Publication

Abstract : This study analysed the monitoring of tea replantation using Linear Imaging Self‐Scanning Sensor (LISS-III) and Cartosat-1 images and identified patterns based on wavelet approaches. Monitoring identifies four phases of replantation and rejuvenation, starting at the time of uprooting and finishing when new plants are planted. The study was conducted within the Dooars region of North East India. The perpendicular vegetation index and perpendicular soil index were derived to measure changes from bare soil reflectances caused by vegetation, whereas the soil index was designed to enhance brightness. Being a multi-resolution study, wavelets such as Haar, Daubechies and Symlet were compared at different levels of decomposition, and information was extracted at different scales. Using topographic and hydrological parameters, informative patterns for each stage of replantation were selected at individual sections within the estate on the basis of spatial correlation. The study showed that levels 3 and 4 gave superior information compared with the other levels. Anisotropic autocorrelation gave constant spatial variation at different scales and in different directions. The selected patterns were weakly correlated with slope, flow accumulation and the compound topographic index, whereas management activities and a small variation in elevation proved less efficient in explaining the extracted patterns. It also showed that hydrological processes could be evaluated using cross-correlations. From the study, it was observed that the asymmetric Daubechies-4 wavelet gave the best results for extraction of fine features, whereas the symmetric Symlet-8 wavelet best represented the extraction of smooth features. Although a strong quantitative linear relationship between the extracted patterns and topographic parameters could not be established, we conclude that wavelets are useful to extract patterns and interpret spatial variations observed at different phases of tea replantation.