Towards Profile Soil Moisture Retrieval From Remote Sensing
Jeffrey P Walker, Garry R Willgoose and Jetse D Kalma
Remote sensing provides a capability to make frequent and spatially distributed measurements of surface soil moisture, whilst recent advances in affordable Time Domain Reflectometry probes allow continuous monitoring of profile soil moisture at specific points. We believe that reliable estimation of the spatial and temporal variation of profile soil moisture on a routine basis will require a combination of calibration and validation of an unsaturated soil moisture model using point measurements, and model updating using remote sensing observations to account for spatial inhomogeneities.
A one-dimensional profile soil moisture and heat transfer model has been updated from surface observations using both hard-updating and Kalman-filtering. With hard-updating, the simulated system states (ie. soil moisture and temperature) are directly replaced with observations, while Kalman-filtering updates with a statistically optimal estimate of the states, based on covariances of the model states and observations.
True soil moisture and temperature profiles were generated for a typical 1m soil column under prescribed surface fluxes. Starting from a poor initial estimate of profile values, the simulation was repeated for the same soil parameters and boundary conditions, but with "observations" of soil moisture over depths of 0, 1, 4 and 10cm and soil temperature at the surface. Updating was executed with temporal resolutions from one-hour to five-days. Results from hard-updating showed that: (1) updating of the surface node was ineffective unless used in conjunction with a direchlet boundary condition; (2) profile recovery rate increased as the updating depth increased; (3) a direchlet boundary condition was required for a greater proportion of the update interval as hard-updates became less frequent; and (4) recovery of soil moisture profile necessitated 7 days of hourly updating, more than 20 days of daily updating, and more than 40 days of 5 day updating, for an observation depth of 10cm. Results from Kalman-filtering showed: (1) it was far superior to hard-updating; (2) Kalman-filter updating once every hour surpassed a continuous direchlet boundary condition with the true surface states; and (3) recovery of soil moisture profile necessitated less than 1 day of hourly updating, 5 days of daily updating, and 30 days of 5 day updating. Results from a field application of the Kalman-filter for profile retrieval in a temperate zone in Australia will also be reported.