Climatic variability of soil water in the American Midwest: Part 1. Hydrologic modeling

Abstract

A conceptual hydrologic rainfall-runoff model that is an adaptation of the operational US National Weather Service hydrologic model is used to simulate the hydrologic processes in large basins of the US upper Mississippi region. In particular, the conceptual rainfall-runoff model is used to produce daily streamflow from daily rainfall, temperature and potential evapotranspiration input over three neighboring headwater basins that span 2° longitude by 2° latitude. When used for simulation of historical flows, the model provides a means of inference of the 40 year time series of unrecorded mesoscale soil water and actual evapo-transpiration for climate studies. In this paper we discuss issues associated with parameter estimation, the reliability and stability of parameter estimates, and the interpretation of soil water estimates. It is concluded that the conceptual hydrologic model may be used to estimate the variability of aggregate soil water over large areas of the Midwestern US provided that: (a) all significant basin inflows and outflows are accounted for; (b) model verification yields good agreement between observed and simulated flows on a daily basis. Parameter sensitivity studies showed that estimating the soil's capacity to hold water is most important for flood event prediction and flow simulation, and, for such parameters, underestimation is more critical than overestimation. Also, uncertainty associated with the parametrization of evapo-transpiration may introduce local errors in the time series of soil water estimates produced by the model. In a companion paper we present a spatio-temporal analysis of the estimated time series of soil water.