CLOUD BASED GEO-PROCESSING PLATFORM FOR ANALYZING LARGE VOLUME TEMPORAL SATELLITE DATA: A STUDY IN PART OF GHAGHARA RIVER BASIN (INDIA) FOR SURFACE WATER SPREAD ANALYSIS

摘要

With the availability of large spatio-temporal Earth Observation satellite data and geospatial layers with concomitant increase in geo-computation requirement, the popularity of cloud based platform for accessing, sharing and processing large volume data is growing day by day among the geospatial community. Google's Earth Engine is one such platform, which provides access to satellite and other geospatial datasets and analysis functionalities to researchers free of cost, for various applications from local to global-scale. This study aims at exploring Google's Earth Engine to study the spatio-temporal dynamics of surface water during the Indian Summer Monsoon season over the last three decades. Top of Atmosphere (TOA) reflectance images available from Landsat-5, 7 and 8 (1984 to 2016.) and Aqua MODIS (2002 to 2016) are used in this process. A part of Ghaghara river basin covering about 8400 km~2 area is selected as the study site. The clouds and their shadows are masked using "Fmask" and "StateQA" quality bands available along with TOA reflectance images for Landsat and MODIS, respectively. Normalized Difference Water Index (NDWI) and Normalized Difference Vegetation Index (NDVI) are calculated and by applying suitable thresholds and combining them, binary surface water images for each date of acquisition are generated. By combining the binary surface water images over the entire study period, a composite long-term surface water map is generated wherein each pixel represents the frequency of the presence of surface water for cloud-free observations. The output is compared with the geomorphology map. Such composite long-term surface water map is extremely useful not only for understanding the spatio-temporal dynamics of surface water but also for other applications e.g., flood hazard zonation and developmental planning, etc. The study demonstrates the potential of cloud based online geo-processing platform in analyzing the time-series data from multiple satellites without any cost implication towards data procurement and processing; the scalability for large-area application may be explored further.