Estimation of reservoir sedimentation using Per-pixel classification approaches

Abstract

For the quantification of sediments deposited in the reservoir, the only thematic information that has to be extracted from the satellite data is water spread area at different water levels of the reservoir. Mostly per-pixel methodology such as MLC, minimum distance, band threshold technique are used to classify the water-spread area from the satellite data. One of the limitations of the per-pixel approach in classifying water spread area is that the border pixels that are mixed in nature, representing soil, vegetation class with moisture are also classified as water pixels, thereby giving inaccurate estimate of the water-spread area. To accurately compute the water-spread area, sub-pixel classification or linear mixture model (LMM) approach has been chosen for classifying the water-spread areas of Singoor reservoir, Telangana, South India. LANDSAT8 (OLI) satellite data pertaining to the year 2014 were used to extract the water-spread area of the reservoir by band threshold and sub-pixel classification approaches. It was ascertained that the estimated capacity of the reservoir from band threshold and sub-pixel approaches were 695.71 Mm3 and 690.53 Mm3 respectively. The difference in capacity between the period 1987 (hydrographic survey) and 2014 (per-pixel and sub-pixel approaches) reveals the amount of sedimentation deposited in the reservoir. If uniform rate of sedimentation is assumed from 1987 to 2014, it is ascertained that the annual rate of sedimentation in Singoor reservoir using per-pixel and sub-pixel approaches are 2.70 Mm3 and 2.71 Mm3 respectively.