The spatial and temporal organization of soil moisture

Abstract

Runoff, infiltration, evaporation and transpiration and-at climatic scales-precipitation are hydrologic processes that strongly depend on soil moisture. From a descriptive viewpoint, soil moisture is. characterized by an extremely high degree of heterogeneity over a wide range of scales in both time and space: a complex picture with nested patterns of variability. From a dynamic perspective many factors and processes influence the evolution of soil moisture at different scales: climate, topography, precipitation, geology, pedology, vegetation. In this thesis it is examined whether the apparently orderless spatial structure of soil moisture can be described using a simple statistical approach and the dynamics of soil moisture is investigated through a model. Experimental large scale data of soil moisture, measured from remote sensors, are analyzed under different resolutions. The variance of soil moisture fields is shown to follow a power law decay as function of the area at which the process is observed. The spatial correlation remains unchanged with the scale and follows a power law decay typical of scaling processes. Soil moisture also shows clear scaling properties on its spatial clustering patterns. A well-defined organization of statistical character links scales from the tens of meters to several kilometers. The dynamics of soil moisture in space and time at climatic scales is modeled including its coupling with the atmosphere through energy and mass fluxes. Water balance for the soil and the atmospheric components of the continental hydrologic subsystem as well as energy balance for the atmosphere, including an additive stochastic forcing term, constitute the model's equations. The system evolves in time and shows to be highly organized reaching a noise-induced statistical balance between positive and negative feedbacks. The generated soil moisture fields are analyzed in both time and space showing some of the typical aspects of experimental data: low pass-filtering of rainfall forcing, persistence of droughts and wet periods. Furthermore the model indicates a scaling spectrum of soil moisture for high frequencies, a strong correlation with the atmospheric moisture and precipitation, and positive spatial correlations ranging up to one-fifth of the domain size.