Estimativa da umidade do solo e caducifolia em plantas da caatinga com uso de sensoriamento remoto

• Autor(a) principal: Lima, José Brenno Carneiro de
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da Universidade Federal do Ceará (UFC)
• Texto Completo: http://www.repositorio.ufc.br/handle/riufc/63777
• topic: Índice de vegetação; Sensoriamento remoto; Semiárido; Landsat; Vegetation index; Remote sensing; Semiarid; Landsat

Soil moisture in the root zones is an environmental variable that has a direct influence on vegetation. During the dry season, the root zone of vegetation in the Caatinga biome is subjected to water stress. To overcome this adverse climatic condition, plants have developed adaptive mechanisms, among them, deciduous, a physiological response to reduce water loss through transpiration. Even though it is a process that impacts both the hydrological cycle and the ecosystem, this vegetation response is still poorly studied at the scale of hydrographic microbasins in the semi-arid region. Thus, this study aims to evaluate soil moisture and caducifolia in caatinga preserved with the use of remote sensing. The study was carried out in the Experimental Basin ofAiuaba (BEA; 12 km²) using 57 images without cloud cover and suitable for analysis, out of the 329 made available by the Landsat5 and Landsat8 projects. For the analysis of the distribution of soil moisture, the methodology was used to estimate soil moisture through the Dry Temperature-Vegetation Index (ISTV). To compare the results, the daily average of soil moisture and the hourly average moisture closest to the satellite pass were used, both measured in the field at a depth of 0-20 cm. For the analysis of deciduous leaves, the average values of NDVI of BEA over time were used in order to determine the seasonality of leaf decay. Statistical analysis of remotely measured and estimated values was performed by applying correlation analysis (R), determination (R²), Nash-Sutcliffe efficiency index (NSE), root mean square error (MEQ) and root mean error quadratic (REMQ). The results show that none of the ISTV variations were successful either in determining soil moisture or in its spatialization, having beenhighly influenced by the effect of relief shadows, which end up masking the current situation of soil moisture. For vegetation decay, the available images were limited to the dry season, with no data on the beginning of decay due to high cloud cover in the rainy season. As a result, a potential model with asymptotic curve was obtained for the relationship between the NDVI and the Julian day from the end of March until the end of August, a period that represents the vegetation without leaves. This equation obtained an R = 0.879; R²= 0.773; NSE = 0.270; CMS =0.005; REMQ = 0.069. The daily decay rate of NDVI in the basin was of the order of 0.015. Furthermore, it was observed that the decay occurs differently when we analyze the soil-vegetation associations (ASVs) of the basin separately, with the NDVI decay equal to 0.015 d-1, 0.018 d-1, 0.014 d-1 for the ASV1, ASV2 and ASV3, respectively. It is concluded that the application of the ISTV model using Landsat images in BEA was not successful in determining soil moisture, with the main factor being the low temperatures of shadow projections that ended up overestimating the moisture values and preventing the correct determination of spatial differences. of soil moisture, ISTV has a limitation of application in areas of uneven relief that have a large projection of shadows; Deciduous leaves in BEA showed a seasonal pattern and occurred non-uniformly among ASVs, with a different decay according to the association of soil and vegetation. The exponential model was the one that best represented the estimate of the NDVI decay as afunction of time.