12t'00'E 122'00"E 
T T 
  
WOON 
T 
We 
L 
TAN 
  
  
  
Mountain Province 
1000"N 
17°0'0"N 
Baguio Cy. Sendhet 
  
  
  
2 N Legend z 
= | Weather Stations LS 
R & Bangag Slaion 5 
0 1020 40 60 8 gie Riesr Nafmek 
Pr ERI Coens 
i 1 
T2 00E 122°00°E 
Figure 1. The Cagayan River Basin, its provinces, the Bangag 
Station, and weather stations. 
2.4. Land use/Land cover: About 37% of the area is covered 
by forest while grassland, agricultural area, and other land use 
such as settlement and water area occupies 34%, 27% and 2%, 
respectively. Of the 741,000 hectares of agricultural area, 94% 
are crop fields while the rest are fruit trees. The crop fields are 
further subdivided into 68% paddy field, 22% corn field and 
10% upland crop field (DPWH & JICA 2001). 
3. MODEL DESCRIPTION 
The Soil and Water Assessment Tool (SWAT) model is a river 
basin scale model specifically used in predicting the effect of 
land management practices, over long periods of time, on 
variables such as flow and sediment to areas of varying soils, 
land use and management conditions. It is physically based, 
uses readily available inputs, computationally efficient and 
enables users to study long-term impacts (Neitsch et. al, 2005). 
The study used the SWAT 2005 version via the ArcSWAT 
interface for ArcGISTM, 
SWAT is a continuous time model and is not designated to 
simulate detailed single-event flood routing (Neitsch et. al, 
2005) and operates on a daily time step (Hao et. al, 2003). To 
predict surface runoff yield, the model uses a modified version 
of the SCS CN method (USDA-SCS, 1972): 
Q - (R-25Yy /(R 0.83) R>025 (1) 
Q-0 Rs028  (Q) 
where Q and R are the daily surface runoff and daily rainfall, 
respectively, both in mm H5O. S is a retention parameter which 
varies spatially under various soil, land use, management and 
slope conditions, and temporally to respond to changes in soil 
water content (Hao et. al, 2003). The retention parameter is 
related to the curve number (CN) and defined as: 
S= 254 120-10) 
GB) 
The model estimates erosion and sediment yield from each sub- 
basin using the Modified Universal Soil Loss Equation 
(MUSLE) (Williams, 1995): 
sed =11.8-(0,.., 4,2 area, -K-C-P-LS-CFRG (4) 
where sed is the sediment yield on a given day in metric tons, 
Osur is the surface runoff volume in mmH,0/ha, Goer 18 the 
peak runoff rate in m°/s, areapy, is the area of the HRU in ha, K 
is the USLE soil erodibility factor, C is the USLE cover and 
management factor, P is the USLE support practice factor, LS 
is the USLE topographic factor and CFRG is the course 
fragment factor. For a detailed description of these variables, 
the reader may refer to the theoretical documentation of SWAT 
2005 (Neitsch et. al, 2005). 
4. METHODOLOGY 
The general procedure used in the study can be seen in Figure 
1. Several datasets are inputted to the model after data 
preparation. The basin is automatically delineated via the 
SWAT model interface (ArcSWAT) using the input DEM 
while subbasins and finer subdivisions in the basin called the 
hydrologic response units (HRU) are defined by setting 
threshold limits for land use/land cover, soil type and slope 
class. Available flow and sediment data were used to calibrate 
and validate the model. The calibrated model was then rerun 
for five scenarios, the results of which are compared and 
became the basis of the author’s final analysis. 
4.1. Data Preparation, Input and Model Setup 
4.1.1. Digital Elevation Model: The Shuttle Radar Topography 
Mission Digital Elevation Model (SRTM-DEM) was used for 
the DEM requirement of SWAT. SRTM data are products of 
processed raw radar signals spaced at different intervals at the 
Jet Propulsion Laboratory (JPL) (USGS, 2011). The DEM 
used was a 3 arc-second medium resolution elevation data 
(approx. 90 m) resampled using cubic convolution interpolation 
and downloaded at the EarthExplorer website (URL: 
http://edcsns17.cr.usgs.gov/NewEarthExplorer). The DEM was 
used to generate percent slope values, to do automatic 
watershed delineation and in defining stream networks and 
gage outlets. 
4.1.2. Land Cover/Land Use Map was generated from two 
sets of Landsat 7 TM and ETM+. CRB covers three Landsat 
scenes within the coverage of path 116, rows 47, 48 and 49. 
These satellite imageries passed through a processing scheme 
as shown in Figure 2. All downloaded Landsat ETM+ images 
were Level 1T products in Geographic Tagged Image-File 
Format (GeoTIFF). These were geo-referenced to include 
terrain correction that corrected parallax error from local 
topographic relief with a digital elevation model (Helmer & 
Ruefenacht, 2007). 
A simple atmospheric correction called the Dark Object 
Subtraction (DOS) technique was applied to the set of final 
images (reference images) covering the CRB before they were 
processed for cloud and cloud shadow masking and filling 
method developed by Martinuzzi et. al (2006). Image 
classification was done in per scene basis. The five general 
land cover classes used in the study are forest, vegetation, water 
bodies, bare soil and built-up areas. There were four supervised 
classifiers (Maximum likelihood, Parallelepiped, Minimum 
Distance and Mahalanobis Distance) and two unsupervised 
classifiers (ISODATA and K-means) that were tested to 
   
    
     
  
  
  
   
    
  
  
   
  
  
    
     
    
   
   
    
  
   
    
   
   
   
    
   
   
   
   
     
   
   
   
   
  
  
   
   
    
  
    
     
     
    
   
     
PRE 
L- 
  
classify 
final cla 
the high 
to a vali 
least 85 
value f 
situatior 
random 
classifie 
ENVITM 
The of 
  
  
  
  
  
  
  
  
  
  
  
  
4.1.4. 
in Cag 
station 
temper 
Philip; 
Servic 
station 
and t 
Auror 
(http:/ 
are sh 
forma 
4.1.5. 
chang 
scena