Modelagem da salinidade do solo com a utilização de técnicas de sensoriamento remoto

Abstract

The management of water resources in arid and semi-arid regions is essential if the use of local resources is to be sustainable. This management will depend on the construction of a base of information on the characteristics of each region. The use of new technologies for the construction of the database is fundamental, considering the technological progress that has occurred in recent years, as well as the elusive, the areas. Remote sensing combined with geoprocessing presents promising techniques in natural resources due to the ability to make large areas, store information, enable data transfer and ease of consultation. When the rules are adjusted to the regions of irrigation use, it becomes a single instrument that meets the needs of agricultural production. Being that, when practiced, it is uncontrolled because it is a degradation of the soil of water and vegetation. Monitoring and evaluation are important are key. With this, the objective was to evaluate the efficiency of the use of remote sensing and geoprocessing techniques in the monitoring of salinity and its effects on soil and vegetation. The research was carried out in the irrigation perimeter of Baixo-Açu, located between the municipalities of Alto do Rodrigues and Afonso Bezerra. Initially, a preliminary analysis of the perimeter was carried out, using satellite imagery and soil sampling in the field. The objective of this analysis was to identify through the production fault images and to verify the variation of the salt concentration in depth, identifying the best correlation between the salinity levels and the response of spectral indices, as well as to perform a temporal analysis of the vigor of the vegetation in the study area. A case study was carried out to evaluate the spectral index that best represents the salinity variation within the irrigated perimeter and to evaluate how the special resolution of the satellite images and the vegetation interfere in the determination of the salinity. In the following chapter, the spectral response of salinized soils was characterized and through the use of multiple regression techniques and spectral analysis were constructed and validated indices for mapping saline soils. Finally, the spectral response of the vegetation of saline areas was characterized and the use of specific indexes for soil salinity mapping was analyzed. The most superficial layer (0-10 cm) is the most suitable for analysis of correlation between soil EC and spectral indexes. It was also identified that several areas within the irrigation perimeter have high salinity imposing limitations to the vegetative development and that based on NDVI analysis assumes that these areas did not present such problems before the creation of the irrigated perimeter. Among the 20 spectral indexes analyzed for soil salinity mapping, SI1 was the one with the best correlation (R² = 0.80). The improvement of the spatial resolution is directly related to the improvement of the correlation results in the determination of the soil salinity when comparing Landsat8 and Sentinel2 images. The vegetation present on the soil surface was shown as a "noise" in the salinity mapping, and it was verified that for the use of spectral soil indexes, the area needs to be without surface vegetation. The band of the MSI / Sentinel2 satellite that best correlates with soil salinity is the green band (B03) with a determination coefficient of 59.85%, and all bands of the visible one show a significant correlation with salinity, soil determined by the use of TIRS / Landsat8 images did not present a good correlation. The elevation of the terrain also showed a significant correlation with 57.21%. The saline areas with exposed soil presented a spectral behavior different from the other areas, with a higher reflectance in the visible region and based on these analyzes it was possible to develop 15 spectral indices of salinity, being the best SA7 used for the mapping of the salinity of the place and validated with a R² of 83.84%. Vegetation indices were not good for soil salinity mapping, and the characteristic vegetation of saline areas (halophytes) presents a distinct reflectance characteristic of the other types of vegetation found in the area, mainly with a higher reflectance in the visible region.