Abstract:
Soil moisture is a key variable for weather and climate prediction, flood forecasting, and the determination of groundwater recharge. But uncertainties related to the heterogeneity of the land surface and the non-linearity of land-atmosphere interactions severely limit our ability to accurately model and predict soil moisture on regional or continental scales. On the other hand, remote-sensing techniques (such as C-band or L-band passive microwaves) can only indirectly measure surface soil moisture, and the data are of limited resolution in space and time.

We present a ``weak constraint'' variational data assimilition algorithm which takes into account model as well as measurement uncertainties and optimally combines the information from both the model and the data by minimizing a least-squares performance index. We achieve a dynamically consistent interpolation and extrapolation of the remote sensing data in space and in time, or, equivalently, a continuous update of the model predictions from the data.

The algorithm is used to run a series of experiments with synthetically generated parameter and system noise. Such experiments are best suited to evaluate the performance of the algorithm as the uncertain inputs are known. All experiments are set up to mimick the conditions during the 1997 Southern Great Plains (SGP97) experiment in central Oklahoma. Specifically, we address three topics which are crucial to the design of an operational soil moisture data assimilation system:

(1) the length of the assimilation time window before re-initialization,

(2) the satellite re-visit interval, and

(3) the effect of different types of vegetation.