Rapid mapping of hydraulic conductivity in tropical watershed

Abstract

Modeling of water and solute dynamics in soils is a major focus in watershed management. In order to model the water flow, soil hydraulic properties have to be known. Spatial information on basic soil data such as particle-size distribution, organic carbon content and bulk density are often available. However, spatial information on soil hydrologic data, which are necessary for soil transport models, is usually not available. Soil hydrologic data are expensive and difficult to obtain on a large area and are often predicted from physical data like particlesize distribution using pedotransfer functions. This study set out to develop a rapid methodology based on relating soil hydraulic conductivity to soil spectral reflectance and using a soil spectral library approach to predict soil hydraulic conductivity in a large area of a watershed. Soil spectral reflectance was correlated to field saturated hydraulic conductivity and the correlation found to be consistent Ks (index of agreement, d = 0.87, and r2 =0.63). This consistent relationship provided a foundation for using a spectral calibration model in conjunction with a georeferenced library to predict surface Ks over a large area. The spectral library contained spectral data for the points in the study area for which direct measurement was not made. In the build up of a map for the whole watershed, spatial variability of surface Ks was studied using classical and geostatistical concepts. Semivariograms were intensively used with actual and kriged estimates to produce an accurate surface interpolation map of surface K, in the watershed. The map, which shows a general trend of surface K, with altitude, will be a great asset for the hydrologic modeling of watershed responses.