978). Based on the findings of other researchers noted above and the results of the on-site measurements presented here, rainfall erosivity information is important especially in agricultural areas as one of the considerations in the planning and timing of field operations. In addition, the temporal pattern of erosivity can be integrated with farm activities in order to minimize the impact of erosive rainfall. From the on-site measurement of runoff and soil loss, cultivated areas such as the cassava/corn plot generated the highest erosion rates as discussed in Section 4.1.1.5. Exposure of disturbed soil surface to the impact of rainfall was considered to be the major factor of high erosion rates from the experimental runoff plots. Erosion is a consequence of cultivation but there are cropping strategies, which may be adopted in order to minimize the rates of erosion. In Section 4.2.1, the effects on soil loss of applying a conservation measure on cultivated lands at varying slope conditions are discussed.

4.1.3 Watershed measurement of flow

Subwatershed flow monitoring was conducted to get a picture of the extent of erosion from a subwatershed as indicated in terms of runoff and sediment yield. Results of subwatershed surface flow monitoring are presented here. The location and subwatershed areas of the two monitoring sites are shown in Figure 4.8. The subwatershed areas of the Bugsok AWaS and Pamacsalan AWaS were delineated using a DEM and their respective geographic locations in ArcGIS. Pamacsalan AWaS had a subwatershed area of 2902 hectares while Bugsok AWaS had a subwatershed area of 1299 hectares.

Figure 4.8. Subwatershed areas of Bugsok AWaS and Pamacsalan AWaS delineated using a DEM and their respective coordinates.

4.1.3.1 Land cover and slope The land cover information for the two catchments was based on a classified image of March 2002. Seven land uses were identified and estimates of the area coverage by land use are presented in Table 4.3. From Table 4.3, Bugsok Subwatershed had 10% more forest area than Pamacsalan Subwatershed. There were more ricefield areas in Pamacsalan than in Bugsok. In terms of area, agricultural areas in Bugsok were twice as much as in Pamacsalan. Grassland covered 20.1% of Pamacsalan area and 15.7% of Bugsok area.

Table 4.3. Land cover distribution of the Bugsok and Pamacsalan Subwatersheds based on March 2002 Landsat-7 ETM+. Bugsok Pamacsalan Landuse Area (ha) % Area (ha) % Agricultural area 174 13.4 308 10.6 Bare soil 67 5.2 108 3.7 Forest 744 57.3 1379 47.5 Grasslands 203 15.7 583 20.1 Ricefields 10 0.7 238 8.2 Shrubland 100 7.7 276 9.5 Water - - 9 0.3 Total 1299 100 2902 100 Source: BSWM, 2005

57 Slope was derived from a 30-m DEM and classified according to BSWM classification. The slope distributions of the two catchments are presented in Table 4.4. Steeplands (>18% slope) account for 39% of Bugsok area and 42% of Pamacsalan area. In terms of hectarage, the undulating to rolling areas in Bugsok was only a third of the hectarage in Pamacsalan. Level to rolling areas covered 61% of Bugsok and 57% of Pamacsalan subwatersheds.

Table 4.4. Slope distribution within Bugsok and Pamacsalan Subwatersheds derived using ArcGIS 9 from a 30-m DEM and classified according to BSWM classification criteria. Bugsok AWaS Pamacsalan AWaS Slope Slope Classification Percent Percent Range (%) Area (ha) Area (ha) (%) (%) A (level to gently 0 - 3 118 9 233 8 sloping) B (Gently sloping to 3 - 8 321 25 409 14 undulating) C (Undulating to 8 - 18 346 27 1028 35 rolling) D (Rolling to 18 - 30 277 21 798 27 moderately steep) E (Steep to very steep) 30 - 50 211 16 326 11 F (Very steep) >50 26 2 106 4 Total 1299 100 2902 100

4.1.3.2 Rainfall and discharge curves The daily rainfall-discharge curves for the Bugsok and the Pamacsalan Subwatersheds are presented in Figure 4.9. The discharge data were taken from the automatic water samplers situated at the outlet point of the delineated subwatersheds. Missing discharge data were indicated by breaks in the discharge curves. On the rainfall axis, breaks in the curve indicate either no rainfall or no record at all. Due to considerable missing data in rainfall and discharge, periods where both rainfall and discharge were available from two monitoring sites were selected for presentation and analysis.

58 Bugsok 0 6.5 20 5.5 Rainfall 40 /s) 3 Discharge 60 4.5 80 3.5 100 120

2.5 140 Rainfall (mm)

Discharge (m 160 1.5 180 0.5 200 31 31 May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

2004 Day s 2005

Pamacsalan 0 14 20 12 40 /s)

3 Rainfall 10 60 Discharge 80 8 100 6 120 140 4 Rainfall (mm)

Discharge (m 160 2 180 0 200 1 31 31 May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May June July Aug Sep Oct Nov Dec

2004 2005 Date

Figure 4.9. Average daily discharge records from automatic water samplers and daily rainfall data from weather stations at Bugsok and Pamacsalan Subwatersheds.

59 Rainfall and discharge records for both sites in December 2005 were complete for the whole month. The events were extracted for further examination. Figure 4.10 shows the average daily discharge and rainfall for the month. As shown in this figure, the response of the Bugsok Subwatershed to rainfall events from December 9-15 is different from that of the Pamacsalan Subwatershed. In Bugsok, the peak discharge was higher and the recession phase was sharper compared to Pamacsalan. The sharper hydrograph response noted for the Bugsok Subwatershed is considered to reflect greater areas of agricultural activity (corn and cassava) and bare soil areas, as compared with the Pamacsalan Subwatershed. In the Pamacsalan Subwatershed, the discharge peak was lower but there was an extended period of almost constant discharge. The minimum and maximum discharges in Pamacsalan are lower compared to Bugsok. The total monthly rainfall measured from nearby weather stations was similar: 512 mm in Bugsok AWeS and 520 mm in Pamacsalan AWeS. An estimate of runoff from both subwatersheds was computed by assuming the lowest discharge as the baseflow. Runoff was calculated as the difference between the discharge and the baseflow. For the month of December, the total runoff from the Bugsok Subwatershed was estimated at 2.8 million m3 and 2.5 million m3 from the Pamacsalan Subwatershed. In terms of runoff depth (mm) over the subwatershed, there was 170 mm of runoff from the Bugsok Subwatershed and 85 mm from the Pamacsalan Subwatershed for the month of December. This means that 33.2% of rainfall in Bugsok was runoff while only 16.3% of rainfall was runoff in Pamacsalan. The low runoff for the Pamacsalan monitoring station is hypothesized to be due to numerous small water impoundments in the upper Pamacsalan Subwatershed area while these do not exist within the Bugsok Subwatershed. This configuration in Pamacsalan Subwatershed would tend to attenuate the hydrograph as represented in Figure 4.10.

Bugsok AWaS December 2005 8 100 rain 90 7 discharge 80 6 70

/s) 5 3 60

4 50

3 (mm) Ranfall

Discharge (m Discharge 30 2 20 1 10

0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Days

Pamacsalan AWaS December 2005 8 100

rain 90 7 discharge 80 6 70 /s)

3 5 60

4 50

3 (mm) Rainfall Discharge (m Discharge 30 2 20 1 10

0 0 0 1 2 3 4 5 6 7 8 9 10111213141516171819202122232425262728293031 Days

Figure 4.10. Rainfall-discharge curves on December 2005 from Bugsok and Pamacsalan monitoring sites.

4.1.3.3 Suspended sediment concentration Two flood events with corresponding sediment yield values at points within the rise and recession limbs of the discharge curve were selected from Bugsok and Pamacsalan records. These events were on different dates so direct comparison between the two is not possible. However, they are presented here in order to gain an understanding of the relationship of sediment concentration and discharge rates under each subwatershed. The flood event on July 22, 2004 in Bugsok AWaS is shown in Figure 4.11. Bugsok AWeS recorded a total of 8.7 mm of rainfall on this day. The increase in

61 discharge cannot be totally accounted for by this rainfall but it is suggested that the discharge reflects heavy rainfall within the four days (from the 18th of July) prior to the sharp increase in discharge and other rainfall events at the upper portion of the subwatershed. Unfortunately there was no discharge record on the 18th of July to indicate changes in the discharge curve over that period. However, the following days showed lower average discharges compared to the discharge on the 22nd. As expected, sediment concentration reached its peak earlier than the peak discharge. Sediment concentration ranged from 21 to 201 mg·L-1. This high sediment concentration is suggested to reflect agricultural activity inside the subwatershed.

8 2 8.0 350 Rainfall 7 7.0 Discharge

300 )

Discharge -1 TSS 6 ) 6.0 )

-1 250 -1 s s 3 3 5 5.0 200 4 1 4.0 150 3 3.0 Rainfall (mm)

Discharge (m 100 Discharge (m Discharge 2 2.0

50 Suspended (mgL Solids 1 1.0

0 0 0.0 0 0 100 200 300 0 100 200 300 Time elapsed (x 5 mins) Time elapsed (x 5 mins)

Figure 4.11. Rainfall, discharge and sediment yield from Bugsok AWaS recorded on July 22, 2004.

Figure 4.12 shows a flood event on July 15-16, 2004 recorded in Pamacsalan AWaS. There were two rainfall events on the 15th of July. The first rainfall was 11.1 mm followed by a 7.8 mm rain event 110 minutes later. Prior to the increase in discharge, there was a rainfall event with 45.5mm rainfall depth on the 13th day of July. Based on the discharge records, there was no other increase in discharge following that event except on the 15th of July. Hence, the increase in discharge can be attributed to the previous events especially the one that occurred on 13th of July. In addition to the rainfall event on the 15th of July, there may have been other rainfall events in the upper part of the watershed that contributed to the discharges as shown by the two discharge peaks.

62 9 2.5 9 900

8 8 Discharge 800 ) Rainfall -1 2.0 ) 7 TSS 700

) 7 -1

-1 Discharge s s 3

3 6 6 600 1.5 5 5 500 4 4 400 1.0

3 Rainfall (mm) 3 300 Discharge (m Discharge (m 2 0.5 2 200

1 1 100 Suspended (mgL Solids 0 0.0 0 0 0 100 200 300 0 100 200 300 Time elapsed (x5 min) Time elapsed (x5 min)

Figure 4.12. Rainfall, discharge rates and sediment yield from Pamacsalan AWaS recorded on July 15-16, 2004.

Sediment concentration ranged from 15 – 782 mg·L-1 in Pamacsalan monitoring site. A very sharp increase in sediment concentration was observed during the rising limb of the discharge curve. During the recession phase, sediment concentration decreased quickly as also observed by Lee et al., (2006) since most of the sediments were eroded during the initial period. The relatively high sediment concentration as noted for the Pamacsalan Subwatershed may reflect a lower forest area compared to the Bugsok Subwatershed. Additionally, there was 10% more undulating to steep slope areas in the Pamacsalan Subwatershed than in Bugsok Subwatershed. Initial results of the runoff and sediment concentration analysis showed that at a subwatershed scale, runoff from Pamacsalan was lower compared to Bugsok but sediment concentration was higher in Pamacsalan than in Bugsok. The lower runoff in Pamacsalan Subwatershed is probably due to water harvesting activities in the upper areas whereas these do not exist in the Bugsok Subwatershed. The water harvesting activities are considered to reflect significantly more widespread economic activities including cultivation in the Pamacsalan Subwatershed than in the Bugsok Subwatershed. It is hypothesized that this is one of the reasons why there was found to be higher concentrations of suspended sediments in the Pamacasalan River. For the measured flood events, the average sediment concentrations from the two subwatersheds exceed the water criteria for conventional and other pollutants contributing to aesthetics and oxygen demand for Class D water usage, which is not more than 60 mg·L-1 (DAO 34, 1990).

4.1.4 Summary of the on-site measurements of erosion

The results of the on-site measurements of soil loss, runoff and sediment yield indicated that, indeed land degradation linked to land use is happening within the Upper Inabanga Watershed. In order to further evaluate and address the issue using the dataset available, the use of an erosion model was considered to be appropriate in developing planning strategies for conservation and watershed management. The following sections present the application of an erosion model (WEPP) in designing appropriate conservation strategies at a hillslope level and in assessing the potential impact of land use changes at a watershed scale.

4.2 Application of WEPP and GeoWEPP erosion models

The capabilities of the WEPP erosion model and its geo-spatial interface as tools for soil and water conservation and planning were applied in the Upper Inabanga Watershed. The WEPP hillslope application was used to simulate erosion processes and predict runoff, soil loss and sediment yield over hillslopes of varying gradients. The effects of adopting conservation strategy were also simulated. The watershed application of WEPP through its geo-spatial interface GeoWEPP was applied to assess potential erosion hazards resulting from land use changes over a watershed. On-site and off-site effects of land use changes were predicted. The following sections describe the results of the model application in the Upper Inabanga Watershed. Input data and the scenarios being modeled were described in Chapter 3 Section 3.3.2. In as many instances as possible, local datasets were used to run the model but for a number of parameters the WEPP databases were utilized for parameters that were not available, especially the crop management inputs.

4.2.1 Erosion assessment at farm level

Based on the on-site erosion monitoring conducted using the experimental plots, the cultivated areas, namely the cassava/corn plot, generated the highest soil compared with the other land uses: as much as six times the losses of the grassland areas. However, agriculture is the major source of income of the community in the

64 Upper Inabanga Watershed and corn is one of the major crops. Thus, conservation strategies for this cropping practice are critical issues for sustainable agriculture. The WEPP hillslope application was used to simulate and predict erosion over a single hillslope cropped with corn. A 12-m slope length was used and slopes varied from, 10%, 50%, 60% and 70%. The inputs were given and described in Chapter 3 Section 3.3.2. The scenarios evaluated were: a) conventional tillage – without any conservation measure, b) application of terraces, and c) use of grass strips. Results of the simulation scenarios are summarized in Table 4.5.

Table 4.5. WEPP simulation of soil, sediment and runoff for non-terraced and terraced conditions under different slopes No terrace 1 terrace 2 terraces Slope Sediment Sediment Sediment Soil loss Runoff Soil loss Runoff Soil loss Runoff (%) yield (t·ha- yield (t·ha- yield (t·ha- (kg·m-2) (mm) (kg·m-2) (mm) (kg·m-2) (mm) 1) 1) 1) 10 1.825 18.25 830.44 1.346 13.456 830.55 1.344 13.44 830.59 50 14.042 140.42 798.82 11.904 107.72 803.7 10.749 102.4 806.98 60 16.992 169.02 792.23 13.826 121.79 797.08 12.802 113.8 801.6 70 17.555 175.55 786.74 16.383 129.51 791.29 14.349 122.4 796.09

4.2.1.1 Increasing slope The simulations shown in Table 4.5 showed that soil loss increased as the slope increased while the runoff depth decreased with increasing slope. The trends on runoff agreed with the field experiments reported by Presbitero (2003) for hedged runoff plots planted with corn on 50%, 60% and 70% slopes. Additionally, the trend for soil loss showed a similar pattern to that measured for soil loss from bare plots in the above noted field experiments of Presbitero (2003).

4.2.1.2 Effects of terracing The WEPP model was run to simulate the effect of a 1 metre terrace situated at the bottom of a hillslope. From Table 4.5, soil loss and sediment yield were reduced with a terrace compared with soil loss and sediment yield on hillslopes without a terrace. The single 1 m width terrace reduced soil loss by 26%, 15%, 14% and 12% on 10%, 50%, 60% and 70% slopes, respectively relative to a no-terrace profile. Sediment yield was reduced by 23%, 24% and 26 % from 50%, 60% and 70% slopes, respectively. Runoff depth on the other hand, increased slightly with terracing.

4.2.1.3 Additional terrace When an additional 1 m terrace wide was placed at the middle of the slope, further decrease in soil loss and sediment yield was observed relative to a no-terrace condition as presented in Table 4.6. At 10% slope, the addition of terraces had no effect on soil loss, sediment yield and runoff. However for slopes at 50%, 60% and 70%, reshaping the landscape to add one more terrace conserved 18-26% of soil loss and 26-30% of sediment leaving the hillslope profile, relative to no-terrace conditions.

Table 4.6. Percent decrease in soil loss, sediment yield and runoff resulting from the use of one and two 1 m terraces relative to a no-terrace condition. 1 terrace 2 terraces Slope (%) Sediment Sediment Soil loss Runoff Soil loss Runoff yield yield 10 26 26 0 26 26 0 50 15 23 -1 23 27 -1 60 14 24 -1 20 29 -1 70 12 26 -1 18 30 -1

4.2.1.4 Grass strips Under conditions in which terraces were replaced with grass strips, simulations were run to test if the grass strips had an impact on soil loss, sediment yield and runoff. Results of the simulations are presented in Table 4.7. The simulation results showed that grass strips were effective at reducing the amount of sediment leaving the hillslope. For instance, at 10% slope, when the bottom terrace was planted with grass, sediment yield was halved: from 26 % with terrace to 55% with grass strips relative to a no conservation measure practice. On the same slope, soil loss was decreased by a further 4% percent with grass strips compared to cropping with corn. The impact of the 1 m grass strip at the bottom of the hillslope, on slopes 50- 70%, was very significant. There was a reduction of 65-66% in sediment yield compared to only 23-25% if the 1 m terrace was cropped with corn.

Table 4.7. Percent decrease in soil loss, sediment yield and runoff when terraces were replaced with grass strips relative to no-terrace conditions. 1 m width grass strips Slope (%) Soil loss Sediment yield Runoff 10 30 55 -1 50 11 66 -6 60 9 66 -7 70 8 65 -8

4.2.1.5 Soil loss graph and deposition points Soil loss graph outputs of the WEPP simulation are presented in Figure 4.13 (A and B) taken from the 50% slope hillslope situation. In Figure 4.13-A, it was noted that deposition occurred along the grass strip width. Contrary to a hillslope without the grass trip, soil loss occurred along the slope line without deposition as shown in Figure 4.13-B. In the longer term, it is suggested that, deposition points in Figure 4.13-A would become bunds and the cropped area will be leveled as more sediments were trapped at the grass strip section. These observations are similar to those of Stark et al. (2003) and have also been noted in some local cropping examples where grass strips have been used. Within two years of the Stark et al. (2003) field research monitoring, natural formation of terraces resulted in the development of almost one-meter high bunds, which further stabilized the cultivated soils.

A B

Figure 4.13. Soil loss graph of a 50% slope (A) with 1 m width grass strip at the bottom of the slope and (B) no terrace.

67 4.2.1.6 Multiple flow simulations Simulation carried out over a hillslope with one and two grass strips was considered a multiple overland flow simulation using the WEPP erosion model. This was illustrated previously in Chapter 3, Figure 3.6. In Figure 4.14, the distribution of soil loss and deposition along the slope for three scenarios is presented. This figure shows the WEPP model simulations on single and multiple flow elements over a hillslope. The points of maximum soil loss and deposition along the hillslope are presented in Table 4.8. One of the interesting points is the initial loss of 10 kg·m-2 of soil on the hillslope, which varied with conditions. In a no grass strip condition, 10 kg·m-2 detachment started at 4.68 m from the top of the hillslope. The distance decreased with the addition of the grass strips. It was noted that for the three conditions, the starting point of a 10 kg·m-2 detachment occurred on the uppermost overland flow element (OFE) with the same cropping system but different slope length (12 m for no grass strip, 11 m for 1 grass strip and 5 m for 2 grass strips). Over the entire hillslope the starting point of 10 kg·m-2 soil losses reflected the effect of multiple OFE simulation. In Figure 4.14, particularly the graph for the 2-grass strip condition, there were two points of deposition, one was on the first upper grass strip and a second was at the bottom grass strip. Prior to deposition, soil detachment rate started to decrease at a distant from the strip. Deposition rate within the width of the grass strip was shown to decrease at a very high rate as shown in Figure 4.14. A critical point after the first deposition was the point of maximum soil detachment. Soil detachment rate was higher on the second cropped OFE. However, this was compensated for by the high deposition rate of the lower grass strip. Although the two 1 m grass strips had the highest deposition values, the sediment leaving the profile was lower by 15% compared to the one 1 m grass strip as shown in Table 4.8.

68 Table 4.8. Location of starting points of deposition and soil loss from the topmost part of the hillslope as determined by WEPP simulation Distance from top of hillslope (m) Start of Point of Point of Start of Conditions 10 (kg·m- maximum soil maximum 2 deposition/amou ) soil -2 detachment/ deposition/amount nt (kg·m ) -2 -2 loss amount (kg·m ) (kg·m ) No grass strip 4.68 No deposition 12 / 28.832 No deposition One grass strip at 4.41 11.03 / 130.406 9.92 / 26.362 11.1 / 136.294 the bottom Two grass strips 3.99 6.06 / 29.867 6.53 / 68.965 11.03 / 377.948

80 Two grass strip 30

-20 024681012 ) 2 -70

-120

-170

-220

-270 Erosion/deposition (kg/m -320

-370

-420

50 One grass strip 30

10 ) 2 -10 024681012

-30

-50

-70

-90 Erosion/deposition (kg/m

-110

-130

-150

50 No grass strip 45

40 ) 2 35

Erosion/deposition (kg/m 10

0 024681Distance, 012

Figure 4.14. Trend of soil loss and deposition along a hillslope for three scenarios where the 2-grass strip condition has four OFEs; 1 grass strip condition has 2 OFEs and the no grass strip condition has one OFE as determined in WEPP simulations.

69 4.2.2. Erosion hazard assessment in the Bugsok Subwatershed

The watershed application of WEPP and its geo-spatial interface, GeoWEPP, was used to predict erosion hazards in the Bugsok Subwatershed. The scenarios simulated and the inputs used were described in Chapter 3, Section 3.3.2.2. Results of the WEPP/GeoWEPP watershed application are presented as on- site effects and off-site effects. The on-site effects are classified in terms of what were considered as tolerable and non-tolerable soil loss rates while the off-site effects were indicated by sediment yield and discharge volume at the watershed outlet. A tolerable soil loss value of 10 t·ha-1·yr-1 was used as the threshold level for evaluating the on-site effects. The threshold value is within the soil loss “tolerable” value of 2 - 11.2 t·ha-1·yr-1 which has been used by previous researchers (e.g. Renard et al., 1996; Morgan, 1995; Lal, 1994). Areas having soil loss rates exceeding the tolerable soil loss rates were considered critical. There were six scenarios simulated and were described previously in Chapter 3 Section 3.3.2.2. The on-site results of the simulations are presented in Table 4.9. Erosion maps are presented in Figure 4.15.

Table 4.9. On-site effects of land use change predicted by WEPP-GeoWEPP and presented as percentage distribution of soil loss under different land cover scenarios1 Scenarios Erosion rates A B C D E F Deposition (t·ha-1·yr-1) 10.7 10.7 13.2 13.8 6.8 3.5 > 10 6.2 5.8 10.6 10.4 6.6 0.4 <= 10 4.5 4.9 2.6 3.4 0.2 3.0 Tolerable soil loss ( 10 t·ha- ≤ 68.1 71.5 33.1 35.5 7.9 92.5 1·yr-1) 0 – 2.5 40.1 44.4 10.9 13.5 0.7 74.6 2.5 – 5.0 21.4 21.0 14.9 14.8 3.1 14.1 5.0 – 7.5 4.4 4.1 5.1 5.1 3.0 2.5 7.5 – 10 2.2 2.0 2.1 2.1 1.0 1.3 Non-tolerable soil loss (>10 21.2 17.8 53.7 50.7 85.3 4.0 t·ha-1·yr-1) 10 – 20 4.1 3.5 4.6 4.0 2.6 2.2 20 – 30 2.0 1.6 2.2 1.9 1.8 0.7 30 – 40 1.2 1.0 1.5 1.4 1.3 0.5 >40 13.9 11.7 45.4 43.4 79.6 0.6 Total 100 100 100 100 100 100

1 Scenario description is given in Table 3.4, Chapter 3

70 A B

C D

E F

Legend: Soil loss values (t·ha-1·yr-1) > 10 (Deposition) < 2.5 5.0 - 7.5 10 - 20 30 - 40 <= 10 (Deposition) 2.5 - 5.0 7.5 -10.0 20 - 30 > 40

Figure 4.15. Erosion maps of the Bugsok Subwatershed with six land use scenarios showing the on-site effects as predicted using GeoWEPP. White areas within the subwatershed are the channels identified by GeoWEPP but were excluded in erosion simulation with the flowpath method.

71 4.2.2.1 Existing land use conditions Scenario A represented the existing land cover conditions in the Bugsok Subwatershed. Land cover and slope distributions of the Bugsok Subwatershed were presented previously in Section 4.1.3.1. On-site effects, in terms of erosion rates (Table 4.9) from the model simulations were grouped into three categories: deposition, tolerable and non- tolerable. To simplify the discussion, the results were grouped into two headings: critical and non-critical zones. The critical areas are those areas with soil loss rates exceeding 10 t·ha-1·yr-1 while the non-critical areas are those areas with deposition and soil loss rates ≤ 10 t·ha-1·yr-1. The two headings are further subdivided into two subheadings: ≤ 18% slope and >18% slope. Results are given in Table 4.10. The 18% slope limit had been designated in the Forestry Code of the Philippines (Philippines, 1975) above which the land should be allocated for forestland.

Table 4.10. Simulation results of Scenario A representing the area (in percent) occupied by each land cover type classified under tolerable and non-tolerable soil loss rates and further classified using the 18% slope criteria. Non-critical Critical Land cover type Total >18% ≤18% >18% ≤18% Agriculture 0.1 2.0 1.4 9.9 13.4 Bare soil 1.6 5.0 0.9 0.1 7.6 Forest 30.3 22.8 4.0 0.1 57.2 Grassland 1.5 13.6 0.4 0.3 15.8 Ricefields 0.2 0.5 - - 0.7 Shrubland - 1.2 0.5 3.6 5.3 Subtotal 33.7 45.1 7.2 14.0 100 Total 78.8 21.2

From Table 4.3, the major land cover under the Scenario A were forest (57.2%) followed by grassland (15.8%) and agriculture (13.4%). The slope distribution for the Bugsok Subwatershed was given in Table 4.4. Under the existing land cover, 78.9 % of the area was under non-critical conditions while 21.2% of the area was under critical conditions. In the non-critical zones, although some areas experienced erosion, erosion rates were under a tolerable level threshold. Most of the areas on the steep lands (>18% slope) which showed low erosion rates were the forest areas (30.3%). There were grasslands (1.5%) and bare soil (1.6%) on steep

72 slopes with low erosion as well. The bare soil on these slopes was hypothesized to be exposed rock areas. The critical areas were distributed into 14.0% on slopes ≤ 18% while 7.2% on slopes >18%. On slopes ≤ 18%, there were 9.9% agriculture and 3.6% shrubland land cover while on slopes >18%, 4.0% were forest. The agriculture land cover type, referred to here, was assumed to be planted with continuous corn as set out in Chapter 3, Table 3.4. Under this type of land cover and even in low slope areas, agriculture activity generated non-tolerable soil loss based on Section 4.2.1. This is where the application of on-farm strategy such as terracing, discussed previously, would particularly be beneficial. On the other hand, there were forest areas on steep slopes that showed high erosion rates. These areas may benefit from consideration of structures such as flow diversion channels, dense vegetative planting strips, or other geotechnical remediation in order to minimize erosion of the slopes. However, the specific presentation of these of these conditions is beyond the scope of the present study.

4.2.2.2 Agriculture ≤ 18% slope> forest With the Philippine government’s restriction on 18% slope agricultural activity, three scenarios were simulated as follows:

• Scenario B- when all the areas >18% were all allocated to forest while the remaining areas were the same as the existing

• Scenario C- when all the areas <18% were allocated to agriculture while the remaining areas were the same as the existing; and

• Scenario D- when the entire area ≤ 18% slope is all allocated to agriculture and the area >18% slope are all allocated to forest used. Scenario B would occur when the policy on >18% slope for forestland will be enforced at the present time. Considering this scenario, there would be a decrease in the critical areas by 3.4%. This decrease in critical areas resulted from the increase in forest area by 6.6%. It should be noted that 40.9% of the area was on slopes >18% and 34.3% were already forested under the existing land cover. Scenario C would happen if slopes below 18% were all utilized for crop production in addition to the existing crop production areas on slopes over 18%. Under this scenario, it was predicted that critical areas would increase by 22.5%, relative to Scenario A. Under this scenario, the forest areas which covered 22.9%

73 would be converted to agriculture that resulted to an increase in the critical zones. In Scenario C, agricultural areas would cover 60.6% of the total area: 1.5% on >18% slopes while 59.1% on ≤ 18% slopes. Under Scenario D, forest would cover 59.1% and agriculture areas would cover 40.9% of the subwatershed. These changes would lead to half of the total area under the critical zones (Table 4.9). Scenarios B-D demonstrated that the areas on slopes ≤ 18% and under agriculture were causing high erosion rates. There were two more scenarios simulated: Scenario E where the entire subwatershed was under agriculture and Scenario F where the whole subwatershed was forested. These two extreme scenarios revealed salient features in the subwatershed. In Scenario E, 85.3% of the area were now under critical conditions while 14.7% were under a non-critical condition. Of the 14.7% experiencing tolerable erosion rates, 9.9% were on slopes ≤ 18% and 12.7% were on slopes >18%. In Scenario F, there was 4.0% of the area identified as critical. Although these areas have been identified in Scenario A, the erosion map produced in running Scenario F positioned the specific location of these critical areas, as shown in Figure 4.13-F. The off-site effect of the scenarios simulated (Table 4.11), in terms of sediment yield, was of the order of 4 t·ha-1·yr-1, 75 t·ha-1·yr-1, 82 t·ha-1·yr-1, 310 t·ha- 1·yr-1, 325 t·ha-1·yr-1 and 923 t·ha-1·yr-1 for Scenario F, Scenario B, Scenario A, Scenario D, Scenario C and Scenario E, respectively. In terms of percent critical areas, the same pattern was observed. The ideal condition was the all forested area situation. However, this scenario is considered extremely unlikely to occur. By considering the slope limit of >18% for forestland, erosion hazard areas under the present land use conditions were predicted to decrease by 3.4%.

Table 4.11. Off-site effects of land use changes predicted by WEPP-GeoWEPP model under each land use scenario Scenario Parameters A B C D E F Sediment yield (t·ha-1·yr-1) 82 75 325 310 923 4 Discharge (mm) 383 381 440 408 503 365

74 4.2.2.3 Application to land use planning Under the existing land cover conditions, only 13.4% of the Bugsok Subwatershed area is utilized for agriculture, while based on the current policy regarding the slope limitation, there is a further 59.1% that can be cultivated for agriculture use. However, even under present conditions, 21.2% of the study area is already experiencing high erosion rates. Two-thirds of these highly erosive areas are on slopes ≤ 18% dominated by agriculture. Though forest use had been identified as the most preferred land use, in terms of soil and water conservation (DENR, 2000), there are still conditions under this land cover that may result in non-tolerable erosion rates, as shown in Figure 4.15-F. In this case, specific conditions leading to high erosion rates would need to be identified in order to design conservation measures for the forest areas. In similar fashion, there are agriculture areas as shown in Figure 4.15-E where erosion rates were predicted under tolerable rates. Conditions in these areas would need to be investigated in order to identify the prevailing specific conditions, which resulted to low erosion rates. Such specific identification was beyond the scope of the current investigation but should be addressed in future studies.

CHAPTER 5 SUMMARY, CONCLUSIONS AND RECOMMENDATIONS

5.1 Summary

The research was conducted with two major objectives: a) to describe the current land use management practices in terms of runoff, soil loss and sediment yield, and b) to apply an erosion model to simulate and predict erosion in the Upper Inabanga Watershed. Satellite image was acquired and the different land cover types were identified and mapped. To meet the first objective, on-site erosion data at plot and watershed scales were used and analyzed. To meet the second objective, the WEPP erosion model and its geo-spatial interface were used.

Objective 1. To describe the current land use and management practices

Plot scale erosion data Erosion was monitored from five experimental erosion plots with agroforest, cassava/corn, forest, grassland and oil palm land cover types. These land cover types represented the major land uses within the watershed. Erosion plots (area ranging from 76.24 m2 to 151.95 m2) were set up along the slope (10-48%) and runoff and soil loss were monitored on a weekly basis. The data consisted of 98 weekly runoff and soil loss data, and were associated to the weekly rainfall from three localized rain gauges. A relationship between weekly rainfall and erosivity was established using the 5-min rainfall data from the automatic weather stations. The relationship was then used to estimate local rainfall erosivity from the rain gauges. Results of the analysis are summarized as follows:

The on-site monitoring of rainfall using three rain gauges clearly demonstrated that there was wide variation of rainfall over a relatively small spatial

76 scale within the watershed. This variation was also noted in the average annual rainfall data from a few closely spaced rain gauges used by Hoyos et al. (2005) to derive seasonal and spatial patterns of erosivity in a tropical watershed of the Columbian Andes. These variations can be explained by elevation difference, local topography and the existing environmental conditions of the rain gauges location (Hoyos et al., 2005). Fornis et al. (2005) reported wide variations in amount, duration and intensity of rainfall events in a study of daily rainfall activity on the adjacent Philippine island of Cebu.

For weekly rainfall not exceeding 60 mm, soil loss from the forest plot was higher than the soil loss from the agroforest plot. The presence of terraces in the agroforest plot was hypothesized to reduce runoff and consequently soil loss during low weekly rainfall. However, soil losses from both types of land use were still far below soil losses from the cassava/corn land use. When properly designed, terraces are effective in reducing soil erosion but could be a source of erosion when risers failed (Sidle et al., 2006).

Erosion under the major land use types demonstrated that under cultivated areas, most notably cassava/corn, erosion rates were very high (43.1 t·ha-1·yr-1), more than six times the erosion rates of the grassland areas (6.4 t·ha-1·yr-1). The high runoff and erosion from this type of land use is linked to farm management practices and poor soil cover. Cassava intercropping and rotational systems were proposed to minimize erosion based from a field study in Sumatra Island, Indonesia (Iijima et al., 2004).

The weekly scale of rainfall, runoff and soil loss associated with the experimental plots constrained the analysis that could be done. The data time frame for the plots was too long to specifically relate rainfall amount with erosion rate. However, the data indicated that few large events could cause significant erosion at a local scale.

77 Watershed scale data River discharge and sediment concentration were monitored from two river subwatersheds using automatic water samplers. The data were linked to rainfall from automatic weather stations. Due to frequent instrument failures, very few datasets were available for analysis. River discharges with corresponding rainfall data and sediment concentration were used for the analysis. Subwatershed delineation and slope classification were carried out using a 30 m DEM created for the study in ArcGIS9 platform. Land cover data was taken from a classified Landsat 7 ETM+ image. Results are summarized as follows:

The area of the Bugsok Subwatershed was less than half of the area of Pamacsalan Subwatershed. In terms of percent area, the Bugsok Subwatershed has 10% more forest areas than the Pamacsalan Subwatershed. In terms of slope distribution, the Bugsok Subwatershed had 3% less rolling to very steep areas (> 18% slope) than the Pamacsalan Subwatershed. For December 2005 total rainfall and with similar rainfall amount, 33.2% and 16.3% ended up as runoff from the Bugsok Subwatershed and Pamacsalan Subwatershed, respectively. During high rainfall periods, the watershed flow monitoring demonstrated high sediment concentrations, as high as 201 mg·L-1 in the Bugsok Subwatershed and 782 mg·L-1 in the Pamacsalan Subwatershed.

Initial results of the runoff and sediment concentration analysis showed that at a subwatershed scale, runoff from Pamacsalan was lower compared to Bugsok but sediment concentration was higher in Pamacsalan than in Bugsok. The lower runoff in Pamacsalan Subwatershed is probably due to water harvesting activities in the upper areas whereas these do not exist in the Bugsok Subwatershed. The water harvesting activities are considered to reflect significantly more widespread economic activities including cultivation in the Pamacsalan Subwatershed than in the Bugsok Subwatershed. It is hypothesized that this is one of the reasons why there was found to be higher concentrations of suspended sediments in the Pamacsalan River.

78 Objective 2. To apply the WEPP-GeoWEPP erosion model

The WEPP erosion model was applied to simulate the erosion over a hillslope with and without conservation measure. The GeoWEPP interface of WEPP was applied to predict soil loss, sediment yield and runoff from a subwatershed under different land use scenarios. Except for most of the crop management parameters, local input data were used to run the model.

Erosion assessment at farm level Erosion was predicted over a 12 metre hillslope at 10%, 50%, 60% and 70% slopes with and without terraces. Erosion was also predicted when grass strips were used as a soil and water conservation strategy. Results of the simulations showed that soil loss decreased with terracing and a further decreased was observed with the use of grass strips.

The significant effect of using terraces in decreasing soil loss can be attributed to decreased length of the hillside slope as well as decreased runoff velocity allowing sediments to settle from runoff water (Schwab et al., 1992). The use of grass strips along the terraces further reduced soil loss and this reduction can be attributed to increased in deposition along the grass strip areas. The simulations showed that in the longer term, deposition areas would become bunds and the cropped areas would be leveled as more sediments were trapped at the grass strip section. Similar observations were reported in the field experiments conducted by Thapa et al. (2000) and Stark et al. (2003) which suggested that the development of bunds could further stabilized cultivated soils. However, grass strips may also fail during concentrated flow conditions. Blanco-Canqui et al. (2004) suggested a combination of stiffed-stemmed grass above vegetative filter strips for lands affected with concentrated flows. Sediment trapping increased with filter strip width (Blanco- Canqui et al., 2004) but this has to be considered carefully since this entails reduction of cropped area that would constrain adoption by smallholder farmers (Thapa et al., 2000).

The WEPP hillslope application enabled an evaluation of conservation strategies such as terracing and use of grass strips at farm level. Simulation results

79 showed that grass strips were effective in reducing sediment yield from steep land. The use of grass strips or other soil stabilization techniques are suggested as valuable tools to minimize runoff and soil loss from common cropping activities, including cassava/corn cropping, which were demonstrated to be linked to particularly high erosion rates.

Watershed land use change and erosion assessment The geo-spatial interface of WEPP erosion model, GeoWEPP, was used to simulate land use scenarios and predict erosion under each scenario. The Bugsok Subwatershed was used as the test subwatershed. Land use scenarios were developed and input data were created in the ArcGIS9 platform. The GeoWEPP model was run in ArcView3.3. Results of the simulations are presented as follows.

Under the current land use cover, model simulation showed that 21.2% of the Bugsok Subwatershed was under critical conditions or with soil erosion exceeding the tolerable rates of 10 t·ha-1·yr-1. The critical areas were distributed as 7.2% on slopes >18% and 14.0% on slopes ≤ 18%. The agricultural areas on ≤ 18% slope generated the highest soil loss. Results from the watershed scenario simulation revealed that as the proportion of area devoted to agriculture increased, on-site erosion rates and sediment yield increased. Ella (2005), using the WEPP model, also predicted an increasing sediment yield as percent cropped area increased from Maagnao Watershed in Mindanao, Philippines. In general, any further increase in the current agricultural areas will increase erosion prone areas and sediment yield from a watershed.

However, even under the all-agriculture scenario, there were areas that experienced tolerable erosion rates (Figure 6-E). These areas need further study in order to identify the conditions that lead to low erosion. Similarly, there were areas under forestry, which showed high erosion rates. These areas should be noted so as to consider appropriate conservation strategies to minimize erosion on these areas.

The model simulations, though subject to further field verification and the availability of crop management data, enabled the evaluation of the potential impact of land use change over a watershed. GIS techniques, such as the display of colored

80 erosion maps, facilitated the interpretation of the spatial variation of soil loss across a highly variable landscape.

5.2 Overall Conclusions

The on-site monitoring of erosion and the model simulations carried out in the Upper Inabanga Watershed resulted in the following conclusions.

The current practice of cassava/corn farming generated high erosion rates compared to other land cover types. The high erosion rates were suggested to be due to tillage operations and exposure of the soil surface to erosive rain. Since agriculture, particularly growing cassava and corn is the common economic activity in the Upper Inabanga Watershed in particular and in the island in general, local authorities should enforce the adoption of conservation measures as a strategy in addressing soil and water resources sustainability. The use of conservation measures such as grass strips was predicted in this study to reduce sediment yield by as much as 65% from the current corn cropping system. The use of grass strips has been noted in some local cropping systems and was considered attractive by farmers due to its low establishment costs and its simplicity of installation (Nelson et al., 1998; Thapa et al., 2000). Increasing soil contact cover such as the addition of mulch was also found effective in reducing soil erosion (Paningbatan et al., 1995). Mulch provided protection against soil detachment by raindrop impact and soil entrainment by runoff or overland flow.

The initial analyses of river discharge and sediment concentration showed that in the Pamacsalan Subwatershed, economic activities in the upper areas reflected the quality and quantity of river water. Water harvesting was suggested to cause lower discharge in the Pamacsalan River. The water harvesting activities also reflected widespread economic activities including agriculture, as supported by the land cover data, and was hypothesized to cause higher sediment concentration of the river flows. In addition to conservation measures at the farm level, the establishment of wide buffer filter grass strips on both sides of the river could decrease sediment concentration in the river as suggested by DENR (2000).

81 The application of the WEPP erosion model and the use of GIS tools demonstrated their usefulness in the planning and managing of resources in the Upper Inabanga Watershed. The ability to simulate and predict the extent of erosion under a wide range of scenarios and to identify specific location of erosion prone areas was deemed valuable for decision-making purposes and for designing conservation strategies. The methodology that was initially developed in the study can be used in the local planning and management of soil and water resources.

5.2 Recommendations

This study led to some useful results and conclusions related to watershed management using the WEPP erosion model and GIS. However, the study also discovered some areas that require further research. The limitations of the current study and areas that need further research relative to the study are described below.

The computer simulations presented in this project were constrained with respect to a number of input parameters, particularly the crop management inputs. Results of the computer simulations could be further adapted to the local conditions by validating the model using local datasets. Thus, there is a need to collect and establish a database of local crop management parameters for use with the WEPP erosion model.

The digital elevation model (DEM) that was used in the simulation was not fully validated. There is a need to more extensively validate the DEM since the accuracy of a number of GIS-analyses that used the DEM was dependent on the accuracy of the DEM itself.

The sustainability of the existing reservoir, the Malinao Dam, is dependent upon the land use and management carried out within the Upper Inabanga Watershed. Of particular importance is the transport of sediments from the upper areas. An erosion model such as the WEPP model could be used to assess and identify potential sources of sediments and thus design appropriate conservation measures targeting the erosion prone areas.

Sidle et al. (2006) emphasized that roads and human trails were major contributors to surface erosion as they were at least an order of magnitude greater in effect when compared to other land uses. In the WEPP model with GIS, roads and trails were not represented because of the coarse representation of the land cover map, i.e. 30 m pixel size. A further study using smaller resolution to include these land cover types could improve our understanding of the processes of soil erosion within a watershed.

Integrating land use changes with socio-economic and biophysical factors will improve understanding of the major drivers and consequences of land use change in the watershed (Mottet et al., 2006). Population pressure, land tenure and agricultural policies are suggested to be major factors affecting land use shift in the watershed. The demographic condition of the Upper Inabanga Watershed is foreseen as a major source of pressure on watershed resources. Uses and allocation of land and water resources will change but the questions of how these changes will be effected and the impact of the changes are a critical challenge to the concerned stakeholders of the Upper Inabanga Watershed.

83 REFERENCES

ACIAR (2004). "Landcare in the Philippines: Stories of people and places." ACIAR Monograph 112, J. Metcalfe, ed., ACIAR, Canberra, 170. Agus, F., Wahyunto, Tala'ohu, S. H., and Watung, R. L. (2004). "Environmental consequences of land use change in Indonesia." ISCO 2004 - 13th International Soil Conservation Organization Conference. Conserving Soil and Water for Society: Sharing Solutions. Brisbane. Aksoy, H., and Kavvas, M. L. (2005). "A review of hillslope and watershed scale erosion and sediment transport models." CATENA, 64(2-3), 247-271. Ananda, J., and Herath, G. (2003). "Soil erosion in developing countries: A socio- economic appraisal." Journal of Environmental Management, 68(4), 343-353. Arnold, J. G., Weltz, M. A., Alberts, E. E., and Flanagan, D. C. (1995). "Chapter 8. Plant growth component." USDA - Water Erosion Prediction Project Hillslope Profile and Watershed Model Documentation., D. C. Flanagan and M. A. Nearing, eds., NSERL Report No.10, USDA-ARS National Soil Erosion Research Laboratory, West Lafayette, Indiana. Ascough II, J. C., Baffaut, C., Nearing, M. A., and Flanagan, D. C. (1995). "Chapter 13. Watershed model channel hydrology and erosion processes." USDA - Water Eroison Prediction Project Hillslope Profile and Watershed Model Documentation., D. C. Flanagan and M. A. Nearing, eds., NSERL Report No.10, USDA-ARS National Soil Erosion Research Laboratory, West Lafayette, Indiana. AusAID (2001). "AusAID's area focused scoping studies: Bohol. Draft Report. 12 August 2001." Manila. Behera, S., and Panda, R. K. (2006). "Evaluation of management alternatives for an agricultural watershed in a sub-humid subtropical region using a physical process based model." Agriculture, Ecosystems & Environment, 113(1-4), 62- 72. BSWM (2005). "Final Review Reports: Integrated Watershed Management for Sustainable Soil and Water Resources Management of the Inabanga Watershed, Bohol Island, Philippines." Bureau of Soils and Water Management, Manila. Chung, S., Kim, H., and Kim, J. S. (2003). "Model development for nutrient loading from paddy rice fields." Agricultural Water Management, 62, 1-17. Cochrane, T. A. (1999). "Methodologies for watershed modeling with GIS and DEMs for the parameterization of the WEPP model," PhD Thesis, Purdue University. Cochrane, T. A., and Flanagan, D. C. (1999). "Assessing water erosion in small watersheds using WEPP with GIS and digital elevation models." Journal of Soil and Water Conservation, 54(4), 678. Cox, C., and Madramootoo, C. (1998). "Application of geographic information systems in watershed management planning in St. Lucia." Computers and Electronics in Agriculture, 20(3), 229-250. Coxhead, I., and Shively, G. (2005). "Economic development and watershed degradation." Land use change in tropical watersheds, I. Coxhead and G. Shively, eds., CABI Publishing, Wallingford, 1-18.

84 Cramb, R. A. (2000a). "Chapter 3. Agricultural land degradation in the Philippine uplands: An overview." Soil Conservation Technologies for Smallholder Farming Systems in the Philippine Uplands: A Socioeconomic Evaluation, R. A. Cramb, ed., Australian Center for International Agricultural Research, Canberra, 23-37. Cramb, R. A. (2000b). "Chapter 12. Evaluation of soil conservation technologies: A summing up." Soil Conservation Technologies for Smallholder Farming Systems in the Philippine Uplands: A Socioeconomic Evaluation, R. A. Cramb, ed., Australian Center for International Agricultural Research, Canberra, 195-212. Craswell, E. T., Sajjapongse, A., Howlett, D. J. B., and Dowling, A. J. (1998). "Agroforestry in the management of sloping lands in Asia and the Pacific." Agroforestry Systems, 38(1-3), 121-137. Cruz, R. V. O. (1999). "Integrated land use planning and sustainable watershed management." Journal of Philippine Development, 26(1), 27-49. Cruz, R. V. O., Saplaco, S. R., Lasco, R. D., Pulhin, F. B., and Avanzado, M. B. (1999). Guidelines for watershed management and development in the Philippines, PCARRD-DOST-DENR-FMB-DA-UPLB-CFNR-FDC/ENFOR, Los Banos, Laguna. Cruz, W., Francisco, H. A., and Tapawan-Conway, Z. (1988). "The on-site and downstream costs of soil erosion." Working Paper Series No 88-11. Philippine Institute for Development Studies, Manila. DAO No. 34 (1990). Department of Environment and Natural Resources Administrative Order. Revised Water Usage and Classification /Water Quality Criteria Amending Section Nos. 68 and 69, Chapter III of the 1978 NPCC Rules and Regulations. Republic of the Philippines. Davenport, T. E. (2003). The Watershed Project Management Guide. Lewis Publishers, CRC Press LLC, Boca Raton. David, W. P. (1988). "Soil and water conservation planning: Policy issues and recommendations." Journal of Philippine Development, 15(1), 47-84. Davison, P., Hutchins, M. G., Anthony, S. G., Betson, M., Johnson, C., and Lord, E. I. (2005). "The relationship between potentially erosive storm energy and daily rainfall quantity in England and Wales." Science of the Total Environment. 344(1-3), 15-25. De Baets, S., Poesen, J., Gyssels, G., and Knapen, A. (2006). "Effects of grass roots on the erodibility of topsoils during concentrated flow." Geomorphology, 76(1-2), 54-67. DeBarry, P. A. (2004). Watersheds: Processes, assessment, and management, John Wiley and Sons, Inc., Hoboken, New Jersey. DENR (2000). "Final management plan of the Inabanga-Wahig Watershed, Province of Bohol." PENRO, Tagbilaran City Ella, V. B. (2005). "Simulating soil erosion and sediment yield in small upland watersheds using the WEPP model." Land use change in tropical watersheds: Evidence, causes and remedies, I. Coxhead and G. Shively, eds., CABI Publishing, Wallingford, 109-125. FAO (1993). "Field measurement of soil erosion and runoff” Soils Bulletin No. 68, N. W. Hudson, ed. Rome. FAO (1995). "The conservation of lands in Asia and the Pacific." Rome. FAO (2000). "Land Resources Potential and Constraints at Regional and Country Levels." Land and Water Development Division. Rome.

85 Fedra, K. (1993). "GIS and environmental modeling." Environmental modeling with GIS, M. F. Goodchild, B. O. Parks, and L. T. Steyaert, eds., Oxford University Press, New York. Flanagan, D. C., Ascough II, J. C., Nicks, A. D., Nearing, M. A., and Laflen, J. M. (1995). "Chapter 1. Overview of the WEPP erosion prediction model." USDA - Water Erosion Prediction Project Hillslope Profile and Watershed Model Documentation., D. C. Flanagan and M. A. Nearing, eds., NSERL Report No.10, West Lafayette, Ind.: USDA-ARS National Soil Erosion Research Laboratory, 1.1-1.12. Flanagan, D. C., and Nearing, M. A. (1995). "USDA-Water Erosion Prediction Project Hillslope Profile and Watershed Model Documentation." NSERL Report No. 10. USDA-ARS National Soil Erosion Research Laboratory. West Lafayette. Fornis, R. L., Vermeulen, H. R., and Nieuwenhuis, J. D. (2005). "Kinetic energy- rainfall intensity relationship for Central Cebu, Philippines for soil erosion studies." Journal of Hydrology, 300(1-4), 20-32. Foster, G. R., Flanagan, D. C., Nearing, M. A., Lane, L. J., Risse, L. M., and Finkner, S. C. (1995). "Chapter 11. Hillslope erosion component." USDA - Water Eroison Prediction Project Hillslope Profile and Watershed Model Documentation., D. C. Flanagan and M. A. Nearing, eds., NSERL Report No.10. USDA-ARS National Soil Erosion Research Laboratory, West Lafayette, 11.1-11.12. Francisco, H. A. (2002). "Why watershed-based water management makes sense." Policy Notes No. 2002-09. Philippine Institute of Development Studies. Manila. Garbrecht, J., and Martz, L. W. (1999). TOPAZ: An automated digital landscape analysis tool for topographic evaluation, drainage identification, watershed segmentation, and subcatchment parameterization. USDA - Agricultural Research Service, Grazinglands Research Laboratory, El Reno, Oklahoma. Hoyos, N., Waylen, P. R., and Jaramillo, A. (2005). "Seasonal and spatial patterns of erosivity in a tropical watershed of the Colombian Andes." Journal of Hydrology, 314(1-4), 177-191. Hudson, N. W. (1995). Soil Conservation, BT Batsford Limited, London. Iijima, M., Izumi, Y., Yuliadi, E., Sunyoto, and Ardjasa, W. S. (2004). "Cassava- based intercropping systems on Sumatra Island in Indonesia: Productivity, soil erosion, and rooting zone." Plant Production Science, 7(3), 347-355. IWMI (2002). "Catchment approach to managing soil erosion in Asia. ADB RETA No. 5803." Technical Assistance (TA) Completion Report, International Water Management Institute, Colombo, 88. Kinnell, P. I. A. (1983). "The effect of kinetic energy of excess rainfall on soil loss from non-vegetated plots." Australian Journal of Soil Research(21), 445-53. Lal, R. (1985). "A soil suitability guide for different tillage systems in the tropics." Soil and Tillage Research, 5(2), 179-196. Lal, R. (1994). Soil erosion research methods, Soil and Water Conservation Society, Ankeny. Lantican, M. A., Guerra, L. C., and Bhuiyan, S. I. (2003). "Impacts of soil erosion in the upper Manupali watershed on irrigated lowlands in the Philippines." Paddy and Water Environment, 1(1), 19-26. Lapar, M. L. A., and Pandey, S. (1999). "Adoption of soil conservation: the case of the Philippine uplands." Agricultural Economics, 21(3), 241-256.

86 Lee, H.Y., Lin, Y.T., and Chin, Y.J. (2006). Quantitative estimation of reservoir sedimentation from three typhoon events. Journal of Hydrologic Engineering. 11(4),362-370. Lenzi, M. A., and Di Luzio, M. (1997). "Surface runoff, soil erosion and water quality modelling in the Alpone watershed using AGNPS integrated with a Geographic Information System." European Journal of Agronomy, 6(1-2), 1- 14. Marcoux, A. (1996). "Linkages between Population, Natural Resources and Environment in China, Philippines, Indonesia and Viet Nam." Sustainable Development Department (SD), Food and Agriculture Organization of the United Nations, Rome. Merritt, W. S., Letcher, R. A., and Jakeman, A. J. (2003). "A review of erosion and sediment transport models." Environmental Modelling & Software, 18, 761- 799. Minkowski, M. (2005). "What about my data? Preparing data for use in GeoWEPP." Department of Geography, University of Buffalo, New York. Moore, I. D., and Burch, G. J. (1986). "Physical basis of the length-slope factor in the Universal Soil Loss Equation." Soil Science Society of American Journal, 50, 1294-1298. Morgan, R. P. C. (2005). “Soil erosion and conservation.” Soil erosion research methods, R. Lal, ed., 3rd Ed Blackwell Publishing, London Mottet, A., Ladet, S., Coque, N., and Gibon, A. (2006). "Agricultural land-use change and its drivers in mountain landscapes: A case study in the Pyrenees." Agriculture, Ecosystems & Environment, 114(2-4), 296-310. Mutchler, C. K., Murphree, C. E., and McGregor, K. C. (1994). Laboratory and field plots for erosion research In: St. Lucie Press, Florida. Nearing, M. A., Lane, L. J., and Lopes, V. L. (1994). "Modeling soil erosion." Soil erosion research methods, R. Lal, ed., Soil and Water Conservation Society, Ankeny, 127-156. Nelson, R. A., Cramb, R. A., Menz, K. M., and Mamicpic, M. A. (1998). "Cost- benefit analysis of alternative forms of hedgerow intercropping in the Philippine uplands." Agroforestry Systems, 39(3), 241-262. NSO (2002). National Statistics Office. Republic of the Philippines Paningbatan, E. P., Ciesiolka, C. A., Coughlan, K. J., and Rose, C. W. (1995). "Alley cropping for managing soil erosion of hilly lands in the Philippines." Soil Technology, 8(3), 193-204. PCARRD (1984). "The Philippines Recommends for Soil Conservation." PCARRD Technical Bulletin Series NO. 28-A, Los Banos, Laguna. PCARRD (1991). State of the art and abstract bibliography: Soil and water conservation in the Philippine upland watersheds, PCARRD, Los Banos, Laguna. Pereira, H. C. (1989). Policy and practice in the management of tropical watersheds Belhaven Press, London. Philippines. (1975). "Presidential Decree No 705. Revised Forestry Code of the Philippines ", Manila. Poudel, D. D., Midmore, D. J., and West, L. T. (1999). "Erosion and productivity of vegetable systems on sloping volcanic ash-derived Philippine soils." Soil Science Society of America Journal, 63(5), 1366-1376. Presbitero, A. L. (2003). "Soil erosion studies on steep slopes of humid-tropic Philippines," PhD thesis, Griffith University, Queensland.

87 Presbitero, A.L., Escalante, M.C., Rose, C.W., Coughlan, K.J. and Ciesiolka, C.A. (1995). Erodibility evaluation and the effects of land management practices on soil erosion from steep slopes in Leyte, the Philippines. Soil Technology. 8(3), 205-213 Putthacharoen, S., Howeler, R. H., Jantawat, S., and Vichukit, V. (1998). "Nutrient uptake and soil erosion losses in cassava and six other crops in a Psamment in eastern Thailand." Field Crops Research, 57(1), 113-126. Renard, K. G., Foster, G. R., Weesies, D. K., McCool, D. K., and Yoder, D. C. (1996). Predicting soil erosion by water: A guide to conservation planning with the Revised Universal Soil Loss Equation (RUSLE). U.S. Department of Agriculture, Agriculture Handbook No. 703. Renschler, C. S. (2003). "Designing geo-spatial interfaces to scale process models: the GeoWEPP approach." Hydrological Processes, 17(5), 1005-1017. Salles, C., Poesen, J., and Sempere-Torres, D. (2002). "Kinetic energy of rain and its functional relationship with intensity." Journal of Hydrology, 257(1-4), 256- 270. Savabi, M. R., and Williams, J. R. (1995). "Chapter 5. Water balance and percolation." USDA - Water Eroison Prediction Project Hillslope Profile and Watershed Model Documentation., D. C. Flanagan and M. A. Nearing, eds., NSERL Report No.10. USDA-ARS National Soil Erosion Research Laboratory, West Lafayette, 5.1-5.14 Schwab, G. O., Fangmeier, D. D., Elliot, W. J., and Frevert, R. K. (1992). Soil and water conservation engineering, John Wiley and Sons, Inc., New York. Sidle, R. C., Ziegler, A. D., Negishi, J. N., Nik, A. R., Siew, R., and Turkelboom, F. (2006). "Erosion processes in steep terrain--Truths, myths, and uncertainties related to forest management in Southeast Asia." Forest Ecology and Management, 224(1-2), 199-225. Stark, M., Itumay, J., and Nulla, S. (2003). "Assessment of natural vegetative contour strips for soil conservation on shallow calcareous soils in the central Philippines." Accomplishment Report. November 2000 - July 2003. ICRAF, Baybay, Leyte, Philippines Steiner, K. G. (1996). Causes of soil degradation and development approaches to sustainable soil management. Pilot Project Sustainable Soil Management. GTZ., Margraf Verlag. Stone, J. J., Lane, L. J., Shirley, E. D., and Hernandez, M. (1995). "Chapter 4. Hillslope Surface Hydrology." USDA - Water Erosion Prediction Project Hillslope Profile and Watershed Model Documentation., D. C. Flanagan and M. A. Nearing, eds., NSERL Report No.10. USDA-ARS National Soil Erosion Research Laboratory, West Lafayette, 4.1-4.20. Thapa, B. B., Garrity, D. P., Cassel, D. K., and Mercado, A. R. (2000). "Contour grass strips and tillage affect corn production on Philippine steepland oxisols." Agronomy Journal, 92(1), 98-105. UNCCD (1994). "United Nations Convention to Combat Desertification: The Convention. Part I. Introduction. Article 1. p5." Bonn. http://www.unccd.int USDA-ARS (1995). "USDA- Water Erosion Prediction Project: Hillslope profile and watershed model documentation." USDA-ARS National Soil Erosion Research Laboratory, West Lafayette. Valentin, C., Poesen, J., and Li, Y. (2005). "Gully erosion: Impacts, factors and control." CATENA, 63(2-3), 132-153.

88 Van Dijk, A. I. J. M. and Bruijnzeel, L. A.(2004). Runoff and soil loss from bench terraces.2. An event-based erosion process model. European Journal of Soil Science. 55, 317-334 van Dijk, A. I. J. M., Bruijnzeel, L. A., and Rosewell, C. J. (2002). "Rainfall intensity-kinetic energy relationships: a critical literature appraisal." Journal of Hydrology, 261(1-4), 1-23. van Noordwijk, M., Lusiana, B., and Khasanah, N. (2004). WANULCAS version 3.1, Background on a model of water nutrient and light capture in agroforestry systems. International Center for Research in Agroforestry (ICRAF), Bogor, Indonesia Walling, D. E. (1994). "Measuring sediment yield from river basins." Soil Erosion Research Methods, R. Lal, ed., Soil and Water Conservation Society, Ankeny White, S. (1995). "Soil erosion and sediment yield in the Philippines." Sediment and Water Quality in River Catchments, I. D. L. Foster, A. M. Gurnell, and B. W. Webb, eds., John Wiley & Sons Ltd., New York. Wischmeier, W. H., and Smith, D. D. (1978). Predicting rainfall erosion losses - a guide to conservation planning, U.S. Department of Agriculture, Agricultural Handbook No. 537. Yu, B., and Rosewell, C. J. (1996). "Rainfall erosivity estimation using daily rainfall amounts for South Australia." Australian Journal of Soil Research, 34, 721- 733. Zeleke, G., Winter, T., and Flanagan, D. (1999). "BPCDG: Break point data generator using observed standard weather data sets." USDA-ARS National Soil Erosion Research Laboratory, West Lafayette, Indiana. Zhang, J. H., Lobb, D. A., Li, Y., and Liu, G. C. (2004). "Assessment of tillage translocation and tillage erosion by hoeing on the steep land in hilly areas of Sichuan, China." Soil and Tillage Research, 75(2), 99-107.

LIST OF APPENDICES

Appendix A Creating a digital elevation model ……….. 91 Appendix B WEPP-GeoWEPP model inputs ……….. 94 Soil properties used as input to WEPP model Appendix B-1 ……….. 94 simulations Sample climate input files for running Appendix B-2 ……….. 96 BPCDG Appendix B-3 GIS-based input data to GeoWEPP model ……….. 97 Appendix C On-site monitoring data ……….. 98 Weekly rainfall, runoff and soil loss from the Appendix C-1 agroforest plot collected from March 2004 – ……….. 98 February 2006 Weekly rainfall, runoff and soil loss from the Appendix C-2 cassava/corn plot collected from March 2004 ……….. 99 – February 2006 Weekly rainfall, runoff and soil loss from the Appendix C-3 forest plot collected from March 2004 – ……….. 100 February 2006 Weekly rainfall, runoff and soil loss from the Appendix C-4 grassland plot collected from March 2004 – ……….. 101 February 2006 Weekly rainfall, runoff and soil loss from the Appendix C-5 oil palm plot collected from March 2004 – ……….. 102 February 2006 Average daily discharge and rainfall records Appendix C-6 ……….. 103 from the Bugsok AWaS and AWeS Average daily discharge and rainfall records Appendix C-7 ……….. 106 from the Pamacsalan AWaS and AWeS

Appendix A. Creating a Digital Elevation Model (DEM)

A DEM is a surface representation of elevation attributes. It is becoming one of the core databases of geographic information system (GIS) applications especially in three-dimensional representation of landforms. A DEM is used in hydrological and ecological modeling, and simulation models of landscape processes (Burrough and Mcdonnell, 1998). It can be represented either in a regular grid of cells or in a triangular irregular network (TIN) format. The grid elevation matrix is the most available form of DEM.

Map digitization

Scanned topographic map sheets with a scale of 1:50,000 were procured from the National Mapping and Resource Information Authority (NAMRIA) and registered to a predefined projected coordinate system. Contour lines at 20m contour intervals were digitized using ArcMap onscreen digitization. Supplementary contours at 10m and 5m intervals at low-lying areas were included. Other information such as rivers (streamlines) and spot elevation points were also extracted from the topographic map.

The TopoToRaster Tool

The TopoToRaster Tool is one of the interpolation methods available in the

Spatial Analyst Extension of ESRI’s ArcGIS 9. The tool is specifically designed to create a hydrologically correct digital elevation model from contour lines, lakes and elevation points utilizing the ANUDEM (Australian National University Digital

Elevation Model) algorithm. The process of creating a DEM using this tool is illustrated in the figure below:

Inputs

DEM

FlowDirection

Sinks

NO Depressionless Any sinks? DEM YES

Fill

The inputs to the TopoToRaster Tool can be:

• Point elevation – a point feature class representing point elevations • Contour lines – a line feature class representing elevation contours • Stream lines – a line feature class representing streams or rivers. The lines are required to point downstream and should be single arc streams

• Sink – a point feature class representing sinks or depressions • Boundary – a polygon feature class that set out the boundary of the output raster

• Lake – a polygon feature class specifying the location of lakes The TopoToRatser tool can accept 6 types of input data with primary inputs being either elevation lines or point features. In this study, the primary inputs were contour elevation lines while the secondary inputs included stream lines and spot elevation points. The stream network was composed of single arcs in a dendritic

92 pattern; all the arcs were pointing down slope and there were no polygons (lakes) or braided streams. Drainage enforcement was applied while other parameters were kept to default. After running the tool, a DEM was produced. To create a sink-free DEM, the following steps were conducted: a) A flow direction raster was created using the FlowDirection tool. The tool assigned one of the eight valid directions (1,2,4,8,16,32,64,128) to each cell. A flow direction raster showing values other than the eight specified values indicated the presence of spurious sinks or errors. Sinks are cells whose elevations are lower than the surrounding cells with flow direction values other than 8 valid directions. b) The Sink tool was used to identify the sinks/depressions created during the interpolation process. c) The Fill Tool was used to remove the sinks. A zero sink surface indicated a depressionless DEM. d) To test the presence of any remaining sinks, a new flow direction surface was created using the filled DEM. The new flow direction surface now only has 8 values indicating the created DEM was free of sinks (depressioness).

93 Appendix B. WEPP-GeoWEPP model inputs

Appendix Table B-1. Soil properties used as input to WEPP-GeoWEPP model simulations. CEC SMU Albedo Layer Depth (cm) Sand (%) Silt (%) OM (%) (meq/100 Code (%) g soil) A 0-25 54.40 21.60 2.00 14.36 22 B 25-40 54.40 17.60 1.38 11.34 BoC1 C 40-60 48.40 19.60 1.10 8.27 D 60-100 44.40 15.60 0.97 7.53 E 100-150 32.40 17.60 1.35 6.94 A 0-14 4.80 36.40 5.46 45.86 23 B 14-37 4.80 32.40 4.99 41.82 C 37-60 4.80 36.40 3.31 42.82 CaA D 60-76 10.80 38.40 3.01 42.89 E 79-96 12.80 38.40 1.68 40.70 F 96-124 14.80 46.40 2.97 36.22 G 124-150 12.80 46.40 2.62 38.45 A 0-18 18.80 30.40 4.56 45.90 23 B 18-38 20.80 32.40 2.37 41.40 C 38-48 32.80 26.40 2.32 42.71 CaB D 48-62 38.80 26.40 1.25 41.38 E 62-74 48.80 24.40 2.54 37.97 F 74-130 44.80 28.40 2.06 40.24 G 130-160 42.80 28.40 1.72 36.56 A 0-21 18.00 28.00 3.96 28.85 23 B 21-47 30.00 24.00 3.12 25.98 CbC1 C 47-66 24.00 18.00 3.26 24.82 D 66-150 34.00 28.00 3.12 15.61 A 0-14 14.00 32.00 3.91 27.97 23 B 14-29 18.00 31.00 4.74 25.10 CbC2 C 29-55 15.00 34.00 3.06 21.18 D 55-82 38.00 26.00 1.79 19.27 E 82-150 44.00 28.00 2.41 17.05 A 0-13 20.40 33.60 7.48 49.26 24 FaB B 13-27 20.40 41.60 1.97 36.31 C 27-60 26.40 39.60 1.91 40.39 A 0-17 30.80 22.40 4.10 42.92 23 FaBoE3 B 17-36 26.80 24.40 2.72 36.35 C 36-60 20.80 24.40 3.10 24.15 A 0-18 22.40 16.20 3.53 32.03 23 FaC2 B 18-50 18.40 12.20 3.24 21.66 A 0-14 34.40 19.60 3.10 32.39 23 B 14-27 32.40 21.60 3.07 29.63 C 27-46 32.40 19.60 2.41 27.23 FaD2 D 46-68 34.40 17.60 1.41 26.38 E 68-90 34.40 13.60 4.19 23.13 F 90-150 42.40 15.60 2.03 21.05 A 0-15 44.80 20.40 3.91 34.69 23 B 15-40 60.80 16.40 4.60 26.14 Lu/Fa/Bo/E3 C 40-55 66.80 12.40 2.75 17.21 D 55-100 60.80 14.40 1.89 12.27

94 A 0-25 44.40 29.60 2.57 14.56 23 B 25-40 46.40 23.60 1.10 12.15 UbB1 C 40-80 44.40 21.60 1.22 10.70 D 80-150 30.40 15.60 1.10 10.08 A 0-12 40.40 31.60 2.50 23..85 23 B 12-23 18.40 31.60 1.53 22.40 C 23-44 36.40 23.60 2.60 19.51 UbB1 D 44-62 22.40 27.60 0.78 14.26 E 62-97 8.40 35.60 1.53 9.09 F 97-150 18.40 58.60 1.10 7.72 A 0-10 36.80 44.40 3.91 10.05 23 B 10-32 60.80 20.40 0.81 13.17 UbC3 C 32-70 38.80 24.40 1.03 9.10 D 70-110 12.80 40.40 1.29 5.09 6.0 Albedo = ( OM *) 100 for the first layer only. e 4.0

95 Appendix Figure B-2. Sample climate input files for running the Breakpoint Climate Data Generator (BPCDG).

BPCDG inputs:

BPCDG output (input to WEPP model)

96 Appendix Figure B-3. GIS-based input data to GeoWEPP model

97 Appendix C. On-site monitoring data

Appendix Table C-1. Weekly rainfall, runoff and soil loss from the agroforest plot collected from March 2004 – February 2006.

Rainfall Runoff Soil loss Rainfall Runoff Soil loss Week Date -1 Week Date -1 (mm) (mm) (t ha ) (mm) (mm) (t ha ) 1 03/21/ - 03/27 04 15 7.1111 0.0441 51 03/06 - 03/12 05 157 20.0943 0.1246 2 03/28 - 04/03 04 0 0.0000 0.0000 52 03/13 - 03/19 05 0 0.0000 0.0000 3 04/04 - 04/10 04 0 0.0000 0.0000 53 03/20 - 03/26 05 0 0.0000 0.0000 4 04/11 - 04/17 04 26 0.0158 0.0001 54 03/27 - 04/02 05 3 0.0010 0.0000 5 04/18 - 04/24 04 0 0.0000 0.0000 55 04/03 - 04/09 05 5 0.0021 0.0000 6 04/25 - 05/01 04 40 0.0182 0.0001 56 04/10 - 04/16 05 7 0.0043 0.0000 7 05/02 - 05/08 04 33 0.0239 0.0001 57 04/17 - 04/23 05 0 0.0000 0.0000 8 05/09 - 05/15 04 171 1.7508 0.0109 58 04/24 - 04/30 05 0 0.0000 0.0000 9 05/16 - 05/22 04 0 0.0000 0.0000 59 05/01 - 05/07 05 0 0.0000 0.0000 10 05/23 - 05/29 04 0 0.0000 0.0000 60 05/08 - 05/14 05 6 0.0013 0.0000 11 05/30 - 06/05 04 27 0.0038 0.0000 61 05/15 - 05/21 05 12 0.0069 0.0000 12 06/06 - 06/12 04 193 35.0150 0.2171 62 05/22 - 05/28 05 31 0.0151 0.0001 13 06/13 - 06/19 04 111 0.5848 0.0036 63 05/29 - 06/04 05 0 0.0000 0.0000 14 06/20 - 06/26 04 0 0.0000 0.0000 64 06/05 - 06/11 05 17 0.0163 0.0001 15 06/27 - 07/03 04 14 0.0080 0.0000 65 06/12 - 06/18 05 22 0.0218 0.0001 16 07/04 - 07/10 04 21 0.0090 0.0001 66 06/19 - 06/25 05 68 0.4736 0.0029 17 07/11 - 07/17 04 112 51.0806 0.3168 67 06/26 - 07/02 05 34 0.3438 0.0021 18 07/18 - 07/24 04 263 149.0253 0.9242 68 07/03 - 07/09 05 133 17.0442 0.1057 19 07/25 - 07/31 04 37 14.8686 0.0922 69 07/10 - 07/16 05 21 3.1306 0.0194 20 08/01 - 08/07 04 13 0.0074 0.0000 70 07/17 - 07/23 05 5 0.0038 0.0000 21 08/08 - 08/14 04 9 0.0065 0.0000 71 07/24 - 07/30 05 150 9.7538 0.0605 22 08/15 - 08/21 04 0 0.0000 0.0000 72 07/31 - 08/06 05 47 0.0742 0.0005 23 08/22 - 08/28 04 0 0.0000 0.0000 73 08/07 - 08/13 05 7 0.0034 0.0000 24 08/29 - 09/04 04 48 0.0188 0.0001 74 08/14 - 08/20 05 13 0.0109 0.0001 25 09/05 - 09/11 04 0 0.0000 0.0000 75 08/21 - 08/27 05 39 0.0382 0.0002 26 09/12 - 09/18 04 28 0.0139 0.0001 76 08/28 - 09/03 05 7 0.0047 0.0000 27 09/19 - 09/25 04 3 0.0010 0.0000 77 09/04 - 09/10 05 53 0.0605 0.0004 28 09/26 - 10/02 04 66 0.5217 0.0032 78 09/11 - 09/17 05 355 77.1320 0.4783 29 10/03 - 10/09 04 11 0.0021 0.0000 79 09/18 - 09/24 05 0 0.0000 0.0000 30 10/10 - 10/16 04 36 0.0154 0.0001 80 09/25 - 10/01 05 28 0.0309 0.0002 31 10/17 - 10/23 04 61 3.0801 0.0191 81 10/02 - 10/08 05 19 0.0185 0.0001 32 10/24 - 10/30 04 0 0.0000 0.0000 82 10/09 - 10/15 05 41 4.3653 0.0271 33 10/31 - 11/06 04 0 0.0000 0.0000 83 10/16 - 10/22 05 31 0.0265 0.0002 34 11/07 - 11/13 04 0 0.0000 0.0000 84 10/23 - 10/29 05 16 0.0118 0.0001 35 11/14 - 11/20 04 16 0.0038 0.0000 85 10/30 - 11/05 05 30 0.0301 0.0002 36 11/21 - 11/27 04 37 0.0255 0.0002 86 11/06 - 11/12 05 38 0.0410 0.0003 37 11/28 - 12/04 04 50 0.0505 0.0003 87 11/13 - 11/19 05 3 0.0010 0.0000 38 12/05 - 12/11 04 116 9.7156 0.0603 88 11/20 - 11/26 05 28 0.0191 0.0001 39 12/12 - 12/18 04 22 0.0176 0.0001 89 11/27 - 12/03 05 37 0.0279 0.0002 40 12/19 - 12/25 04 39 0.0316 0.0002 90 12/04 - 12/10 05 121 71.3690 0.4426 41 12/26 - 01/01 05 52 0.0591 0.0004 91 12/11 - 12/17 05 182 73.0991 0.4533 42 01/02 - 01/08 05 22 0.0218 0.0001 92 12/18 - 12/24 05 131 71.0385 0.4405 43 01/09 - 01/15 05 0 0.0000 0.0000 93 12/25 - 12/31 05 28 1.9977 0.0124 44 01/16 - 01/22 05 5 0.0010 0.0000 94 01/01 - 10/07 06 23 0.0143 0.0001 45 01/23 - 01/29 05 35 0.0187 0.0001 95 01/08 - 01/14 06 121 70.7180 0.4385 46 01/30 - 02/05 05 0 0.0000 0.0000 96 01/15 - 01/21 06 18 0.0076 0.0000 47 02/06 - 02/12 05 0 0.0000 0.0000 97 01/22 - 01/28 05 4 0.0010 0.0000 48 02/13 - 02/19 05 20 0.0111 0.0001 98 01/29 - 02/04 06 0 0.0000 0.0000 49 02/20 - 02/26 05 7 0.0086 0.0001 TOTAL 3850 694.2007 4.3050 50 02/27 - 03/05 05 0 0.0000 0.0000

Appendix Table C-2. Weekly rainfall, runoff and soil loss from the cassava/corn plot collected from March 2004 – February 2006.

Rainfall Runoff Soil loss Rainfall Runoff Soil loss Week Date -1 Week Date -1 (mm) (mm) (t·ha ) (mm) (mm) (t·ha ) 1 03/21/ - 03/27 04 122 101.2882 6.1719 51 03/06 - 03/12 05 163 2.0516 0.1250 2 03/28 - 04/03 04 11 0.0030 0.0002 52 03/13 - 03/19 05 28 0.1390 0.0085 3 04/04 - 04/10 04 0 0.0000 0.0000 53 03/20 - 03/26 05 21 0.2559 0.0156 4 04/11 - 04/17 04 2 0.0000 0.0000 54 03/27 - 04/02 05 1 0.0000 0.0000 5 04/18 - 04/24 04 0 0.0000 0.0000 55 04/03 - 04/09 05 4 0.0036 0.0002 6 04/25 - 05/01 04 65 0.1282 0.0078 56 04/10 - 04/16 05 0 0.0017 0.0001 7 05/02 - 05/08 04 15 0.0252 0.0015 57 04/17 - 04/23 05 0 0.0000 0.0000 8 05/09 - 05/15 04 84 36.7550 2.2396 58 04/24 - 04/30 05 10 0.0099 0.0006 9 05/16 - 05/22 04 7 0.0062 0.0004 59 05/01 - 05/07 05 0 0.0000 0.0000 10 05/23 - 05/29 04 34 0.1306 0.0080 60 05/08 - 05/14 05 11 0.0166 0.0010 11 05/30 - 06/05 04 10 0.0109 0.0007 61 05/15 - 05/21 05 70 21.4277 1.3057 12 06/06 - 06/12 04 148 66.6874 4.0635 62 05/22 - 05/28 05 0 0.0001 0.0000 13 06/13 - 06/19 04 28 8.0821 0.4925 63 05/29 - 06/04 05 26 1.9712 0.1201 14 06/20 - 06/26 04 1 0.0000 0.0000 64 06/05 - 06/11 05 18 3.7199 0.2267 15 06/27 - 07/03 04 11 0.0342 0.0021 65 06/12 - 06/18 05 54 1.8634 0.1135 16 07/04 - 07/10 04 9 0.0084 0.0005 66 06/19 - 06/25 05 44 14.6256 0.8912 17 07/11 - 07/17 04 71 4.0673 0.2478 67 06/26 - 07/02 05 82 54.3910 3.3143 18 07/18 - 07/24 04 47 5.7750 0.3519 68 07/03 - 07/09 05 42 8.2267 0.5013 19 07/25 - 07/31 04 49 10.2608 0.6252 69 07/10 - 07/16 05 85 60.9986 3.7169 20 08/01 - 08/07 04 6 0.0237 0.0014 70 07/17 - 07/23 05 76 42.8544 2.6113 21 08/08 - 08/14 04 4 0.0075 0.0005 71 07/24 - 07/30 05 158 14.7353 0.8979 22 08/15 - 08/21 04 2 0.0137 0.0008 72 07/31 - 08/06 05 20 0.0643 0.0039 23 08/22 - 08/28 04 0 0.0000 0.0000 73 08/07 - 08/13 05 0 0.0000 0.0000 24 08/29 - 09/04 04 19 0.2184 0.0133 74 08/14 - 08/20 05 65 5.9342 0.3616 25 09/05 - 09/11 04 47 2.8685 0.1748 75 08/21 - 08/27 05 41 5.6701 0.3455 26 09/12 - 09/18 04 49 1.7404 0.1061 76 08/28 - 09/03 05 23 4.1572 0.2533 27 09/19 - 09/25 04 7 0.0067 0.0004 77 09/04 - 09/10 05 68 12.7065 0.7743 28 09/26 - 10/02 04 123 65.1953 3.9726 78 09/11 - 09/17 05 156 42.8896 2.6134 29 10/03 - 10/09 04 15 0.7286 0.0444 79 09/18 - 09/24 05 57 2.3380 0.1425 30 10/10 - 10/16 04 225 1.1432 0.0697 80 09/25 - 10/01 05 1 0.0022 0.0001 31 10/17 - 10/23 04 42 10.3991 0.6337 81 10/02 - 10/08 05 116 21.8680 1.3325 32 10/24 - 10/30 04 8 0.0678 0.0041 82 10/09 - 10/15 05 25 3.9386 0.2400 33 10/31 - 11/06 04 0 0.0000 0.0000 83 10/16 - 10/22 05 0 0.0000 0.0000 34 11/07 - 11/13 04 3 0.0000 0.0000 84 10/23 - 10/29 05 82 7.2944 0.4445 35 11/14 - 11/20 04 0 0.0000 0.0000 85 10/30 - 11/05 05 2 0.0011 0.0001 36 11/21 - 11/27 04 41 2.4940 0.1520 86 11/06 - 11/12 05 14 0.5561 0.0339 37 11/28 - 12/04 04 48 3.7743 0.2300 87 11/13 - 11/19 05 52 5.4746 0.3336 38 12/05 - 12/11 04 58 7.1176 0.4337 88 11/20 - 11/26 05 22 0.4440 0.0271 39 12/12 - 12/18 04 7 0.0090 0.0005 89 11/27 - 12/03 05 0 0.0000 0.0000 40 12/19 - 12/25 04 20 0.6320 0.0385 90 12/04 - 12/10 05 134 71.3791 4.3494 41 12/26 - 01/01 05 52 8.3450 0.5085 91 12/11 - 12/17 05 461 344.0925 20.9669 42 01/02 - 01/08 05 27 0.0651 0.0040 92 12/18 - 12/24 05 231 62.5255 3.8099 43 01/09 - 01/15 05 12 0.0865 0.0053 93 12/25 - 12/31 05 255 113.6797 6.9270 44 01/16 - 01/22 05 10 0.0056 0.0003 94 01/01 - 10/07 06 38 4.1333 0.2519 45 01/23 - 01/29 05 12 0.0302 0.0018 95 01/08 - 01/14 06 98 20.8919 1.2730 46 01/30 - 02/05 05 18 0.0401 0.0024 96 01/15 - 01/21 06 24 10.1848 0.6206 47 02/06 - 02/12 05 0 0.0000 0.0000 97 01/22 - 01/28 05 7 0.0091 0.0006 48 02/13 - 02/19 05 14 0.0092 0.0006 98 01/29 - 02/04 06 45 5.1628 0.3146 49 02/20 - 02/26 05 0 0.0000 0.0000 TOTAL 4515 1310.97 79.88 50 02/27 - 03/05 05 1 0.0003 0.0000

Appendix Table C.3. Weekly rainfall, runoff and soil loss from the forest plot collected from March 2004 – February 2006.

Rainfall Runoff Soil loss Rainfall Runoff Soil loss Week Date -1 Week Date -1 (mm) (mm) (t ha ) (mm) (mm) (t ha ) 1 03/21/ - 03/27 04 15 0.0279 0.0008 51 03/06 - 03/12 05 157 0.1003 0.0028 2 03/28 - 04/03 04 0 0.0000 0.0000 52 03/13 - 03/19 05 0 0.0000 0.0000 3 04/04 - 04/10 04 0 0.0000 0.0000 53 03/20 - 03/26 05 0 0.0000 0.0000 4 04/11 - 04/17 04 26 0.0309 0.0009 54 03/27 - 04/02 05 3 0.0000 0.0000 5 04/18 - 04/24 04 0 0.0000 0.0000 55 04/03 - 04/09 05 5 0.0010 0.0000 6 04/25 - 05/01 04 40 0.0681 0.0019 56 04/10 - 04/16 05 7 0.0021 0.0001 7 05/02 - 05/08 04 33 0.0340 0.0010 57 04/17 - 04/23 05 0 0.0000 0.0000 8 05/09 - 05/15 04 171 0.1153 0.0033 58 04/24 - 04/30 05 0 0.0000 0.0000 9 05/16 - 05/22 04 0 0.0000 0.0000 59 05/01 - 05/07 05 0 0.0000 0.0000 10 05/23 - 05/29 04 0 0.0000 0.0000 60 05/08 - 05/14 05 6 0.0032 0.0001 11 05/30 - 06/05 04 27 0.0188 0.0005 61 05/15 - 05/21 05 12 0.0042 0.0001 12 06/06 - 06/12 04 193 0.2302 0.0065 62 05/22 - 05/28 05 31 0.0214 0.0006 13 06/13 - 06/19 04 111 0.1134 0.0032 63 05/29 - 06/04 05 0 0.0000 0.0000 14 06/20 - 06/26 04 0 0.0000 0.0000 64 06/05 - 06/11 05 17 0.0155 0.0004 15 06/27 - 07/03 04 14 0.0028 0.0001 65 06/12 - 06/18 05 22 0.0171 0.0005 16 07/04 - 07/10 04 21 0.0191 0.0005 66 06/19 - 06/25 05 68 0.0333 0.0009 17 07/11 - 07/17 04 112 0.1908 0.0054 67 06/26 - 07/02 05 34 0.0158 0.0004 18 07/18 - 07/24 04 263 9.5397 0.2710 68 07/03 - 07/09 05 133 0.0899 0.0026 19 07/25 - 07/31 04 37 0.1470 0.0042 69 07/10 - 07/16 05 21 0.0483 0.0014 20 08/01 - 08/07 04 13 0.0156 0.0004 70 07/17 - 07/23 05 5 0.0021 0.0001 21 08/08 - 08/14 04 9 0.0117 0.0003 71 07/24 - 07/30 05 150 0.0886 0.0025 22 08/15 - 08/21 04 0 0.0000 0.0000 72 07/31 - 08/06 05 47 0.0816 0.0023 23 08/22 - 08/28 04 0 0.0000 0.0000 73 08/07 - 08/13 05 7 0.0057 0.0002 24 08/29 - 09/04 04 48 0.0496 0.0014 74 08/14 - 08/20 05 13 0.0225 0.0006 25 09/05 - 09/11 04 0 0.0000 0.0000 75 08/21 - 08/27 05 39 0.0327 0.0009 26 09/12 - 09/18 04 28 0.0278 0.0008 76 08/28 - 09/03 05 7 0.0026 0.0001 27 09/19 - 09/25 04 3 0.0021 0.0001 77 09/04 - 09/10 05 53 0.0486 0.0014 28 09/26 - 10/02 04 66 0.0809 0.0023 78 09/11 - 09/17 05 355 0.2577 0.0073 29 10/03 - 10/09 04 11 0.0059 0.0002 79 09/18 - 09/24 05 0 0.0000 0.0000 30 10/10 - 10/16 04 36 0.0449 0.0013 80 09/25 - 10/01 05 28 0.0238 0.0007 31 10/17 - 10/23 04 61 0.0783 0.0022 81 10/02 - 10/08 05 19 0.0144 0.0004 32 10/24 - 10/30 04 0 0.0000 0.0000 82 10/09 - 10/15 05 41 0.0236 0.0007 33 10/31 - 11/06 04 0 0.0000 0.0000 83 10/16 - 10/22 05 31 0.0197 0.0006 34 11/07 - 11/13 04 0 0.0000 0.0000 84 10/23 - 10/29 05 16 0.0197 0.0006 35 11/14 - 11/20 04 16 0.0271 0.0008 85 10/30 - 11/05 05 30 0.0181 0.0005 36 11/21 - 11/27 04 37 0.0362 0.0010 86 11/06 - 11/12 05 38 0.0297 0.0008 37 11/28 - 12/04 04 50 0.0397 0.0011 87 11/13 - 11/19 05 3 0.0000 0.0000 38 12/05 - 12/11 04 116 0.0580 0.0016 88 11/20 - 11/26 05 28 0.0086 0.0002 39 12/12 - 12/18 04 22 0.0223 0.0006 89 11/27 - 12/03 05 37 0.0036 0.0001 40 12/19 - 12/25 04 39 0.0416 0.0012 90 12/04 - 12/10 05 121 0.0536 0.0015 41 12/26 - 01/01 05 52 0.0788 0.0022 91 12/11 - 12/17 05 182 0.3280 0.0093 42 01/02 - 01/08 05 22 0.0171 0.0005 92 12/18 - 12/24 05 131 0.0527 0.0015 43 01/09 - 01/15 05 0 0.0000 0.0000 93 12/25 - 12/31 05 28 0.0241 0.0007 44 01/16 - 01/22 05 5 0.0021 0.0001 94 01/01 - 10/07 06 23 0.0046 0.0001 45 01/23 - 01/29 05 35 0.0395 0.0011 95 01/08 - 01/14 06 121 0.0527 0.0015 46 01/30 - 02/05 05 0 0.0000 0.0000 96 01/15 - 01/21 06 18 0.0027 0.0001 47 02/06 - 02/12 05 0 0.0000 0.0000 97 01/22 - 01/28 05 4 0.0000 0.0000 48 02/13 - 02/19 05 20 0.0093 0.0003 98 01/29 - 02/04 06 0 0.0000 0.0000

49 02/20 - 02/26 05 7 0.0034 0.0001 3850 12.8036 0.3637 TOTAL 50 02/27 - 03/05 05 0 0.0000 0.0000

100

Appendix Table C-4. Weekly rainfall, runoff and soil loss from the grassland plotollected from March 2004 – February 2006.

Rainfall Runoff Soil loss Rainfall Runoff Soil loss Week Date -1 Week Date -1 (mm) (mm) (t·ha ) (mm) (mm) (t·ha ) 1 03/21/ - 03/27 04 122 38.1173 0.4835 51 03/06 - 03/12 05 163 4.4195 0.0561 2 03/28 - 04/03 04 11 0.0002 0.0000 52 03/13 - 03/19 05 28 0.0115 0.0001 3 04/04 - 04/10 04 0 0.0000 0.0000 53 03/20 - 03/26 05 21 0.0177 0.0002 4 04/11 - 04/17 04 2 0.0000 0.0000 54 03/27 - 04/02 05 1 0.0000 0.0000 5 04/18 - 04/24 04 0 0.0000 0.0000 55 04/03 - 04/09 05 4 0.0044 0.0001 6 04/25 - 05/01 04 65 0.0253 0.0003 56 04/10 - 04/16 05 0 0.0009 0.0000 7 05/02 - 05/08 04 15 0.0082 0.0001 57 04/17 - 04/23 05 0 0.0000 0.0000 8 05/09 - 05/15 04 84 3.0819 0.0391 58 04/24 - 04/30 05 10 0.0070 0.0001 9 05/16 - 05/22 04 7 0.0022 0.0000 59 05/01 - 05/07 05 0 0.0000 0.0000 10 05/23 - 05/29 04 34 0.0358 0.0005 60 05/08 - 05/14 05 11 0.0168 0.0002 11 05/30 - 06/05 04 10 0.0032 0.0000 61 05/15 - 05/21 05 70 0.0244 0.0003 12 06/06 - 06/12 04 148 27.5206 0.3491 62 05/22 - 05/28 05 0 0.0013 0.0000 13 06/13 - 06/19 04 28 0.5022 0.0064 63 05/29 - 06/04 05 26 0.0121 0.0002 14 06/20 - 06/26 04 1 0.0000 0.0000 64 06/05 - 06/11 05 18 0.0131 0.0002 15 06/27 - 07/03 04 11 0.0087 0.0001 65 06/12 - 06/18 05 54 0.0207 0.0003 16 07/04 - 07/10 04 9 0.0054 0.0001 66 06/19 - 06/25 05 44 0.0535 0.0007 17 07/11 - 07/17 04 71 0.1167 0.0015 67 06/26 - 07/02 05 82 26.1808 0.3321 18 07/18 - 07/24 04 47 2.0055 0.0254 68 07/03 - 07/09 05 42 0.0181 0.0002 19 07/25 - 07/31 04 49 3.5854 0.0455 69 07/10 - 07/16 05 85 55.9625 0.7099 20 08/01 - 08/07 04 6 0.0019 0.0000 70 07/17 - 07/23 05 76 19.4168 0.2463 21 08/08 - 08/14 04 4 0.0026 0.0000 71 07/24 - 07/30 05 158 31.8284 0.4037 22 08/15 - 08/21 04 2 0.0025 0.0000 72 07/31 - 08/06 05 20 0.0339 0.0004 23 08/22 - 08/28 04 0 0.0000 0.0000 73 08/07 - 08/13 05 0 0.0000 0.0000 24 08/29 - 09/04 04 19 0.0193 0.0002 74 08/14 - 08/20 05 65 2.6287 0.0333 25 09/05 - 09/11 04 47 0.2825 0.0036 75 08/21 - 08/27 05 41 3.2290 0.0410 26 09/12 - 09/18 04 49 0.1065 0.0014 76 08/28 - 09/03 05 23 0.9814 0.0124 27 09/19 - 09/25 04 7 0.0034 0.0000 77 09/04 - 09/10 05 68 8.4481 0.1072 28 09/26 - 10/02 04 123 74.4510 0.9444 78 09/11 - 09/17 05 156 63.7886 0.8092 29 10/03 - 10/09 04 15 0.0041 0.0001 79 09/18 - 09/24 05 57 1.6021 0.0203 30 10/10 - 10/16 04 225 0.0084 0.0001 80 09/25 - 10/01 05 1 0.0017 0.0000 31 10/17 - 10/23 04 42 5.2813 0.0670 81 10/02 - 10/08 05 116 78.4945 0.9957 32 10/24 - 10/30 04 8 0.0030 0.0000 82 10/09 - 10/15 05 25 2.0283 0.0257 33 10/31 - 11/06 04 0 0.0000 0.0000 83 10/16 - 10/22 05 0 0.0000 0.0000 34 11/07 - 11/13 04 3 0.0000 0.0000 84 10/23 - 10/29 05 82 18.1444 0.2302 35 11/14 - 11/20 04 0 0.0000 0.0000 85 10/30 - 11/05 05 2 0.0000 0.0000 36 11/21 - 11/27 04 41 0.0422 0.0005 86 11/06 - 11/12 05 14 0.0026 0.0000 37 11/28 - 12/04 04 48 0.1967 0.0025 87 11/13 - 11/19 05 52 6.7994 0.0863 38 12/05 - 12/11 04 58 6.1520 0.0780 88 11/20 - 11/26 05 22 0.0000 0.0000 39 12/12 - 12/18 04 7 0.0035 0.0000 89 11/27 - 12/03 05 0 0.0000 0.0000 40 12/19 - 12/25 04 20 0.1403 0.0018 90 12/04 - 12/10 05 134 0.0778 0.0010 41 12/26 - 01/01 05 52 3.1056 0.0394 91 12/11 - 12/17 05 461 295.9821 3.7545 42 01/02 - 01/08 05 27 0.0080 0.0001 92 12/18 - 12/24 05 231 130.9021 1.6605 43 01/09 - 01/15 05 12 0.0051 0.0001 93 12/25 - 12/31 05 255 4.5602 0.0578 44 01/16 - 01/22 05 10 0.0013 0.0000 94 01/01 - 10/07 06 38 0.0883 0.0011 45 01/23 - 01/29 05 12 0.0036 0.0000 95 01/08 - 01/14 06 98 23.5530 0.2988 46 01/30 - 02/05 05 18 0.0030 0.0000 96 01/15 - 01/21 06 24 7.6853 0.0975 47 02/06 - 02/12 05 0 0.0000 0.0000 97 01/22 - 01/28 05 7 0.0021 0.0000 48 02/13 - 02/19 05 14 0.0041 0.0001 98 01/29 - 02/04 06 45 0.2591 0.0033 49 02/20 - 02/26 05 0 0.0000 0.0000 TOTAL 4515 952.15 12.08 50 02/27 - 03/05 05 1 0.0002 0.0000

101

Appendix Table C-5. Weekly rainfall, runoff and soil loss from the oil palm plot collected from March 2004 – February 2006.

Rainfall Runoff Soil loss Rainfall Runoff Soil loss Week Date -1 Week Date -1 (mm) (mm) (t·ha ) (mm) (mm) (t·ha ) 1 03/21/ - 03/27 04 11 3.4858 0.0650 51 03/06 - 03/12 05 38 0.1352 0.0025 2 03/28 - 04/03 04 0 0.0000 0.0000 52 03/13 - 03/19 05 16 0.0121 0.0002 3 04/04 - 04/10 04 0 0.0000 0.0000 53 03/20 - 03/26 05 0 0.0000 0.0000 4 04/11 - 04/17 04 14 0.0914 0.0017 54 03/27 - 04/02 05 0 0.0000 0.0000 5 04/18 - 04/24 04 0 0.0000 0.0000 55 04/03 - 04/09 05 0 0.0000 0.0000 6 04/25 - 05/01 04 0 0.0000 0.0000 56 04/10 - 04/16 05 0 0.0000 0.0000 7 05/02 - 05/08 04 15 0.0656 0.0012 57 04/17 - 04/23 05 0 0.0000 0.0000 8 05/09 - 05/15 04 45 0.0266 0.0005 58 04/24 - 04/30 05 14 0.0219 0.0004 9 05/16 - 05/22 04 0 0.0000 0.0000 59 05/01 - 05/07 05 4 0.0075 0.0001 10 05/23 - 05/29 04 48 6.3983 0.1193 60 05/08 - 05/14 05 0 0.0000 0.0000 11 05/30 - 06/05 04 21 0.6094 0.0114 61 05/15 - 05/21 05 0 0.0000 0.0000 12 06/06 - 06/12 04 163 68.3827 1.2747 62 05/22 - 05/28 05 8 0.0087 0.0002 13 06/13 - 06/19 04 4 0.0050 0.0001 63 05/29 - 06/04 05 123 6.6687 0.1243 14 06/20 - 06/26 04 0 0.0000 0.0000 64 06/05 - 06/11 05 0 0.0000 0.0000 15 06/27 - 07/03 04 7 0.0042 0.0001 65 06/12 - 06/18 05 0 0.0000 0.0000 16 07/04 - 07/10 04 26 0.0164 0.0003 66 06/19 - 06/25 05 34 0.0563 0.0010 17 07/11 - 07/17 04 103 24.4512 0.4558 67 06/26 - 07/02 05 115 4.1839 0.0780 18 07/18 - 07/24 04 58 20.5123 0.3824 68 07/03 - 07/09 05 8 0.0365 0.0007 19 07/25 - 07/31 04 22 0.0324 0.0006 69 07/10 - 07/16 05 5 0.0141 0.0003 20 08/01 - 08/07 04 86 0.1129 0.0021 70 07/17 - 07/23 05 0 0.0000 0.0000 21 08/08 - 08/14 04 27 0.0783 0.0015 71 07/24 - 07/30 05 43 0.1042 0.0019 22 08/15 - 08/21 04 0 0.0000 0.0000 72 07/31 - 08/06 05 21 0.0542 0.0010 23 08/22 - 08/28 04 0 0.0000 0.0000 73 08/07 - 08/13 05 9 0.0107 0.0002 24 08/29 - 09/04 04 41 1.5722 0.0293 74 08/14 - 08/20 05 49 0.0641 0.0012 25 09/05 - 09/11 04 15 1.5702 0.0293 75 08/21 - 08/27 05 28 0.0177 0.0003 26 09/12 - 09/18 04 12 0.0137 0.0003 76 08/28 - 09/03 05 0 0.0000 0.0000 27 09/19 - 09/25 04 14 0.0273 0.0005 77 09/04 - 09/10 05 15 0.0320 0.0006 28 09/26 - 10/02 04 41 4.5612 0.0850 78 09/11 - 09/17 05 53 0.1084 0.0020 29 10/03 - 10/09 04 0 0.0000 0.0000 79 09/18 - 09/24 05 616 1.7117 0.0319 30 10/10 - 10/16 04 0 0.0000 0.0000 80 09/25 - 10/01 05 21 0.0295 0.0005 31 10/17 - 10/23 04 0 0.0000 0.0000 81 10/02 - 10/08 05 94 7.2428 0.1350 32 10/24 - 10/30 04 0 0.0000 0.0000 82 10/09 - 10/15 05 41 0.0763 0.0014 33 10/31 - 11/06 04 0 0.0000 0.0000 83 10/16 - 10/22 05 3 0.0021 0.0000 34 11/07 - 11/13 04 0 0.0000 0.0000 84 10/23 - 10/29 05 80 1.1653 0.0217 35 11/14 - 11/20 04 0 0.0000 0.0000 85 10/30 - 11/05 05 91 1.5374 0.0287 36 11/21 - 11/27 04 55 0.0288 0.0005 86 11/06 - 11/12 05 5 0.0102 0.0002 37 11/28 - 12/04 04 24 0.0116 0.0002 87 11/13 - 11/19 05 335 7.1922 0.1341 38 12/05 - 12/11 04 32 0.0238 0.0004 88 11/20 - 11/26 05 0 0.0000 0.0000 39 12/12 - 12/18 04 28 2.7504 0.0513 89 11/27 - 12/03 05 31 4.3892 0.0818 40 12/19 - 12/25 04 29 0.0387 0.0007 90 12/04 - 12/10 05 34 0.0700 0.0013 41 12/26 - 01/01 05 77 27.2672 0.5083 91 12/11 - 12/17 05 391 19.6032 0.3654 42 01/02 - 01/08 05 29 0.0404 0.0008 92 12/18 - 12/24 05 521 29.1232 0.5429 43 01/09 - 01/15 05 17 0.0169 0.0003 93 12/25 - 12/31 05 34 0.0510 0.0010 44 01/16 - 01/22 05 9 0.0096 0.0002 94 01/01 - 10/07 06 10 0.0103 0.0002 45 01/23 - 01/29 05 7 0.0054 0.0001 95 01/08 - 01/14 06 75 3.1787 0.0593 46 01/30 - 02/05 05 26 0.0226 0.0004 96 01/15 - 01/21 06 703 11.7417 0.2189 47 02/06 - 02/12 05 0 0.0000 0.0000 97 01/22 - 01/28 05 0 0.0000 0.0000 48 02/13 - 02/19 05 0 0.0000 0.0000 98 01/29 - 02/04 06 271 4.0602 0.0757 49 02/20 - 02/26 05 0 0.0000 0.0000 TOTAL 5044 264.9639 4.9392 50 02/27 - 03/05 05 0 0.0000 0.0000

102 Appendix Table C-6. Average daily discharge and rainfall records from the Bugsok AWaS and AWeS. Month May-04 Jun-04 Jul-04 Aug-04 Sep-04 Oct-04 Nov-04 Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain Day (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) 1 2.442 0.911 2.1 2.060 52.5 2 2.396 0.833 1.573 1.4 0.8 3 1.083 2.369 0.3 0.848 1.200 1.4 4 1.085 2.234 0.3 0.850 1.137 1.7 5 1.080 1.717 10.0 1.963 0.805 1.056 2.0 6 1.075 1.736 2.000 5.7 0.786 1.015 3.0 7 1.747 2.042 0.991 25.3 1.003 6.3 8 1.657 1.971 7.5 1.104 2.0 1.103 1.2 9 1.328 1.612 2.012 0.960 1.069 10 1.508 1.668 26.5 1.923 0.6 0.6 1.008 1.183 11 1.970 10.7 1.885 5.9 0.980 1.307 12 1.527 2.316 26.8 1.824 10.5 0.8 0.979 0.6 1.407 0.6 13 1.413 2.525 41.7 1.740 1.0 1.660 5.6 1.485 14 1.245 5.5 0.890 1.873 2.2 1.533 15 1.212 20.4 0.851 1.2 6.0 1.309 0.3 1.573 3.4 16 1.247 0.5 2.1 0.882 1.2 18.8 1.162 1.380 17 1.254 10.3 0.893 0.9 0.3 1.128 4.4 1.311 18 1.222 5.5 82.8 0.861 0.3 1.431 10.2 1.325 19 1.160 2.420 6.4 0.843 1.2 1.391 1.465 0.3 20 2.386 1.8 0.862 18.8 1.455 1.122 21 2.307 2.9 0.851 1.693 1.137 13.6 22 3.550 8.7 0.817 0.869 1.300 1.8 1.169 2.3 23 3.762 43.8 0.813 0.854 1.197 2.0 1.208 24 3.421 14.8 0.794 0.853 1.7 1.168 10.2 1.175 25 3.747 1.2 0.807 0.845 1.139 1.191 20.8 26 3.080 0.811 0.898 4.9 1.121 1.310 11.4 27 2.818 11.3 0.787 1.258 1.098 3.3 1.211 28 10.9 3.043 2.4 0.782 1.255 24.3 1.066 1.3 1.236 3.3 29 0.6 3.064 22.9 0.854 44.7 13.8 0.5 1.278 30 3.6 2.912 0.3 0.901 4.6 1.314 31 2.584 0.910 13.1 0.3 Appendix Table C-6… continuation… Month Dec-04 Jan-05 Feb-05 Mar-05 Apr-05 May-05 Jun-05

103 Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain Day (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) 1 1.457 23.1 2.733 0.9 4.9 0.839 1.1 1.001 1.432 2 1.313 0.3 2.499 0.6 0.841 0.986 1.108 3 1.146 2.756 6.9 0.3 0.831 0.986 1.208 4 1.119 3.055 6.1 0.843 0.917 1.050 5 1.072 2.372 0.3 1.3 0.865 0.9 0.905 0.959 6 1.090 2.835 4.9 0.902 0.900 0.945 7 1.202 26.7 3.131 0.6 1.595 0.900 0.920 8 1.686 15.2 3.453 7.7 3.750 0.893 0.897 1.8 9 2.807 15.4 3.635 1.714 0.893 0.905 10 10 2.105 6.1 3.638 1.322 0.882 0.905 13.7 11 1.985 1.4 3.843 14.4 1.182 0.891 0.940 12 2.007 4.822 1.106 0.901 0.905 13 2.367 0.6 3.980 1.078 0.997 0.924 0.906 9.3 14 2.249 1.4 2.812 7.6 1.044 0.987 0.904 0.982 4.8 15 2.163 5.1 4.018 0.3 1.025 0.979 0.909 0.942 16 3.068 25.1 4.058 0.317 1.5 1.023 0.962 0.892 0.926 1.9 17 3.206 0.3 3.196 0.314 1.020 0.965 0.889 0.957 5.4 18 2.567 3.356 0.311 1.016 0.971 0.908 0.973 1.8 19 2.626 2.7 3.913 0.317 1.013 0.963 0.894 0.946 20 2.654 2.564 0.312 1.481 0.959 0.907 0.941 21 6.0 1.010 2.8 0.295 1.463 0.960 0.901 22 5.8 1.017 5.1 0.291 1.226 0.961 0.899 0.8 23 1.014 0.3 0.294 1.160 0.955 0.897 2.4 24 6.5 1.003 0.290 1.108 1.203 0.893 19.6 25 9.1 1.014 4.9 0.293 1.084 1.710 0.898 33.2 26 39.3 1.042 0.296 1.060 1.556 0.900 0.3 27 1.5 1.016 0.281 1.052 1.219 0.886 18.9 28 1.4 1.003 0.274 1.050 0.988 7 29 0.971 1.7 1.023 0.975 3.5 30 0.982 1.021 0.937 2.2 31 6.6 1.045

104 Appendix Table C-6.. continuation… Month Jul-05 Aug-05 Sep-05 Oct-05 Nov-05 Dec-05 Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain Day (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) 1 28.1 1.835 0.976 32.9 1.257 3.6 10.7 0.960 3.9 2 37.9 1.626 1.002 1.198 0.3 0.3 0.961 11 3 0.3 1.467 1.186 32.7 0.3 0.975 10.1 4 1.473 0 1.162 0.975 0.3 5 1.282 3.6 1.154 1.4 0.6 0.963 6 0.6 1.191 0 0.6 1.183 32.4 0.945 7 1.116 2.8 8.4 1.212 2.8 0.935 2 8 1.291 17.1 1.081 0 0.3 1.163 4.3 0.964 6.9 9 1.181 1.5 1.048 0 5.1 1.175 1.9 0.971 0.5 10 1.136 0.3 0.983 0 8 1.121 0.952 47.9 11 1.053 0.960 0.3 1.180 0.3 16.2 2.005 43.2 12 0.995 0.961 0 57.8 1.554 4.1 0.9 1.362 8.5 13 0.984 11.4 0.964 5.7 34.3 1.526 3.2 1.198 3.2 14 1.018 0.3 1.005 24.2 1.420 16.2 1.541 15.2 15 0.939 1.211 0.6 1.558 0.3 2.076 21.3 1.508 27.5 16 0.896 1.069 0 1.421 1.733 1.2 4.545 69.5 17 0.881 4.9 0.997 18.4 1.268 1.452 2.197 1 18 0.873 0.3 0.962 0 1.212 21.1 1.385 1.644 11.8 19 0.845 0.997 4.8 1.396 16.2 1.290 0.3 1.378 20 0.856 0.980 2.9 1.565 3.8 1.239 1.4 1.255 21 0.835 0.966 9.3 1.890 3.3 1.198 1.8 1.1 1.207 14 22 0.826 1.2 1.017 0 1.605 4.7 1.261 18.9 1.5 1.796 51.6 23 0.845 1.2 1.025 0 1.346 0.3 3.829 40.1 24 0.888 27.5 0.976 9.5 1.201 0.970 0.3 2.809 10.7 25 1.615 11.3 1.219 2 1.154 3.6 1.017 9.8 3.797 89.1 26 1.511 4.4 1.176 4.5 1.109 45.8 1.067 1.2 4.419 13.9 27 1.291 1.103 0 1.111 7.4 8.2 1.014 1.4 2.251 0.3 28 1.139 0.6 1.096 0 1.567 33.7 4.2 0.991 0.3 1.754 29 1.082 3 1.054 0 1.531 14.9 6.1 0.981 1.569 25.5 30 1.443 1.018 0 1.400 0.969 1.612 4.3 31 3.713 0.996 3.7 1.416

105 Appendix Table C-7. Average daily discharge and rainfall records from the Pamacsalan AWaS and AWeS. Month May-04 Jun-04 Jul-04 Aug-04 Sep-04 Oct-04 Nov-04

Day Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) 1 1.201 3.431 1.7 1.852 44.5 2 1.185 1 1.313 0.9 3 1.167 0.9 1.311 0.6 4 1.151 1.203 2.116 1.295 2.2 5 1.045 1.262 4 2.101 1.225 3.9 6 0.980 1.294 0.3 2.281 7.2 1.059 1.8 7 1.186 1.227 2.408 0.3 10.2 3.9 8 3.772 1.225 2.220 5.6 0.9 1.2 9 6.059 2.148 1.277 10 2.853 3.4 1.959 18.1 1.274 11 1.512 1.692 14.4 2.160 4 2 1.340 12 1.413 2.600 0.3 8.5 0.9 0.3 1.312 13 1.183 3.225 1.513 44.5 0.9 0.3 1.285 14 1.062 1.466 2.280 4.1 6.555 1.306 15 1.127 1.569 3.214 16.6 1.473 0.3 6 4.705 16 1.188 1.484 2.738 12.2 1.530 18.2 3.694 0.3 17 1.195 1.256 1.800 11.3 1.445 0.6 5 3.767 18 1.194 3.5 10.374 64 1.506 0.3 5.765 1.366 19 1.176 12.913 9.6 1.654 0.9 4.927 1.377 20 1.127 3.6 1.540 0.3 0.3 2.943 1.5 21 1.128 0.9 1.197 1.112 22 1.119 5.8 1.183 1.125 23 0.3 29.1 1.106 24 6.9 1.146 3.327 25 4.3 1.160 2.965 1.621 1.2 26 4.7 2.769 1.589 0.6 27 4.176 11.5 1.381 2.763 1.423 28 3.1 4.174 2.6 1.372 0.6 2.795 1.574 0.3 29 1.9 6.502 17.9 1.077 5.5 2.839 1.511 30 4 5.371 1.114 2.4 0.9 1.449 31 3.801 1.4 2.847 25.9

106 Appendix Table C-7. continuation… Month Dec-04 Jan-05 Feb-05 Mar-05 Apr-05 May-05 Jun-05

Day Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) 1 1.673 1.245 3.6 1.133 7.4 1.217 2.1 nd nd 2 2.500 1.243 1.8 1.003 0.6 1.112 4 1.200 3 1.386 0.3 1.229 1.2 1.049 1.079 1.205 1.5 4 1.353 1.5 1.122 1.047 1.224 0.3 5 1.385 9.9 1.134 0.3 1.105 1.261 1.8 1.8 6 1.255 8.7 1.166 1.036 16.9 1.278 7 1.196 18.5 1.172 5.266 17.4 1.279 2.6 8 1.303 18.3 1.183 9.9 9 6.128 1.338 0.3 1.213 6.093 6 0.3 10 2.482 1.110 0 2.053 3.6 0.3 11 1.654 1.380 18.3 1.561 2.1 5.4 12 1.306 1.242 0 1.207 1.5 0.6 13 1.445 0.3 1.090 0 1.065 0.6 1.409 1.328 14 3.637 1.028 0 10 1.065 0.6 1.443 0.3 1.341 15 0 1.213 0.3 1.147 0.6 1.479 1.362 16 0 1.350 3.6 1.269 0.3 1.512 1.369 17 0 1.282 0.6 1.422 0.3 1.387 18 1.309 0 1.174 1.278 1.398 19 1.191 1.301 0 1.072 0.3 1.382 20 1.147 1.240 0 1.073 0.3 1.376 21 1.109 1.198 1.4 1.119 0.3 1.826 22 1.451 1.342 3.9 1.134 0.3 1.150 23 0.9 1.398 0.3 0.3 2.1 1.118 24 0.3 0.962 0 1.085 0.3 0.6 1.127 25 0.3 1.010 7 1.098 0.3 1.126 26 0.3 0.957 0 1.097 0.3 1.139 27 0.6 0.957 0 1.104 0.3 1.157 28 0.6 0.958 0 1.008 0.3 0.3 1.157 29 1.8 0.996 1.9 1.157 30 1.239 1.2 1.016 0 1.256 31 1.262 6.3 1.095 11.3 1.225

107 Appendix Table C-7. continuation… Month Jul-05 Aug-05 Sep-05 Oct-05 Nov-05 Dec-05

Day Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain Discharge Rain (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) (m3/s) (mm) 1 no data 2.2 0.659 36.6 0.500 6.4 1.555 8.1 0.387 3.8 2 0.434 0.3 0.429 0.3 1.491 0.450 16 3 2.8 0.395 0.541 21.5 1.053 1.013 27.6 4 2.351 0.9 0.381 0.520 0.9 0.879 0.620 5 2.026 1.6 0.381 0.421 0.9 0.791 0.464 6 1.924 0.382 1.2 0.763 58.1 0.780 0.3 0.459 2.4 7 1.651 0.3 0.385 5.4 1.037 1.8 0.767 0.490 1.1 8 1.622 0.3 0.388 0.718 2 0.757 0.570 8.7 9 1.617 1.3 0.382 2.6 0.685 6.6 0.751 0.510 1.3 10 1.617 0.484 8 0.669 0.766 0.516 48.6 11 1.615 0.565 17.6 0.575 2.1 1.240 6.4 2.138 36.7 12 1.616 0.732 0.778 19.7 1.057 0.6 1.006 8.6 13 3.467 9.7 2.230 1.030 9.7 0.788 2.7 0.738 0.9 14 4.057 1.8 1.069 0.699 0.833 9.6 1.026 10.3 15 7.649 7 0.682 0.630 1.4 0.985 18 1.024 20.6 16 2.479 0.503 0.602 0.3 1.942 1.2 3.141 89.7 17 2.183 0.402 0.479 0.973 2.498 18 13.167 0.457 20.4 0.461 2.9 0.933 2.498 12.1 19 17.904 0.966 16.7 0.558 3 0.913 0.3 2.498 0.3 20 5.682 0.760 3.1 0.474 0.919 2.498 21 4.051 1.146 3.2 0.429 1.059 3.8 1.809 11.6 22 8.048 0.679 3.2 0.407 3.7 1.023 2 1.295 44.4 23 3.714 0.470 0.392 0.885 0.6 2.077 40.7 24 2.459 0.393 0.365 0.3 0.872 0.8 1.648 12 25 5.322 0.379 0.350 2.6 0.943 8.9 2.219 85.5 26 6.541 0.398 1.9 0.543 39.6 0.912 2.4 2.206 19.4 27 3.783 0.398 1.7 0.942 12.1 0.911 0.3 1.253 0.3 28 2.957 0.847 9 0.780 4.6 1.014 0.6 0.948 29 48 2.918 0.809 2.2 0.849 10.6 1.033 0.815 12.1 30 94 2.911 0.588 1.6 0.782 1.017 1.027 4.9 31 14.9 2.904 0.530 0.3 0.762

108

109