Online Vegetation Parameter Estimation in Passive Microwave Regime for Soil Moisture Estimation

Authors: Fitzmaurice, Jean; Crow, Wade

Submitted to: BARC Poster Day
Publication Type: Abstract Only
Publication Acceptance Date: 4/4/2010
Publication Date: 4/23/2010
Citation: Fitzmaurice, J.A., Crow, W.T. 2010. Online vegetation parameter estimation in passive microwave regime for soil moisture estimation [abstract]. Abs. 13, BARC Poster Day.

Interpretive Summary:

Technical Abstract: Remote sensing observations in the passive microwave regime can be used to estimate surface soil moisture over land at global and regional scales. Soil moisture is important to applications such as weather forecasting, climate and agriculture. One approach to estimating soil moisture from remote sensing observations is filtering, which combines (weights) soil hydrology models and passive microwave observations. A radiative transfer model is required which maps the soil moisture content and land surface properties to a remote sensing radiobrightness temperature. Vegetation canopies attenuate the remote sensing signal, and the vegetation parameter is key information for the radiative transfer model. The vegetation parameter is obtained outside of the framework, for example monthly climatology or other datasets such as visible/near infrared remote sensing vegetation indices. We propose an online parameter estimation method for the vegetation parameter where the parameter is added to the soil moisture state vector, called state-augmentation in the adaptive signal processing literature. A two-layer soil hydrology model is used with a radiative transfer model and a particular filter, the ensemble Kalman filter. Synthetic remote sensing radiobrightness observations are generated for a 184 day period for daily remote sensing observation filter updates. The filter assimilates the synthetic uncertain observations and estimates both soil moisture and the vegetation parameter. Satisfactory results are obtained for both static and idealized time-varying vegetation parameter cases. Persistent excitation, adding small mean zero Gaussian noise, is required in the time-varying case for the parameter estimate to converge close to the true time-varying vegetation value, which coincides with mathematical theory. The implications of this research are that there may be additional information, especially vegetation information, available to be extracted from passive microwave observations using the filtering approach. Future research will test this method using real data such as AMSR-E satellite data, launched in 2002.