Modeling Forest Carbon Estimation Using Sentinel-2 Derived Indices in Yayu Afro-Montane Forest, South West Ethiopia

Abstract Background: Empirical analyses were common methods for forest carbon estimation. Lately, satellite images are popularly used to study different attributes of forest vegetation. Sentinel-2 image provides a significant improvement in spectral coverage, spatial resolution and temporal frequency to assess forest biomass. This study assessed the potential of vegetation indices and biophysical variables derived from Sentinel-2 images in modeling above ground biomass (AGB) and carbon stock in the Yayu forest biosphere reserve. Method: About twenty variables extracted from the Sentinel-2 image were used in this study. Forest stand parameters such as DBH and tree height were used to calculate AGB using allometric equations. The correlation between the biomass values measured from plots and the variables extracted from Sentinel-2 images were examined using the Pearson correlation coefficients. A regression analysis was applied to select determinant variables for predicting AGB. The regression model results were validated based on the coefficients of determination between the observed and the predicted values.Results: A strong correlation (r = 0.65 - 0.74) was found between the biophysical variables from Sentinel-2 image and AGB measured from sampling plots. The multispectral (MS) Band 4, the biophysical vegetation variables from Sentine-2 images were strongly correlated with the AGB. The variables MS Band 4, IRECI, LAI, FCOVER and FAPAR are good predictors of the forest AGB. The model goodness of fit between the observed and predicted values of the AGB showed a coefficient of determination (r2) value of 0.74 and root mean square error (RMSE) of 0.16 ton C/pixel. Conclusion: The developed AGB prediction model was applied to successfully quantify and map the AGB and carbon stock of the forest in the biosphere reserve. Vegetation indices from Sentinel-2 images can effectively predict AGB in forest landscapes and can avoid costly ground surveys to quantify AGB and carbon stock in difficult terrains.