Principal component analysis and biophysical parameters in the assessment of soil salinity in the irrigated perimeter of Bahia, Brazil

Abstract

Soil salinity is an environmental concern which can lead to land desertification. It is a common problem in the semiarid region of northeastern Brazil and its impacts extend from the environment to the economy. The objective of this study was to evaluate soil salinity in the dry period in an agricultural area of irrigated Maniçoba perimeter, in Juazeiro, State of Bahia, Brazil. The study was carried out in an agricultural area of the irrigated perimeter in Juazeiro of Bahia, Brazil, where 98 disturbed soil samples were collected in different areas, being 33 samples to obtain Electrical Conductivity (EC) (dS m−1) and regression models, and 65 samples to validate the models. Images from the Landsat-8 and Sentinel-2 satellites were used to obtain the biophysical parameters: NDVI, SAVI, EVI, and GDVI vegetation indices, SI-1, SI-2, SI-3, and IB salinity indices, albedo, temperature (TSUP), and real evapotranspiration (ETR). Descriptive statistics and principal component analysis (PCA) were applied to identify majors EC and biophysical parameters relationships, and select the best predictors to generate equations capable of estimating soil salinity across the irrigated perimeter. The EC showed a high standard deviation and coefficient of variation compared to the other variables, with higher concentrations observed in areas of bare soil, followed by natural vegetation and agricultural vegetation. TSUP and GDVI were the predictors most strongly related to EC, for images Landsat-8 and Sentinel-2 respectively, with R2 of 73% and 71%. The application of the generated models presented satisfactory validation for two distinct areas. The maps obtained were similar in terms of the distribution of soil saline levels, confirming what was observed in the field scale, with a greater level of detail for the map from Sentinel-2.