An alternative approach for estimating above ground biomass using Resourcesat-2 satellite data and artificial neural network in Bundelkhand region of India Dibyendu Deb & J. P. Singh & Shovik Deb & Debajit Datta & Arunava Ghosh & R. S. Chaurasia Received: 6 July 2017 /Accepted: 12 October 2017 # Springer International Publishing AG 2017 Abstract Determination of above ground biomass (AGB) of any forest is a longstanding scientific endeav- or, which helps to estimate net primary productivity, carbon stock and other biophysical parameters of that forest. With advancement of geospatial technology in last few decades, AGB estimation now can be done using space-borne and airborne remotely sensed data. It is a well-established, time saving and cost effective technique with high precision and is frequently applied by the scientific community. It involves development of allometric equations based on correlations of ground- based forest biomass measurements with vegetation indices derived from remotely sensed data. However, selection of the best-fit and explanatory models of bio- mass estimation often becomes a difficult proposition with respect to the image data resolution (spatial and spectral) as well as the sensor platform position in space. Using Resourcesat-2 satellite data and Normalized Difference Vegetation Index (NDVI), this pilot scale study compared traditional linear and nonlinear models with an artificial intelligence-based non-parametric technique, i.e. artificial neural network (ANN) for for- mulation of the best-fit model to determine AGB of forest of the Bundelkhand region of India. The results confirmed the superiority of ANN over other models in terms of several statistical significance and reliability assessment measures. Accordingly, this study proposed the use of ANN instead of traditional models for deter- mination of AGB and other bio-physical parameters of any dry deciduous forest of tropical sub-humid or semi- arid area. In addition, large numbers of sampling sites with different quadrant sizes for trees, shrubs, and herbs as well as application of LiDAR data as predictor variable were recommended for very high precision modelling in ANN for a large scale study. Keywords Above ground biomass . Allometric equation . Artificial neural network . Normalized difference vegetation index . Satellite image Introduction Above ground biomass (AGB) is widely considered to be a key indicator of forest vegetal health and related seral stages (Brown et al. 1997; Yen 2015; Luo et al. 2017). In spite of the fact that direct measurement of AGB of a forest area is the most accurate technique with respect to other ones, it is considerably laborious, time consuming, and expensive at one end as well as involves Environ Monit Assess (2017) 189:576 https://doi.org/10.1007/s10661-017-6307-6 D. Deb : J. P. Singh : R. S. Chaurasia Indian Grassland and Fodder Research Institute, Gwalior Road, Jhansi 284 003, India S. Deb (*) Department of Soil Science and Agricultural Chemistry, Uttar Banga Krishi Viswavidyalaya, Cooch Behar 736 165, India e-mail: shovikiitkgp@gmail.com D. Datta Department of Geography, Jadavpur University, Kolkata 700032, India A. Ghosh Department of Agricultural Statistics, Uttar Banga Krishi Viswavidyalaya, Cooch Behar 736 165, India