Assimilating Remotely Sensed Soil Moisture in the CLM Using 1D Kalman Filter
Wenge Ni-Meister
Jeff Walker
Paul Houser

Background
Extended 1-D Kalman filter (EKF) has been implemented into the catchment model and  was validated using identical twin experiment in North America by Jeff Walker
Ensemble Kalman filter (EnKF) has been implemented into the  catchchment model and was validated using identical twin experiement in  North America by Rolf Reichle
Currently we are assimilating the soil moisture data from SMMR into the the catchement model and validate the above two data assimilation methods using in-situ measurements in Eurasia and using the new catchment definitions

Approaches
Run the catchment model without data assimilation (case I)
Run the catchment model with data assimilation (case II)
Compare modeled soil moisture from the above two cases with the in situ SM measurements

Inputs
New catchment definitions,  soil parameters, and topographic information are from Sarith Mahanama.
Vegetation parameters:
Use Max Suarez’s SIB vegetation types + 9 year AVHRR greenness and LAI datasets from Pierre
 Generate MOSAIC vegetation parameters  based on the code given by Randy Koster
Convert to catchment-based vegetation parameters by choosing the dominant vegetation types
Compare the histograms of the land cover types both from the new  and old catchment definitions  to validate our algorithm
Forcings: ISLSCP forcings for the period of 1987-1988

Land Cover types

North America Runs

North America Runs

Eurasia Catchments

Model Predictions

Model Predictions

Model Predictions

Model Predictions

Dryness Index

Three Dataset Comparisons
Model predictions
 Moisture content at the surface layer (2cm), root zone layer (1m) and the whole soil profile
Period : 1987-1988
SMMR
at the very thin top surface (1cm), period: 1979-87
In-situ measurements
Plant-available SM: originally recorded as percent wetness by mass of dry soil->volumetric SM –> subtract  volumetric SM at wilting level -> Plant-available SM
Different datasets  (Russia (130 stations, 1978-85, Mongolia
(42 stations, 1964-93), China (43 stations, 1981-91) cover different years
Three datasets with different soil moisture definitions, how will we compare them?

SMMR Soil Moisture

SMMR Soil Moisture

Vegetation Attenuation from Polarization Ratio

In-situ Soil Moisture Data

In-situ Soil Moisture Data

Soil Moisture Comparison

Future Works
Use the ECMWF forcings from Jay’s group
Use the new snow model
Reprocess the in-situ data to check the seasonality of in-situ measurements
Run the data assimilation