Generalization and Testing of a Soil Moisture Budget for Different Drainage Conditions

Abstract

An empirical site-specific water balance model was generalized to account for cropland drainage effects on soil moisture and evapotranspiration (ET). In predicting soil moisture for well-drained (WD) soils, usually total water use is equated to infiltrated precipitation, stored soil moisture, and ET. In much of the eastern U.S. Corn Belt, however, row crop production is on soils which are poorly drained (PD) and underlain with perched water tables. These provide an additional source of soil water, capillary flow (C) into the crop root zone. Consequently, ET from PD soils is usually greater than that from WD soils. At West Lafayette, Indiana, from 1970 through 1974, for a PD soil the shallow water table furnished about a fourth of the total water used by late-planted corn (Zea mays L.) and a fifth of the water used by early-planted corn. A soil moisture budget model accounting for shallow water table influences, developed experimentally for a tile-drained Typic Argiaquoll soil, was generalized for use with other PD soils and, by voiding the capillary component, also for WD soils. For PD soils the C component was estimated as a function of the depth (G) to the shallow water table and a relative soil moisture gradient, or deficit, 1 ? PAV, where PAV is the fraction of plant available soil water in the corn root zone. In turn, changes in depth of the water table were predicted as a function of C and G, assuming a fixed basal leakage. For both PD and WD soils, daily ET was estimated and the pattern of relative soil moisture extraction from each 15 cm soil layer was established as a function of the relative soil moisture deficit in the top 30 cm and the age of the corn crop. The generalized simulation of the soil moisture balance (SIMBAL) model provided excellent results when tested on independent experimental data. It provided reasonable agreement when tested over a large area, using measurements taken at 108 location-dates in July and August in Indiana, Illinois, Minnesota and Nebraska in 1978-80, in cooperation with the Control Data Corporation AGSERV project. Scatter in the predicted versus measured PAV comparisons was attributed mainly to sampling errors in estimating the daily precipitation and pan evaporation inputs needed for the model predictions of soil moisture.