Vol. 3, Issue 5 (2016)
Forest carbon stock assessment at Barkot Flux tower Site (BFS) using field inventory, Landsat-8 OLI data and geostatistical techniques
Author(s): T Watham, SPS Kushwaha, S Nandy, NR Patel, S Ghosh
Abstract: Quantification of forest biomass is of vital importance to assess productivity - a critical information for carbon budget accounting, carbon flux monitoring and for understanding the forest ecosystem response to climate change. In this present study, we used field measured aboveground biomass (AGB), Landsat 8 OLI (operational land imager) derived variables and geostatistical tools for spatial total biomass and carbon stock mapping surrounding Barkot Flux Site (BFS), Uttarkhand, India. For this purpose, different AGB prediction maps were produced using on Ordinary Kriging (OK), Universal Kriging (UK), Co-Kriging (CoK) and Regression Kriging (ReK) methods and tested the models’ accuracy. Biomass estimated using OK and UK had root mean square error (RMSE) of 121.78 and 139.48 Mg ha-1, respectively. Total 16 variables were tested one by one as an auxiliary variable in the CoK technique. CoK with Land Surface Water Index (LSWI) had the lowest RMSE of 58.77 Mg ha-1 (R2=0.63). LSWI performed best because of its sensitivity to leaf moisture. Also, ReK method was tested using top three variables (based on RMSE value) achieved in CoK method. However, ReK was not able to improve the accuracy, attained by CoK. This may be due to high spectral variability or the limitation of the typical OK method. Under performance by OK and the UK must have been due to their prerequisite of a large number of well-distributed sample points to capture adequately the spatial variability. Therefore, selection of site-specific suitable variables and method can help in improving the accuracy of biomass assessment. Hence, CoK method with LSWI as an auxiliary was considered for biomass and carbon stock estimation. The total carbon stock (47 % of the total biomass) in the study area was estimated to be 2240797.37 Mg C, with an average of 276.13 Mg C ha-1. Furthermore, the estimated spatial biomass/carbon will be useful in complementing Eddy Covariance (BFS) and remote sensing based carbon dynamics studies being carried out in the same study area.