Assessment of salt-affected soils using multisensor radar data: A case study from northeastern Patagonia (Argentina)

H. F. DEL VALLE, P. D. BLANCO, W. SIONE, C. M. ROSTAGNO AND N. O. ELISSALDE

Remote Sensing of Soil Salinization: Impact on Land Management.

CRC Press, a Taylor & Francis Group

Año: 2009; p. 155 - 173

Dryland and irrigation salinity are widespread in the lower Chubut valley, northeastern Patagonia. The latter is the southernmost valley of Argentina, where forage, vegetables, fruits and flowers are cultivated under intensive land management. The overall aims of this study were: (1) to evaluate the usefulness of radar-derived parameters for detecting and mapping salt-affected soils within the irrigated portion of the valley, and (2) to classify the main landuses/landcovers using an object-oriented image classification approach. Imagery from two radar systems was used to this end: RADARSAT-1 (C-band, HH polarization, acquired in 2005 with a spatial resolution of 8 m), and SIR-C (C- and L-bands, multi-polarization, acquired in 1994 with a pixel resolution of 25 m). The use of L-band can be considered as a test for the future use of the SAOCOM (Satellites for Observation and Communications) data. Variables related to the radar system design (mainly wavelength, number of looks and polarization mode) and to the environmental setting of the area were analyzed to better understand the behavior of the radar backscattering. Four factors resulted to be significant when analyzing the variations of the backscattering coefficients in the environmental setting considered: soil type (soil texture), tillage conditions or soil surface aspect, soil moisture or surface water, and salt-affected soils. The average backscattering values for all salt-affected soil classes were higher in the L-band than in the C-band of the SIR-C, when the same polarization modes were compared. The radar response in terms of backscattered intensity and the interaction with salt-affected soils are dependent upon a number of system and target parameters. All parameters interact and influence each other. In this sense, both radar systems (RADARSAT-1 and SIR-C) used in this study showed different identification capabilities.