Assessment of the FCD model for estimating canopy density in Zagros forests using Landsat 8 and Sentinel 2 images
Subject Areas : Environmental managment
Ahmad abbasiwand
1
,
Hormoz Sohrabi
2
1 - MSc. Graduated, Forest Science and Engineering, Faculty of Natural Resources, Tarbiat Modares University, Tehran, Iran.
2 - Associate Professor, Forest Science and Engineering, Tarbiat Modares University, Tehran, Iran. *(Corresponding Author)
Keywords: Forest Canopy Density, Zagros Forests, Sentinel 2, Landsat 8.,
Abstract :
Background and Objective: One of the important criteria for studying, assessing changes, and monitoring forests is the canopy cover. Considering the role of Zagros forests in water and soil conservation and the significance of canopy cover in these forests, timely mapping canopy cover and cover changes are of special importance.
Material and Methodology: this research selected a total of 27 study areas throughout the Zagros growth region. A total of 2637 plots of 400 meters were examined, and the actual percentage of canopy cover was extracted. To estimate the canopy cover for Landsat 8 and Sentinel-2 satellites, data from the calculation of four integrated indices were combined, and the Forest Canopy Density (FCD) map was calculated. The model's performance was evaluated using statistical error metrics. Pearson correlation was used to assess the correlation between Landsat 8 and Sentinel 2 derived FCD indices and measures canopy cover.
Findings: The results showed that AVI and SI, had a direct relationship, while BI and TI had an inverse relationship with the canopy cover in all regions. Moreover, the highest coefficient of determination among all regions was obtained for the Sentinel-2 satellite (86.64%) in the densely forested area of northern Zagros. The lowest statistical error values, including RMSE (17.41%), RMSE% (31.57%), Bias (1.57%), Bias% (2.70%), MAE (12.60%), and MAE% (24.28%), were all obtained for the Sentinel-2 satellite and areas with dense canopy cover. Additionally, the statistical error metrics in northern Zagros showed lower values compared to central and southern Zagros, indicating the model's efficiency in densely forested areas.
Discussion and Conclusion: In general, it was not possible to achieve canopy estimation with high accuracy by the FCD model. The results obtained from the Sentinel-2 satellite performed better than the Landsat 8 satellite. Overall, this model is not suitable for the estimation of canopy covers of sparser semi-dense forests in Zagros.
