Large Scale Remote Sensing Data Mining for Biomass Monitoring: Recent Advances and Future Challenges Ranga Raju Vatsavai, Varun Chandola, and Budhendra Bhaduri Computational Sciences and Engineering Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA. Email: vatsavairr@ornl.gov, chandolav@ornl.gov, bhaduribl@ornl.gov 1. Introduction Recent advances in remote sensing instrumentation, and commercialization of remote sensing technology has lead to the unprecedented growth in the acquisition and archival of high resolution (spatial, spectral, and temporal) imagery. There is a great demand for analyzing high-resolution (both spatial and temporal) imagery data combined with GIS data layers, in order to extract useful information (e.g., land-use/land-cover) which allows better management of the environment. However, processing these large datasets poses several constraints and challenges. In this paper, we take a closer look at some of these challenges, summarize the recent progress, and also point the community to some of future research challenges. As a case study we use biomass monitoring application at regional and global scales. 2. Biomass Monitoring Monitoring biomass over large geographic regions for identifying changes is an important task in many applications. With recent emphasis on biofuel development for reducing dependency on fossil fuels and reducing carbon emissions from energy production and consumption, the landscape of many countries is going to change dramatically in coming years. Already there are several preliminary reports that address both economic and environmental impacts of growing energy crops. In the United States continuous corn production is becoming a dominant cropping pattern as more and more soybean and wheat rotations are replaced by continuous corn production. It is also expected that more and more pasture lands will be converted to Switchgrass in the coming years, which may positively impact climate change because of its superior carbon uptake properties. These changes are not limited to the United States alone. Developing countries like India, the rural areas are facing increasing demand for energy. It is expected that energy crops like Jatropha curcas are going to be widely planted in Asian countries. Recent FAO report (Richard, 2010) indicates a threefold increase in the area planted to jatropha from 4.72 million ha in 2010 to 12.8 million ha by 2015. In order to understand the changing landscape and complex interactions between biomass and climate, biogeophysical variables on a continuous basis, we need novel techniques for monitoring biomass and scalable algorithms for species-level information extraction from high-resolution images. To meet these challenges, we have developed a novel biomass- monitoring framework consisting of two key change detection algorithms. 2.1 Continuous Change Detection The launch of NASA's Terra satellite in December of 1999, with the MODIS instrument aboard, introduced a new opportunity for continuous monitoring of biomass over large geographic regions. MODIS data sets represent a new and improved capability for terrestrial satellite remote sensing aimed at meeting the needs of global change research (Justice, 1998).