b. Rain water that fail to infiltrate moves up to velocities 500x that of 
infiltrated water. 
c. The antecedent soil moisture conditions plays a dominant role in 
determining areas where surface runoff will originate. 
d. Areas with swamps, bogs, etc. have the potential to saturate quickly 
and produce overland flow. 
e. Subsurface contributions to storm flow are highly dependent on many 
geologic and topographic factors. 
The method requires the following steps: 
a. Identification of physical factors. 
b. Assigning weights to factors. 
c. Risk probability analysis. 
IDENTIFICATION OF PHYSICAL FACTORS 
Classification of Images 
Land cover data may be derived from satellite products in two ways, 
frcm the imagery and fran the CCT's. For this type of study, a computer- 
enhanced 1:250000 satellite images of false-oolor composites are used. 
Classification is the critical step since it is here that pattern 
recognition algorithms are applied to the preprocessed data in order to 
generate useful land cover information. 
Digital napping techniques are based upon the assumptions that 
different land covers have different spectral properties. By applying 
certain pattern recognition algorithms to satellite data the various 
pixels may be statistically separated into clusters or classes defined by 
their spectral properties. This require the usage of two or more band. 
Due to the limitation of hardware used, a three spectral bands of image 
are used. To determine these classes, training site data must be 
established either through the use of supervised or unsupervised 
clustering technique. These data are employed to set the spectral limits 
for the various classes. 
The nBximum likelihood pattern recognition method of the minimum 
distance to mean was used. This particular algorithm calculates the mean 
distance to the mean for each pixel. The distance is computed by squaring 
the difference between the pixel value and the class mean, and then 
deviding the amount by the standard deviation. Due to the limited 
capability of the hydrological model used, only four land use pattern are 
classified. These are: 
a. Forest. 
b. Grass. 
c. Urban. 
d. Bare. 
670