Efficient Earth Observation Data Representation for Scalable Clustering

Abstract

In the recent scenario of Earth Data, information on a broad variety of topics (e.g.,Population, land use, agriculture, soils, climate) is growing rapidly in five eras i.e. volume, variety, value, velocity and veracity. These data exceeds their boundary in form of storage, manage, compute, access and analysis called Big Data. Moreover, the fast growth rate of such large data generates numerous challenges, such as the rapid access growth of value, transfer speed, diverse data, processing speed and security. However, there exist compact data structure based techniques which directly access the compressed data without the need of decompressing it and also indexing its content. In this work, we have presented the compact data structure based technique. We applied this technique to OCM2-NDVI GeoTIFF single and stacked image to develop compressed dataset for efficient representation. Then we applied scalable clustering on compressed datasets. However, when the datasets have larger values developed compact data structure work very well in both the compact representation of data as well as on scalable clustering.