SMOS L2 Retrieval Results over Forests and Tests
Ferrazzoli, Paolo¹; Rahmoune, Rachid¹; Singh, Yogesh Kumar¹; Kerr, Yann²; Richaume, Philippe²; Al Bitar, Ahmad²; Moisy, Christophe^3
1Tor Vergata University, DICII, ITALY; ²CESBIO, FRANCE; ³INRA, FRANCE

The SMOS satellite carries the MIRAS L-band radiometer, aimed at monitoring soil moisture and ocean salinity. For land applications, a retrieval algorithm was developed, which subdivides land into several categories. For low vegetation and forests, the algorithm estimates soil moisture and optical depth as outputs of the retrieval process (Kerr et al. , IEEE TGRS, vol. 50, pp. 1367-1383, 2012). For the specific case of developed forests, the optical depth is relatively high (in the range 0.6-1.2) which makes the retrieval of soil moisture difficult. For optical depths typically found for forests at L band, a soil moisture variation of 0.04 m^3/m^3 corresponds to a brightness temperature variation in a range 2.4-3.0 K, which can be not sufficiently higher than the noise in the measurements. In spite of these problems, several efforts have been done to improve the performance of the forward model and the retrieval algorithm. This produced an improvement in the soil moisture retrieval performance, at least under moderately developed forests and mixed pixels. Moreover, the retrieved optical depth can be considered as a product by itself. An accurate mapping of this variable can be useful for further exploitation of SMOS, for planning future L band systems, and as a new information source about forest biomass (or height), to be cross-checked with already available information sources.
In this work, the retrieval algorithm and its recent advances are briefly summarized, and last results about retrieved optical depth and soil moisture are presented. Figure 1 shows a map of retrieved optical depth over forests obtained by averaging the results collected from July 1, 2011 to July 4, 2011. The spatial trends agree with expectations, since the higher values are observed in dense rain forests of Amazonia and Indonesia, and intermediate values are found in forests on North American continent. Unfortunately, many pixels of Europe and Asia are severely affected by RFI. Optical depth values were compared against forest height measurements collected by ICESAT GLAS Lidar (Simard et al., Journal of Biophysical Research, vol. 116, 2011). An interesting correlation was observed, and there is no evidence of saturation effects for the highest forests.
Results of soil moisture retrieval were tested against ground measurements of the SCAN/SNOTEL network, in the US (Schaefer et al., J. Atmos. Ocean. Technol., vol. 24, pp.2073-2077, 2007). Test results are critically analyzed and directions of further work are indicated.