Heyvaert, Z., Scherrer, S. A., Dorigo, W. A., Bechtold, M., & De Lannoy, G. (2023). Joint assimilation of SMAP soil moisture and AMSR2 vegetation optical depth retrievals into the Noah-MP land surface model. In EGU General Assembly 2023. EGU General Assembly 2023, Wien, Austria. EGU. https://doi.org/10.5194/egusphere-egu23-8638
E120 - Department für Geodäsie und Geoinformation E120-08 - Forschungsbereich Klima- und Umweltfernerkundung
EGU General Assembly 2023
EGU General Assembly 2023
23-Apr-2023 - 28-Apr-2023
remote sensing; soil moisture; vegetation
As soil moisture and vegetation water content both affect the emissivity from the land surface, each of them can be derived from satellite-based passive microwave measurements. In this study, we use soil moisture retrievals from the 36 km SMAP L2 product and X-band vegetation optical depth (VOD) from AMSR2 LPRM version 6. VOD is a proxy for vegetation water content, linked to the leaf area index (LAI). We developed a machine learning-based observation operator to map LAI to VOD.
We assimilate the SMAP and AMSR2 products into the Noah-MP land surface model (LSM) with dynamic vegetation. This is done by means of a one-dimensional Ensemble Kalman Filter (EnKF) within the NASA Land Information System (LIS). SMAP soil moisture retrievals update soil moisture in each of the four soil layers of the LSM, while AMSR2 VOD retrievals update the LAI. A cumulative distribution function (CDF) matching approach rescales the soil moisture retrievals to the model climatology. Model LAI is mapped to VOD by means of the above-mentioned observation operator. The resulting data assimilation (DA) system produces consistent estimates of all land surface variables on a quarter-degree regular grid over the European continent from 1 April 2015 through 31 March 2022.
This joint SMAP and AMSR2 DA system is validated by assessing a number of geophysical variables. The surface and root-zone soil moisture estimates are evaluated using in situ observations from the ISMN. Gross primary production (GPP) and evapotranspiration are evaluated using FLUXNET data. Estimates for LAI are compared with optical satellite data from MODIS. The results are compared with open loop (model-only), and SMAP- and AMSR2-only DA experiments.
SMAP-only DA primarily improves soil moisture estimates, while AMSR2-only DA mainly improves estimates of GPP and ET. Preliminary results indicate that the joint DA has the potential to combine the improvements of both individual assimilation systems.
Environmental Monitoring and Climate Adaptation: 100%