Upscaling soil moisture measurements from in situ sensors

Abstract

Accurate, field-scale soil moisture information is needed to match the spatial scale of land and water management decisions related to agricultural production and environmental protection. Soil moisture measurements at the field-scale are limited because the resolution of most satellite-based soil moisture products is too coarse, while most in situ monitoring networks can provide only point-scale, not field-scale, data. This research attempts to develop a broadly applicable upscaling approach for observations from in situ soil moisture sensors using data from the Marena, Oklahoma, In Situ Sensor Testbed (MOISST) and a cosmic-ray neutron rover. The landscape at the MOISST site is predominantly grassland with some deciduous trees and eastern redcedar intermixed. Cosmic-ray neutron rover survey data were used to measure average soil moisture for the ~64 ha site on 12 dates in 2019-2020. The relationships between the point-scale in situ data and the field-scale rover data were examined using data from six in situ stations. Statistical modeling was used to identify the soil, terrain, and vegetation properties that influence these relationships. Site-specific linear upscaling models estimated the field average soil moisture with root mean squared error (RMSE) values ranging from 0.014 - 0.022 cm3 cm-3, but these models are not transferable to other sites. A general upscaling model using soil texture data was developed and achieved RMSE values ranging from 0.017 - 0.038 cm3 cm-3 for four calibration sites and values ranging from 0.015 - 0.021 cm3 cm-3 for two validation sites. The general upscaling model demonstrated accuracy better than the commonly used threshold of 0.04 cm3 cm-3 and should be further tested to evaluate its suitability as a broadly applicable upscaling approach for point-scale in situ monitoring stations.