Spatiotemporal patterns of hydroclimatic drivers and soil-water storage: Observations and modeling across scales

Elin M Jacobs, Purdue University

Abstract

Soil-water storage is a critical regulator of water and energy balances at the land surface-atmosphere boundary. Root-zone soil-water in particular provides a key link between soils, vegetation and climate with consequent impacts on water and biogeochemical cycles and on ecosystem functions. The overarching aim of this dissertation is to contribute to the understanding of spatial and temporal patterns and properties of hydroclimatic variables, with focus on soil-water storage and its key hydroclimatic drivers in the Midwestern United States. Spatial and temporal patterns and trajectories are analyzed based on long-term field observations and high-resolution, numerical, land surface modeling. A simple stochastic model was also used and evaluated as an alternative to the numerical models. The stochastic model generates a probability density function p(s) representative of long-term variability of the state variable, s, or soil-water storage. Empirical and analytical probability density functions were compared to evaluate the sufficiency of the simple stochastic model. Additionally, seasonality in hydroclimatic forcing was explicitly considered. While the studied region has undergone a change in hydroclimatic forcing over the past three decades, this has not significantly impacted average or extreme water-storage states or the persistence of dry and wet conditions. Spatially, the region is found to be largely homogeneous in terms of hydroclimatology despite heterogeneity in surface properties. The surface properties tend to drive spatial variability in soil-water storage mainly during dry periods. Because the soil-vegetation-climate system is complex and the future is uncertain, models need to be able to account for this uncertainty. Current land surface models attempt to provide detailed parametrizations of all necessary physical processes but are still unable to do so, particularly for root zone water content. The findings from this research suggest that inclusion of stochastic elements in modeling and use of a probabilistic framework for spatiotemporal data analysis can help shed light on what can be expected in a variable and changing climate while reducing the computational effort required.

Degree

Ph.D.

Advisors

Rao, Purdue University.

Subject Area

Hydrologic sciences|Climate Change|Environmental engineering

COinS