Use of limited soil property data and modeling to estimate root zone soil water content

Abstract

Estimation of soil water content in the root zone with time in different parts of a watershed is important for both strategic and tactical management of water resources, as well as of agricultural production, water quality, and soil resources. This estimation requires detailed knowledge of rainfall intensities and meteorological variables over space and time, as well as the physical and hydraulic properties of the soil horizons and plant growth information. However, all this detailed spatial information is extremely expensive and time consuming to obtain. New technologies are helping to increase the spatial sampling of rainfall and other meteorological variables, but spatially detailed measurement of soil properties is still not practical. The best we can obtain from the existing soil survey database is the spatial distribution of soil textural class. We investigated the use of a hierarchy of limited soil input data, ranging from soil textural class of soil horizons alone, to measured soil texture and bulk densities of horizons, additional lab or field measurement of −33 kPa soil water content, to additional field measurement of average saturated hydraulic conductivity. These five modeling scenarios, along with meteorological and plant information, were input to the Root Zone Water Quality Model (RZWQM) to estimate 0–60 cm soil water content over a 30-day period in 1997 at the Little Washita River Experimental Watershed in Oklahoma. The estimated water contents were compared with time-domain reflectometry (TDR) profile measurements and gravimetric samplings of soil surface moisture. In addition to the five scenarios using limited input data, a more detailed set of data based on laboratory measured soil water retention curves and field measured saturated conductivity was supplied to the model for all Brooks–Corey function parameters (full description mode). Estimates of root zone soil water content using detailed input were compared to estimates obtained using minimum input data. Adjustments in specific hydraulic parameters were also made in an effort to calibrate the model to the soils in this region. Overall, reasonable agreement was found between TDR-measured and RZWQM-predicted average water contents for 0–60 cm depths. Surprisingly, the smallest errors in the predicted water contents were achieved using either the textural class only or the hydraulic properties determined in situ, with root mean square errors ranging from 0.012 to 0.018 m 3 m −3. Hence, the model provided adequate estimates of average profile soil water content based on textural class-name only which was considered the most limited input data condition.