| Abstract: |
Microwave remote sensing represents a very useful tool to monitor soil moisture at different spatial and temporal scales, including both active and passive observations, depending on the sensor mode of operation, with different swath widths and temporal and spatial resolutions. Microwave scatterometers (active sensors) and radiometers (passive sensors) can be used to monitor the radar backscattering and the surface emission, respectively.
Then, a validation of remotely sensed data is of extreme importance for a better exploitation of the data themselves. In this work, a comparison study was performed among different remotely sensed surface soil moisture datasets, including the L-band Soil Moisture and Ocean Salinity (SMOS) radiometer, the C-band Advanced Scatterometer (ASCAT) onboard METOP and the Soil Moisture Active and Passive (SMAP) satellite developed by NASA. These satellite retrievals were compared with the simulations of the ERA-Interim model produced by the European Centre of Medium Range Weather Forecasts (ECMWF) and with ground measurements available in the frame of the International Soil Moisture Network (ISMN). The reprocessed L2 SMOS SMC products, produced by the last version 6.50 of the ground processor, and the SMAP passive L2 moisture products version 4 are considered. As for ASCAT, since each data represents a soil moisture relative index between 0 and 100% (i.e., driest and wettest conditions), the retrievals were converted into volumetric soil moisture using the porosity map available from the Global Land Data Assimilation System website. All the data were collocated in time and in space over the SMOS grid in the period from April 2015 to July, 2017.
A novel Extended Quadruple Collocation (E-QC) was developed (Pierdicca et al., 2017) and used in order to consider the possibility of an error cross-correlation between soil moisture products, identifying automatically the couple of error cross-correlated systems. The method was able to correctly retrieve the error of each system and identified the presence of error cross-correlation between satellite products, even if derived by different algorithms and operation principles.
With respect to the past, in this work the last release of the SMOS and SMAP data have been considered in the E-QC analysis and a longer period has been investigated. In addition, the problem of the representativeness error, in case the compared maps have not the same spatial resolution has been investigated. In fact, common small-scale features that are detected in the products at higher resolution can be misinterpreted as a noisy contribution. In order to investigate this aspect, a geospatial analysis has been carried out, trying to disjoint the correlated error from a common small-scale component by analysing the semivariograms of the different products. This analysis is at an early stage but some preliminary results are expected at the time of the workshop.
N. Pierdicca, F. Fascetti, L. Pulvirenti, R. Crapolicchio, “Error Characterization of Soil Moisture Satellite Products: Retrieving Error Cross-Correlation Through Extended Quadruple Collocation”, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, in press, 2017
|
