Improving mesoscale soil moisture mapping with in situ networks and cosmic-ray neutron probes

Abstract

Soil moisture plays a vital role in the water and energy balance of rainfed agricultural and hydrological systems in the U.S. Great Plains. With the rise of mesoscale environmental monitoring networks that include soil moisture as part of the standard suite of measurements, this dissertation addresses a central question: Can we upscale point-level soil moisture observations from sparse monitoring networks by integrating spatial model estimates and in situ observations through data assimilation? The first chapter investigates a data-fusion method to leverage existing mesoscale rootzone soil moisture to create new, high-spatial resolution soil moisture maps for the state of Kansas. These maps are extensively validated using cross-validation and regional surveys conducted with a roving cosmic-ray neutron detector. The second chapter presents CRNPy, a Python library that compiles common correction methods for converting raw neutron counts into volumetric soil water content. The CRNPy library is then used to validate soil moisture estimates at watershed and state levels. The third chapter explores the use of a machine learning observation operator to translate soil moisture observations from stations of the Kansas Mesonet typically located in areas with perennial grassland vegetation to represent soil moisture conditions in nearby cropland fields. Collectively, these chapters offer new insights for leveraging in situ and proximal soil moisture data to advance our understanding of regional and mesoscale soil moisture, with a focus on harnessing the expanding infrastructure of mesoscale monitoring networks.