A synergistic use of multi-temporal ASAR and PALSAR data for wheat mapping and monitoring the underlying soil moisture content
G. Satalino(1), A. Balenzano(1), F. Mattia(1) and M. Rinaldi(2)
(1) CNR-ISSIA, Via Amendola 122/D, Bari, Italy
(2) CRA-SCA, Via Celsio Ulpiani 5, Bari, Italy
This paper investigates the use of multi-temporal L & C band SAR data for wheat mapping and for the retrieval of the underlying soil moisture content. Its final aim is to contribute at defining and assessing retrieval strategies for monitoring agricultural crops using current and future satellite SAR data.
Past studies have shown that a condition for successful retrieval algorithms consists of a thorough understanding and an appropriate modelling of the relationships between SAR measurements and soil and vegetation parameters. Besides, it is necessary to have updated information about land cover, at least in terms of main crop classes (e.g. small stems/broad leaves crops), in order to apply the appropriate retrieval approach. Then, in view of this consideration, the development of retrieval and classification methods are strictly connected. In this respect, the use of multi-frequency and multi-polarimetric SAR data can be beneficial due to their different sensitivity to vegetation and soil components. For instance, for crop canopies characterized by a predominant vertical structure (e.g. cereal crops), it has been observed that the interaction between the vegetation layer and the SAR signal at L-band and HH polarization can be disregarded, thus simplifying the retrieval of soil moisture content. Therefore, the HH/VV backscatter ratio at C-band can be employed to map cereal fields because their canopy attenuates differently the H and V polarizations, whereas this is not observed for other canopies characterised by more branched structures or broad leaves (e.g. tomato or sugar beet).
In this context, the objective of this paper is to assess a combined use of ASAR AP and PALSAR data for wheat mapping and for retrieving the underlying soil moisture content. More precisely, the ASAR HH/VV backscatter ratio is firstly employed to identify the wheat fields, and then multi-temporal PALSAR HH data on the obtained wheat map are used as input to the soil moisture retrieval algorithm. In order to improve the robustness and the accuracy of the retrieval algorithm, a constrained minimization technique integrating a priori information on surface parameters is exploited.
The analysed data set has been collected over an agricultural area of approximately 700 km2 located in the Capitanata plain, close to the Foggia town (Puglia region, Southern Italy). The area has a flat topography and is mainly devoted to wheat cultivation. On this area, temporal series of ground and ASAR and PALSAR data at fine resolution have been acquired from 2006 to 2008.
Results indicate that the proposed combined use of ASAR and PALSAR data permits to achieve classification and soil moisture retrieval accuracies of approximately 80% and 5%, respectively. Future work will be dedicated to integrate the proposed monitoring approach into a crop growth model and to assess its potential for the wheat yield mapping.