Evaluation of four algorithms for estimation of canopy cover of mangrove forests by using aerial imagery

Today, it is important to use the ecological indicators, such as canopy cover for recognizing the special status of ecosystems, such as mangrove forests and also monitoring and evaluating changes through a specific period. This study aimed to investigate the sufficiency of parametric and nonparametric algorithms using the spectral data with high spatial resolution in the evaluation of canopy cover in the mangrove forest in the Bushehr province. The vegetative characteristics were studied at 20×20 square meter sample plots. 50 Sample plots were studied for the proposed vegetative characteristic such as diameter, Height and percentage of canopy cover of mangrove forest. The camera UltraCamX digital images which used in this study were harvested to the shooting operation on 2012/01/10. After conducting some proper Preprocessing and processing, the digital values corresponding to the ground samples were extracted from spectral bands and were considered as the independent variables while and the crown canopy percent per plot were considered as the dependent variable. Modeling was carried out based on 75 percent of sample plots using K-Nearest Neighbor methods, support vector machine, random forest and General linear model methods and the results were cross-validated using the remaining 25 percent. The results showed that the best estimates were obtained from the crown canopy percent with method Random Forest, k-NN, SVM and General linear model methods with a root mean square error of 13.57, 13.95, 14.88 and 17.73 percent and relative bias of -3.88, -4.62, -5.05 and -2.88 percent that Random Forest method had the best performance. The results of this study showed UltraCam X Arial spectral data had the high ability for estimating of canopy cover percent.