Consistent estimates of soil moisture information can be obtained in various ways (i.e., remote sensing, modeling). However, these estimates are not perfect and they all have characteristic uncertainties. It is therefore often desirable to merge these independent realizations to obtain a more-accurate estimate. Data assimilation using Kalman Filter-based methodologies are one of the most commonly used approaches for merging different products while taking into account the relative uncertainties of each product. However, these methodologies often rely on ad-hoc estimates of the relative errors of the products. For instance, in land data assimilation studies, uncertainties of observations are predefined by the user, by often neglecting a rigorous justification. Furthermore in these studies, the magnitude of the model uncertainty estimates heavily depend on ensemble representation of the given variable which is closely tied to the user-defined uncertainty of the forcings that drive the land surface models. As a result, in majority of the current land data assimilation studies, both observations and model uncertainty estimates that drive the merging process are obtained through some user-defined estimates.

In this study, we are proposing a new merging methodology that combines different soil moisture products using the error estimates of each product without integrating any user-defined parameter. Specifically, we are merging soil moisture estimates that are obtained from Noah land surface model, ALEXI energy balance model using thermal infrared remotely sensed images, and AMSRE based retrievals using microwave remotely sensed images, while the relative errors of each product are obtained through triple collocation.

In general, the primary idea of the merging algorithms is to preserve the existing climatology and the trends while reducing the random noise that exists in different products. However, not taking into account the climatology and trend differences reduces the potential benefit that can be obtained through the merging algorithm. In the proposed methodology we are also taking into account the differences that exist in different soil moisture products, particularly in their climatology and trends using standardized anomalies. This proposed methodology will add value to the current and future soil moisture satellite missions (i.e, SMOS: Soil Moisture and Ocean Salinity; SMAP, Soil Moisture Active Passive) by reducing the errors of these microwave remote sensing-based soil moisture estimates with infrared based observations and land surface models.

Given an improved root-zone soil moisture product is obtained through this methodology, the anomalies of the final merged estimate will be used as a new drought product on weekly and monthly time-scales. This new product will also improve the currently available existing drought products with the improved accuracy.