News & Updates
The 14th JCSDA Workshop on Satellite Data Assimilation
JCSDA News and Newsletters
SMAP Data Assimilation at the GMAO
From the Winter 2016 issue of the JCSDA Quarterly
The NASA Soil Moisture Active Passive (SMAP) mission (Entekhabi et al. 2010) has been providing L-band (1.4 GHz) passive microwave brightness temperature (Tb) observations since April 2015. These observations are sensitive to surface (0-5 cm) soil moisture. Several of the key applications targeted by SMAP, however, require knowledge of deeper-layer, "root zone" (0-100 cm) soil moisture, which is not directly measured by SMAP. The NASA Global Modeling and Assimilation Office (GMAO) contributes to SMAP by providing Level 4 data, including the Level 4 Surface and Root Zone Soil Moisture (L4_SM) product, which is based on the assimilation of SMAP Tb observations in the ensemble-based NASA GEOS-5 land surface data assimilation system (Reichle et al. 2015). The L4_SM product offers global data every three hours at 9 km resolution, thereby interpolating and extrapolating the coarser- scale (~40 km) SMAP observations in time and in space (both horizontally and vertically). Since October 31, 2015, beta-version L4_SM data have been available to the public from the National Snow and Ice Data Center (NSIDC; http://nsidc.org/data/smap) for the period March 31, 2015, to near present, with a mean latency of ~2.5 days.
The land surface model component of the assimilation system, the NASA GEOS-5 Catchment land surface model (Koster et al. 2000), is driven with observations-based surface meteorological forcing data, including precipitation (Reichle and Liu 2014), which is the most important driver for soil moisture. The model also encapsulates knowledge of key land surface processes, including the vertical transfer of soil moisture between the surface and root zone reservoirs. The radiative transfer model to simulate L-band brightness temperatures was calibrated using observations from the Soil Moisture and Ocean Salinity mission (De Lannoy et al. 2013). The horizontally distributed ensemble Kalman filter update step considers the respective uncertainties of the model estimates and the observations.
Figure 1. L4_SM analysis for 29 May 2015, 0z. The top row shows O-F Tb residuals for (a) H-pol and (b) V-pol. Analysis increments are shown for (c) surface soil moisture, (e) root zone soil moisture and (g) surface soil temperature. The resulting analysis fields are shown for (d) surface soil moisture, (f) root zone soil moisture and (h) surface soil temperature.
The quality of the L4_SM product is assessed against in situ measurements from watersheds with locally dense sensor networks and continental-scale sparse networks (Reichle et al. 2015). These comparisons indicate that the current beta version of the L4_SM data product meets the mission accuracy requirement, which is formulated in terms of the ubRMSE: the RMSE after removal of the long-term mean difference. For the period March 31 to October 25, 2015, the overall ubRMSE of the 3-hourly L4_SM surface soil moisture at the 9 km scale is 0.037 m3/m3. The corresponding ubRMSE for L4_SM root zone soil moisture is 0.024 m3/m3. Both of these metrics are well within the 0.04 m3/m3 ubRMSE requirement.
The assessment of the L4_SM product further includes a global evaluation of the internal diagnostics from the assimilation system, including statistics of the observation-minus-forecast (O-F) residuals and the analysis increments (Reichle et al. 2015). The instantaneous soil moisture and soil temperature increments are within a reasonable range and result in spatially smooth soil moisture analyses (Figure 1). The O-F Tb residuals exhibit only small biases on the order of 1-3 K between the (rescaled) SMAP Tb observations and the L4_SM model forecasts, which indicates that the assimilation system is largely unbiased (Figure 2a). The average (RMS) magnitude of the O-F residuals is 5.7 K, which reduces to 2.6 K for the observation-minus-analysis (O-A) residuals (not shown), reflecting the impact of the SMAP observations on the L4_SM system.
Figure 2. (a) Mean O-F Tb residuals from the current beta-version L4_SM algorithm for 11 Apr 2015, 0z, to 25 Oct 2015, 0z. (b) Same as (a) but for the standard deviation of the normalized O-F Tb residuals.
Finally, the standard deviation of the normalized O-F residuals measures the consistency between the expected (modeled) errors and the actual errors. Specifically, the O-F residuals are normalized within the standard deviation of their expected total error, which is composed of the error in the observations (including instrument errors and errors of representativeness) and errors in the brightness temperature model forecasts. The parameters that determine the expected error standard deviations are key inputs to the ensemble-based L4_SM assimilation algorithm. If they are chosen so that the modeled errors are fully consistent with the actual errors, the metric should be unity. If the metric is less than one, the actual errors are overestimated by the assimilation system, and if the metric is greater than one, the actual errors are underestimated. Averaged globally, the time series standard deviation of the normalized O-F residuals is 1.2 K/K and close to unity (Figure 2b). Regionally, however, the metric deviates considerably from unity, which indicates that the L4_SM assimilation algorithm either over- or underestimates the actual errors that are present in the system.
Several limitations of the beta-version L4_SM data product and science algorithm calibration will be addressed prior to the release of the validated data product scheduled for 2016. Planned improvements include revised land model parameters, revised error parameters for the land model and for the assimilated SMAP observations, and revised surface meteorological forcing data for the operational period and the underlying climatological data. Nevertheless, the current beta version of the L4_SM product is sufficiently mature for release to the larger science and application communities.
R. Reichle, G. De Lannoy, Q. Liu, and J. Ardizzone (NASA GMAO)
The authors gratefully acknowledge the contributions by many SMAP project team members and Calibration/Validation (Cal/Val) Partners at the Goddard Space Flight Center, Jet Propulsion Laboratory, NSIDC, and several universities and agencies. Computational resources are provided by the NASA High-End Computing Program through the NASA Center for Climate Simulation.
De Lannoy, G.J.M., Reichle, R.H., and Pauwels, V.R.N., 2013. Global calibration of the GEOS-5 L-band microwave radiative transfer model over nonfrozen land using SMOS observations. Journal of Hydrometeorology, 14, 765-785.
Entekhabi, D., Njoku, E.G., O'Neill, P.E., Kellogg, K.H., Crow, W.T., Edelstein, W.N., Entin, J.K. , Goodman, S.D., Jackson, T.J., Johnson, J., Kimball, J., Piepmeier, J.R., Koster, R.D., Martin, N., McDonald, K.C., Moghaddam, M., Moran, S., Reichle, R., Shi, J.-C., Spencer, M.W., Thurman, S.W., Leung Tsang, Van Zyl, J., 2010, The Soil Moisture Active and Passive (SMAP) Mission. Proceedings of the IEEE, 98, 704-716.
Koster, R.D., Suarez, M.J., Ducharne, A., Stieglitz, M., and Kumar, P., 2000, A catchment-based approach to modeling land surface processes in a general circulation model, 1: Model structure. J. Geophys. Res, 105, 24809-24822.
Reichle, R.H., and Liu, Q., 2014, Observationcorrected precipitation estimates in GEOS-5. NASA Technical Report Series on Global Modeling and Data Assimilation. NASA/TM-2014-104606, 35, 18 pp. Available at: http://gmao.gsfc.nasa.gov/pubs/.
Reichle, R.H., De Lannoy, G., Liu, Q., Colliander, A., Conaty, A., Jackson, T., Kimball, J., and Koster, R.D., Soil Moisture Active Passive (SMAP) project assessment report for the beta-release L4_SM data product. NASA Technical Report Series on Global Modeling and Data Assimilation, NASA/TM-2015-104606, 40, 63 pp. Available at: http://gmao.gsfc.nasa.gov/pubs/.