Three-Decade Land Use Transitions in a Coastal Deltaic Region: Integrating Google Earth Engine and Random Forest Classification for Multi-Temporal Change Detection

Three-Decade Land Use Transitions in a Coastal Deltaic Region: Integrating Google Earth Engine and Random Forest Classification for Multi-Temporal Change Detection | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 8 January 2026 V1 Latest version Share on Three-Decade Land Use Transitions in a Coastal Deltaic Region: Integrating Google Earth Engine and Random Forest Classification for Multi-Temporal Change Detection Authors : Md. Jubayer Ahamed [email protected] , Pranto Kumar Sarker 0009-0009-9577-7093 , and Sujoy Dey 0009-0008-5281-4467 Authors Info & Affiliations https://doi.org/10.22541/au.176789260.02273034/v1 114 views 95 downloads Contents Abstract Supplementary Material Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Coastal Bangladesh represents one of the world's most dynamic and vulnerable deltaic landscapes, experiencing intensive anthropogenic pressures from agricultural expansion, aquaculture development, urbanization, and climate-related environmental changes. The southwestern coastal region, encompassing districts adjacent to the Sundarbans mangrove forest, faces particularly acute land use pressures driven by population growth, economic development, and integration into global supply chains. Despite the critical importance of this region for ecosystem services, food security, and climate resilience, comprehensive multi-decadal assessments of land use and land cover (LULC) change patterns across large spatial extents remain limited. This study quantifies LULC dynamics across 11,700 km² in southwestern coastal Bangladesh over three decades (1995, 2010, 2025), employing multitemporal satellite imagery, machine learning classification, and rigorous transition analysis.​ We processed cloud-free Landsat imagery from multiple sensors (Landsat 5 Thematic Mapper, Landsat 7 ETM+, Landsat 8 OLI, and Landsat 9 OLI-2) in Google Earth Engine, applying strict quality filters (cloud cover <3%) and selecting images from the dry season (November-December) to ensure temporal consistency. Spectral indices, including Normalized Difference Vegetation Index (NDVI), Modified Normalized Difference Water Index (MNDWI), Modified Soil Adjusted Vegetation Index (MSAVI), and Random Forest classification, were implemented with six LULC classes: agricultural land, closed waterbodies, seasonal grazing land, mangrove forest, river/canal, and settlement. Informative bands and spectral indices have led to achieving robust classification accuracy. Classification achieved high accuracy across all periods: 99.7% overall accuracy (Kappa = 0.99) for 1995, 98.8% (Kappa = 0.96) for 2010, and 99.9% (Kappa = 0.99) for 2025, validated using stratified sampling. Producer's accuracies ranged from 90.3% to 99.9% across land cover classes, with user's accuracies between 92.1% and 99.9%, demonstrating robust discrimination between categories. Hyperparameter tuning via RandomizedSearchCV optimized model performance, with validation conducted using stratified train-test splits. Post-classification analysis included transition matrix computation, swap analysis following the methodology of Pontius et al. (2004) to distinguish between location and quantity changes, chi-square statistical tests to assess systematic versus random transitions, and mapping of change intensity.​ Results reveal a profound transformation of the landscape over the 30-year period. Agricultural land expanded dramatically from 620.5 km² in 1995 to 1044.8 km² in 2025, representing a 68.4% increase. Annual change rates demonstrate temporal heterogeneity: agricultural expansion occurred at a rate of 4.0% per year from 1995 to 2010, decelerating to 0.3% per year from 2010 to 2025, suggesting potential land availability constraints or policy interventions may have contributed. Settlement areas grew from 421.4 km² to 576.3 km² (36.8% increase), with relatively consistent annual rates of 1.8% and 0.5% for the two periods, respectively. In contrast, seasonal grazing lands experienced severe contraction from 1893.20 km² to 982.60 km² (48.1% decline), resulting in a loss of area at rates of 2.9% per year (1995–2010) and 0.6% per year (2010–2025). Mangrove forests exhibited remarkable stability, maintaining approximately 4,082-4,096 km² with only 0.3% net change over 30 years. River and canal systems expanded by 8.7% from 3,418.2 km² to 3,715.8 km², with accelerating rates (0.09% per year increasing to 0.48% per year), potentially reflecting river channel migration, erosion, or the development of aquaculture ponds. Closed waterbodies showed non-monotonic trends, initially increasing 24.1% (1,265.0 km² to 1,569.8 km²) and then declining 17.1% (to 1,300.9 km²), suggesting complex interactions between shrimp pond expansion, abandonment, and hydrological changes.​ Chi-square tests yielded values of 20,037.5 (1995–2010) and 20,808.9 (2010–2025), both with p<0.05, confirming that observed transitions represent systematic directional changes rather than random fluctuations. Transition matrices reveal that agricultural land primarily originated from seasonal grazing lands (391.0 km²) and closed waterbodies (243.1 km²) during 1995-2010, while settlement expansion occurred predominantly at the expense of seasonal grazing lands (188.7 km²) and agricultural areas (60.8 km²).​ These findings document the accelerated intensification of agriculture and urbanization in southwestern coastal Bangladesh, driven by population pressure, economic incentives for shrimp aquaculture and rice intensification, and shifting livelihood strategies. The dramatic decline in seasonal grazing lands raises concerns for pastoral communities, livestock-based livelihoods, and ecosystem services, including carbon storage and flood buffering. Mangrove stability at the landscape scale can mask localized dynamics, requiring targeted conservation interventions. Results provide quantitative evidence for land-use planning, coastal zone management, and climate adaptation strategies in vulnerable deltaic regions worldwide. Session: IN51C. Advancing Artificial Intelligence for Remote Sensing: Overcoming Data Scarcity and Domain Shift III Poster Presentation Type: Poster Final Poster Number: IN51C-VR8981 Day/Time: Friday, 19 December 2025: 08:30 - 12:00, NOLA CC, Hall EFG (Poster Hall) Supplementary Material File (land_use.pdf) Download 13.60 MB Information & Authors Information Version history V1 Version 1 08 January 2026 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords coastal region hyperparameter tuning land use land cover landsat random forest Authors Affiliations Md. Jubayer Ahamed [email protected] View all articles by this author Pranto Kumar Sarker 0009-0009-9577-7093 View all articles by this author Sujoy Dey 0009-0008-5281-4467 View all articles by this author Metrics & Citations Metrics Article Usage 114 views 95 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Md. Jubayer Ahamed, Pranto Kumar Sarker, Sujoy Dey. Three-Decade Land Use Transitions in a Coastal Deltaic Region: Integrating Google Earth Engine and Random Forest Classification for Multi-Temporal Change Detection. Authorea . 08 January 2026. DOI: https://doi.org/10.22541/au.176789260.02273034/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.176789260.02273034/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9fe4dfd96eb1ad07',t:'MTc3OTIxMjI5Nw=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();