Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → Brésil → Estimativa da umidade do solo na região semiárida do Brasil pelo método do triângulo universal

Instituto Nacional de Pesquisas Espaciais – INPE (2017)

Estimativa da umidade do solo na região semiárida do Brasil pelo método do triângulo universal

Costa Tássia Alves

Titre : Estimativa da umidade do solo na região semiárida do Brasil pelo método do triângulo universal

Auteur : Costa Tássia Alves

Université de soutenance : Instituto Nacional de Pesquisas Espaciais – INPE

Grade : Doutorado em Sensoriamento Remoto 2017

Soil moisture is a key variable of several processes in the earth system, with emphasis on water, energy and nutrient cycles in the soil-plant-atmosphere continuum. Direct and indirect methods have been developed to estimate its spatial and temporal variability, in situ or via remote sensing products. In the context of remote sensors, one of the first alternative proposed for estimation of surface soil moisture was the so-called ’method of the universal triangle’. This method presumes the existence of a relationship between the Normalized Difference Vegetation Index (NDVI), the Land Surface Temperature (LST) and the surface soil moisture. While most of the remote sensors used in estimating soil moisture work on scales of kilometers, the triangle method, in principle, allows working on the desired scale (that is, the same as the NDVI and LST data). The requirement, however, is to select an area with the widest possible range of NDVI and LST values. In the present work, it were used MODIS data, MOD13A2 (NDVI) and MOD11A2 (LST), with a spatial resolution of 1 km, to test the ability of the method to estimate the surface soil moisture (Mo) in a MODIS scene, in the Brazilian semi-arid region. The results were validated in comparison with in situ soil moisture data obtained from two different research sites, São João and Serra Talhada, representing, respectively, the Caatinga ecoregions known as the Borborema Plateau and Southern Sertaneja Depression. Two alternatives were tested for the development of the relationship between Mo, NDVI, and LST : a simple geometric solution and a third-degree polynomial. The comparison with in situ data resulted in correlation coefficients (R²) of 0.18 and 0.78, respectively, for both solutions. The polynomial equation was validated in two ways : through the conventional validation process, where part of the data is used to build the equation, and the remainder to verify it ; and by cross-validation procedure. The R2 values obtained in both processes were, respectively, 0.78 and 0.91. The high correlation coefficient achieved with the cross-validation process indicates the existence of a strong dependence on the surface soil moisture about the NDVI and LST parameters, as recommended by the method applied. Thus, the polynomial solution was used to spatialize the surface soil moisture estimated within the MODIS scene, for two different periods : dry and rainy seasons. For the dry period, soil moisture ranged from 0.1 to 0.3 cm3 cm-3 throughout the scene. For the rainy season, the variation obtained was 0.3 to 0.5 cm3 cm-3. Based on these results, we consider that the triangle method presents a potential for surface soil moisture estimates for the Brazilian semi-arid region.

Mots Clés  : Soil moisture. Scatterplot NDVI x LST. MODIS data. Triangle method. Caatinga. Brazilian semiarid

Version intégrale

Page publiée le 14 avril 2019