Predicting Forest Soil Hydraulic Properties for Hydrological Modeling Using Digital Soil Mapping and Existing Soil Data.

The development of distributed hydrological models has brought to focus the need for spatially explicit soil hydraulic properties. However, most of the soil hydraulic properties are measured at point scale, which does not support the needs for distributed models. Distributed Hydrology Soil Vegetation Model (DHSVM) developed for forested catchment in Pacific Northwest is one of the models used for predicting various hydrological processes such as stream flow, soil moisture and water table depth.

Saturated hydraulic conductivity (K_sat) is one of the most important soil hydraulic property model inputs as it controls the rate of soil water movement with direct impacts on stream flow. Field measurements of K_sat are expensive with highly variable results mostly due to preferential flow. In this study we combined field measured K_sat at point, hillslope and small catchment scales with digital soil mapping techniques and existing soil data to generate continuous inputs for DHSVM to predict stream flow for a HUC12 watershed. The study site is located in southern Indiana and is part of Crawford Upland physiographic sub region of high relief bedrock hills with slopes varying from 2 to 60%. The watershed was once dominated by native broadleaf forest which currently occupies only 1/3 occurring mostly on steep slopes with the remaining 2/3 divided between permanent pasture on broad flat summits and croplands in valleys and floodplains. The DHSVM captured hydrograph rise, peaks and falling limbs for the majority of precipitation events. Daily hydrographs of observed and simulated stream discharge showed a good performance as indicated by the Nash-Sutcliffe (N-S) overall model efficiency of 0.52 without any model calibration. However, the N-S varied from 0.32 during dry season to 0.72 for winter season due to the variability in soil moisture deficit and evapotranspiration. The results demonstrate the utility of combining DSM techniques with field measurements and existing soil information in improving streamflow predictions that could be used for flood forecast and mitigation.