A coupled model of greenhouse gas emissions from erosion and accretion prone zones of mangrove ecosystem, Sundarban, India

• Published in: Geoderma Regional Vol. 40; p. e00928
• Main Authors: Das, Nilanjan, Pal, Debarati, Chakrabortty, Rabin, Pal, Subodh Chandra, Mandal, Sudipto
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.03.2025
• Subjects: aquaculture; Bay of Bengal; carbon dioxide; Climate variables; deforestation; digital elevation models; estuaries; General Circulation Models; geographical distribution; greenhouse gas emissions; greenhouse gases; India; land cover; Landsat; mangrove ecosystems; Mangroves; neural networks; radar; rain; Random forest model; regression analysis; risk; runoff; RUSLE model; Soil erosion; soil quality; topography; Bay of Bengal; India; Random forest model; Mangroves; RUSLE model; Climate variables; Soil erosion
• ISSN: 2352-0094; 2352-0094
• DOI: 10.1016/j.geodrs.2025.e00928

Soil erosion (SE) and accretion process resulting from anthropogenic and natural causes has a substantial impact on soil quality and functionality, thereby influencing greenhouse gas (GHG) emissions. Here, the determination of the extent of SE, resultant accretion and emissions of GHGs like CO2, CH4, and N2O was quantified from the Sundarban mangrove ecosystem (SME). A Random Forest (RF) spatial model was proposed to predict the geographical distribution of GHG emissions throughout the Sundarban. Gas samples were gathered from the mangrove bed in 2022 using the enclosed box technique. Next, the Shuttle Radar Topographic Mission Digital Elevation Model (SRTM DEM), and Landsat 8 data were used to produce thematic inputs for the RUSLE model in a GIS platform. Both models were coupled to observe the match between erosion and GHG emission. The result showed that coastal most western part of Sundarban is prominent for CO2 (28.29–31.29 mmol m−2 d−1) and CH4 (0.281–0.329 mmol m−2 d−1) emission while N2O fluctuated more (0.137–0.169 mmol m−2 d−1) at central eastern part due to high deforested agriculture and aquaculture practices. The study revealed that the islands of Sundarban, located at the edge of the Bay of Bengal (BoB) have an increased risk of SE (>12 t ha−1 yr−1) as these islands encounter high oceanic water surges and cyclones yearly. The accuracy of the models were adjudged by the estimation of SE and GHG emission. The measured precision and area under the curve of receiver operating characteristics was 0.796 for RUSLE and 0.784 for RF models, respectively. An “Automated Linear Regression (ALR) model” showed that N2O was the most sensitive (Normalized Importance: 0.55) to erosion. Regression results showed association between GHGs and erosion were weakly correlated (R2 for CO2, CH4, and N2O were 54.79 %, 43.51 %, and 55.08 %). According to the “Artificial Neural Network (ANN) model, rainfall and runoff erosivity factor (R)” was the prime governing factor (normalized importance 100 %) for SE. The “RUSLE model” with “Coupled Model Intercomparison Project-6 (CMIP6)” “rainfall data”, a “Global Circulation Model (GCM)” and the historical climatic data from 1960 to 1989 revealed the CO2, CH4 and N2O emissions would be 7.13 %, 39.25 %, and 38.18 % respectively by 2100. Accretion phenomena was more on the upstream regions of the Sundarban estuary. The work will help the environmental managers in identifying the erosion-prone zones, thereby reducing the land cover changes for anthropogenic benefits that promoted SE. [Display omitted]