Surveying of Saline Soils with Spectral Analysis of LISS III Sensor and Field DataCase Study: Fars Province (Tashk & Bakhtegan Lake)

In this research the data of LISS-IIIsensor used for mapping saline soils on the around of Tashk and Bakhtegan lake, Fars Province. After the corrections of the imagery, the training and control pixels selected and the soil samples (depth of 5 cm) prepaired. Then pixels of the imagery classified in different approaches by maximum likelihood algorithm. Analysis of spectral graphs of classes indicated the most reflectance of saline soils was in visible spectral. The highest of correlation coefficient between salinity values and digital numbers of bands (0.45) observed in band 2. Investigation of feature space graphs indicated, almost single clusters formed in all of the classes except in low and moderate saline soils. The white salt crusts lied in the highest portion of soil line and black salt crusts in lower parts. The roughness and moisture were the main factors in reflectance decreasing of salt crusts in studied region. In non-saline soils the roughness produced by large clods in ploughed areas was the effective factor of reflectance decreasing. Presence of plant cover and residual of the straw caused error in identifying of saline soils. Results of classification of different approaches indicated the highest accuracy of procedure, user, overall and kappa index was observed in integration of all the bands. In all of the approaches, soils with low and moderate salinity had low accuracy. The produced map of LISS-III sensor indicated salinity values above of 16 dS/m there were on the around soils of lake and decreased one with distance, in center of the plain and increased in agriculture lands again.