Assessment of LST and NDMI indices using MODIS and Landsat images in Karun riparian forest

Abstract

Riparian forest plays a significant role in ecosystems. Also, research on land surface temperature and soil moisture is essential in earth science and forest studies. Because measuring methods are difficult to apply in large areas and especially in dense forests, in this study normalized difference moisture index (NDMI) and land surface temperature (LST) were estimated using the infrared thermal method by data of Landsat 8 and Moderate Resolution Imaging Spectroradiometer (MODIS) in the Karun riparian forest that is of ecological importance in the Khuzestan province of Iran. The results showed that the accuracy for estimated NDMI and LST was appropriate (root mean square error = 3.45). In addition, the used polynomial support vector machine algorithm for classification by four classes (forest, agriculture, river, and others) and the validity of classification in these areas were suitable (overall accuracy = 95%, kappa coefficient = 0.93). Also, the NDMI index was dependent on changes in LST and Pearson coefficients were 0.94 and 0.84 for Landsat 8 and MODIS data, respectively. The average temperature of the area was obtained as 43.22 and 42.77 for Landsat 8 and MODIS, respectively. Finally, more protection of this forest against LST enhancement and reduction in soil moisture is necessary.