ESTIMATION OF MANGROVE CARBON STOCK WITH HYBRID METHOD USING IMAGE SENTINEL-2

Abstract

Field survey data combined with remote sensing data were an ideal and practical method for
estimating carbon stocks. The objective of this research was to get an estimation model of mangrove carbon
stock with good accuracy. Modeling used hybrid methods, by combining satellite image analysis and field
data. The result of this research was to get the mangrove carbon estimation model. Model 1 merging between
NNIP vegetation index equation using regression of power/geometry and six variables multiple regression
(NDRE or WVVI vegetation index, sediment depth, soil density,% C soil depth 0-15 cm, 15-50 cm and >50
cm). RMSE test resulted 0.4778 t 100 m-2 and % RMSE 16.12%. Model 2 NNIP vegetation index and three
variable regression (VIRRE vegetation index, sediment depth, soil density). RMSE test resulted 0.5639 t 100
m-2 and % RMSE 19.03%. Model 3 uses NNIP vegetation index and two variable regression (NDRE
vegetation index and sediment depth). RMSE test resulted 0.7295 t 100 m-2 and RMSE % 24.63%. Model 4
incorporation of NNIP vegetation index and multiple regression of 3 variables (VIRRE vegetation index,
average sediment depth value 100.63 cm, soil density value 1.02 g cm-3). RMSE test resulted 1.0043 t 100 m2 and % RMSE 33.89%.