Submitted to: Latin American Symposium on Remote Sensing
Publication Type: Proceedings
Publication Acceptance Date: November 7, 1995
Publication Date: N/A
Interpretive Summary: In Mexico, 1.8-4.0 million hectares of irrigated land are saline enough to reduce yields of crops. In this study, procedures developed by ARS scientists at Weslaco, TX were applied as a pilot test to wheat in the EL Carrizo Irrigation District near Los Mochis, Sinaloa, Mexico. We made combined use of satellite digital spectral observations, ground sampling of frepresentative fields for soil salinity and yield of wheat, regression analyses, and unsupervised image classification. Digital data for the same locations in the field where soil salinity and yield samples were taken enabled us to develop regression equations that estimated soil salinity and yield from the digital counts of the satellite data. Unsupervised classification of the satellite data from the wheat fields established categories or classes of wheat that "looked different" in the satellite data. The mean values for the classes, when inserted in the regression equations, defined the spectral classes for both soil salinity and crop yield. We determined the acreage and yield of each salinity and yield category and prepared salinity and yield maps for the whole district. Each unit increase in salinity above 4 dS/m (desisiemens per meter) reduced yield by 430 kg/ha (kilograms per hectare) or by about 10,000 metric tons for the whole irrigation district. The procedures will be applied to additional irrigation districts.
In this study, satellite images were used to identify the salinity of the soil and to estimate the losses in yield of wheat (Triticum aestivum) cultivated in fall-spring 1003-1994 season in Irrigation District No. 076, Carrizo Valley, Sinaloa, Mexico, which is an area with soil salinity problems. At the following stage of the wheat, one Landsat Thematic Mapper r(TM) and one Spot panchromatic image were obtained. In salt-affected fields observation points were established for taking soil samples for salinity and grain yield samples. The digital counts for the observation points were extracted for TM bands 2, 3 and 4 and correlated with the salinity and yield of the sample points. That permitted us to obtain regression equations to estimate the salinity and grain yield of all the area sown to wheat. A nonsupervised classification of the marked out wheat fields was also obtained. The mean values of the spectral categories were substituted into the regression equations to estimate the salinity and yield of the spectral classes and to produce maps of the salinity and yield classes. It was determined that salinity reduced yield of wheat by approximately 10,000 metric tons for the whole irrigation district.