Estimating the Forest Carbon Storage of Chongming Eco-Island, China, Using Multisource Remotely Sensed Data

Urban forests are highly heterogeneous; information about the combined effect of forest classification scale and algorithm selection on estimation accuracy for urban remains unclear. In this study, we chose Chongming eco-island in mega-city Shanghai, a national experimental carbon neutral construction plot China, as study object. Remote sensing models (simple regression vs. machine learning models) density were constructed across different scales (all forests, types, dominant tree species) based high-resolution aerial photographs Sentinel-2A remote images, large number field surveys optimal screened by ten-fold cross-validation. The results showed that (1) early 2020, total area storage 307.8 km2 573,123.6 t, respectively, among which areal ratios evergreen broad-leaved forest, deciduous warm coniferous 51.4% 53.3%, 33.5% 32.8%, 15.1% 13.9%, respectively. (2) average was 18.6 t/ha, no differences detected three types (i.e., 17.2–19.2 t/ha), opposite to what observed species 14.6–23.7 t/ha). (3) Compared simple models, an improvement performance all scales, with rRMSE rBias values decreasing 29.4% 53.1%, respectively; compared all-forests scale, algorithms decreased 25.0% 45.2% at forest-type 28.6% 44.3% We concluded refining classification, advanced prediction procedures, could improve estimates forests.