B1453
Title: Spatiotemporal data fusion method for soil moisture data
Authors: Weiyue Zheng - University of Glasgow (United Kingdom) [presenting]
Marian Scott - University of Glasgow (United Kingdom)
Claire Miller - University of Glasgow (United Kingdom)
Andrew Elliott - University of Glasgow (United Kingdom)
Abstract: High-resolution soil moisture data have great value in many different application areas. Soil moisture can be measured in various ways, including in-situ sensors and satellites. In-situ sensor networks can provide accurate and stable long-term soil moisture values but typically have limited spatial coverage. Satellite images typically provide good spatial coverage but less frequent temporal coverage. Typically, high spatiotemporal resolution data cannot be obtained from a single instrument because of the trade-off between high spatial and temporal resolutions. In general, every data source has its advantages and disadvantages; neither can simultaneously provide soil moisture with high accuracy and high spatiotemporal resolution. A spatio-temporal data fusion method is developed using an SPDE (stochastic partial differential equation) approach to generate detailed soil moisture maps from in-situ sensors and satellite data. The innovation includes accommodating both misaligned and non-misaligned covariates in a spatio-temporal perspective and integrating diverse data sources of the same variable, which can be compounded by differences in spatial and temporal resolution. The preliminary results are presented both in a detailed simulation and in the real data application from the Elliot Water in Scotland, UK.