Comparison of GIS Based Models for Estimating Monthly Global Overland Flow

Zhang, X., Drake, N.A., Wainwright, J. and Mulligan, M.
Department of Geography, King's College London, Strand, London WC2R 2LS, United Kingdom

Large scale models of monthly overland flow are important for understanding and managing the global hydrologic cycle, agriculture, soil erosion and environmental change. GIS is a logical framework within which to develop such models. To estimate overland flow at this scale there are two important conditions. One is that the selected hydrological models are relatively simple, and the other is that the required model parameters are available or can be calculated from available data. Data from the Global Ecosystem Database and the World Weather Guide are used here.

Two models are implemented, developed and compared, the empirically based SCS model, and the physically based model of Carson & Kirkby (1972, 1976). The SCS model is based on individual rainfall events and has been converted to work with monthly averages by assuming a exponential rainfall distribution. The Carson and Kirkby model is implemented in its original form and developed in three stages of increasing complexity to include, root depth, evapotranspiration, and subsurface drainage. In the models most complex form overland flow is generated when rainfall exceeds the soil water storage. This storage is controlled by vegetation and soil types and the initial soil moisture. To calculate the initial soil moisture a hydrological cycle among rainfall, overland flow, actual evapotranspiration, infiltration and subsurface flow from January to December has been set up. The initial soil water content in each month is adjusted iteratively in each pixel until equilibrium starting conditions are attained. The basis of this model is:

OF(i)=R(i)e -(rc - Q(i))/ro(i)
SF(i)=DIF(i)/N(i)+Q(i)-FC               (DlF(i)/N(i)+Q(i)>=FC)
SF(i)=0                                 (DlF(i)/N(i)+Q(i)<FC)
Q(i+l)=Q(i)+DlF(i)/N(i)-SF(i) -AET(i)   (0<=Q(i+l)<=FC)

Where N(i) is number of rain days per month, ro(i) is mean rain intensity per rain day each month (mm/day), R(i) is monthly total rainfall (mm), Q(i) is total initial soil water (mm), rc is water storage capacity (mm), OF(i) is monthly overland flow (mm), DlF(i) is amount of monthly infiltration (mm), i is the i-th month, SF(i) is mean subsurface flow in a daily rainfall (mm), FC is field capacity (mm), and AET(i) is actual evapotranspiration between two rainfall events (mm). Several submodels for estimating parameters are used. The Brutseart-Stricker model is selected to calculate AET. The soil water storage capacity is substituted by using soil storage capacity in the rooting zone, which is computed on the basis of the soil properties in different soil horizons. The subsurface flow is controlled by infiltration and field capacity of the different horizons. All the model and submodels are written in C++ and are integrated with GIS software.

Monthly overland flow has been estimated for Eurasia and North Africa to test the effectiveness of the models in various climate zones. Substantial variations in overland flow are found between many of the models particularly in semi-arid environments. Runoff coefficients from several catchments are currently being used to validate the findings and determine a method that is robust globally.