Assessment of Landsat-8 and Sentinel-2 Imagery for Estimation of Aboveground Biomass and Carbon Stocks in Chure Forests of Nepal

Abstract

Forests play a critical role in global carbon cycles and biodiversity conservation. Quantifying forest aboveground biomass (FAGB) helps in assessing carbon emission and sequestration and can reduce uncertainty in monitoring global carbon cycles and climate change. Remote sensing techniques have proved to be a cost-effective way to estimate FAGB with timely and repeated observations and recommended by United Nations Framework on Climate Change (UNFCC) to assess FAGB and carbon stock for Reduction of Emissions from Degradation and Deforestation Program (REDD+). This research presents an assessment of freely available Landsat-8 and Sentinel-2 imagery for estimating FAGB in the Chure forests of Nepal using Multiple Regression Analysis. Sentinel-2 and Landsat-8 missions are similar, but Sentinel-2 has higher spatial resolution and collects data from the red-edge region of the electromagnetic spectrum while Landsat has one of the largest temporal ranges of imagery range from 1970s. Twelve variables derived from Landsat-8 and fifteen variables derived from Sentinel-2 imageries were used in the study. Forest inventory data of 225 plots collected by the Forest Research and Training Centre (FRTC) in 2017 and 2018 through field measurements, were used to train and validate the model. The correlation of FAGB measured in each plot and variables extracted from the Landsat-8 and Sentinel-2 optical imageries were assessed by the Pearson correlation coefficients. Multiple Linear Regression was applied based on chosen variables to develop the models for estimating FAGB for the whole study area. The R-squared values of 0.78 and 0,68 and standard error of estimate 45.98 and 40.29 were obtained respectively for the Sentinel-2 and Landsat-8 based estimation models. The estimated results were validated by considering R² and RMSE values between observed and estimated FAGB. The values of R² and RMSE between observed and estimated FAGB were 0.70 and 45.02 tons/ha and 0.77 and 34.33 tons/ha for Landsat-8 and Sentinel-2 images respectively. The results of the study showed that both Sentinel-2 and Landsat-8 imageries were viable for FAGB and carbon stocks estimation. However, with high R² and low RMSE value Sentinel-2 based model outperforms the Landsat-8 based model and it seems to be better suited for FAGB estimation in Chure area for the studied year. Overall, the research highlights the potential of Sentinel-2 and Landsat-8 imagery for FAGB estimation and emphasizes the importance of utilizing multi-source data and advanced modelling techniques for accurate and reliable FAGB mapping. The study also revealed that Chure forests are of considerable significance as they store substantial amounts of carbon despite disturbance from different anthropogenic activities.