Smos retrieval over forests: Exploitation of optical depth and tests of soil moisture estimates

TitleSmos retrieval over forests: Exploitation of optical depth and tests of soil moisture estimates
Publication TypeJournal Article
Year of Publication2016
AuthorsVittucci, C, Ferrazzoli, P, Kerr, Y, Richaume, P, Guerriero, L, Rahmoune, R, G. Laurin, V
JournalRemote Sensing of Environment
Date Published2016///
ISBN Number0034-4257

This research aims to test data obtained by level 2 retrieval algorithm of SMOS over land, in order to provide information regarding vegetation and soil moisture over forested areas. Results presented in this paper were obtained using the last 620 version of the algorithm.The correlation between the new vegetation optical depth (VOD) product and the height of the forest estimated by ICES at GLAS lidar on a global scale is investigated. Over South American and African forests a good correspondence between the two variables is observed, with saturation occurring above about 30 m height. Moreover, the comparison between the VOD and the height of the forest shows good spatial and temporal stability, and the r2 correlation coefficient is within a 0.59–0.69 range. Conversely, discrepancies are observed in some Indonesian islands, particularly New Guinea. Over specific areas, the trends vs. forest height obtained with SMOS VOD are compared with the corresponding trends of AMSR-E VOD. Results are also validated at country-level scale. To this aim, accurate estimates of forest biomass derived from airborne lidar over selected forests of Peru, Columbia and Panama are used.
Finally, the soil moisture retrieved over forests is investigated, reporting continental maps for Tropical areas and comparisons with ground measurements in selected forests of the US. Continental maps obtained with the new level 2 V620 algorithm cover almost all forest areas, and show seasonal variations which are dependent on climatic zones. Comparisons between soil moisture retrievals in forests and ground measurements of the US SCAN network produce worse RMSE values with respect to low vegetation areas. Significant improvements however are achieved after averaging among close nodes of the ground network.

Short TitleRemote Sensing of Environment