Spatiotemporal evolution and uncertainty analysis of mangrove ecosystem carbon storage based on multi-source remote sensing and Monte Carlo error propagation: a case study of the mangrove reserves in Hainan Province
Abstract
Mangrove forests are the planet's most carbon-dense coastal blue-carbon ecosystem, which plays an important role in global forest carbon sequestration capacity. With the background of the China policy of a dual-carbon, this research provides a multidimensional evaluation of the carbon-stock dynamics of mangroves on Hainan Island of China, in the period between 2000 and 2024. This study used winter cloud-free 30 m time-series dataset spanning 2000–2024 constructed on the Google Earth Engine platform using Landsat-5/7/8/9 and Sentinel-2 MSI imagery. Above- and below-ground biomass carbon were co-retrieved by coupling an NDVI-AGB allometric growth equation with root-to-shoot ratios and carbon-conversion coefficients. A Monte Carlo-Bayesian framework was embedded to explicitly propagate uncertainties and attribute variance via Sobol decomposition. The results revealed a ‘decline-recovery-net-increase’ trajectory, having total carbon storage declining from 37.09 ± 8.01 TgC in 2000 to 27.79 ± 4.16 TgC in 2004 while rebounding sharply thereafter, and reaching 68.97 ± 21.69 TgC in 2024, representing a net increase of 85.7% over 24 years. The propagation of errors at a fine spatial resolution led to an expansion of estimation uncertainty, with the 95% confidence interval widening from 26.61–52.05 TgC in 2000 to 41.46–110.05 TgC in 2024.