Monitoring And Spatio-temporal Characteristics Of Rubber Plantations Using Sentinel-2 In The China-Laos-Myanmar Border Region

Since the 1990 s,rubber plantations with agroforestry attributes have rapidly expanded in northern Indochina,exerting a series of impacts on the region’s climate,carbon storage,hydrological processes,biodiversity,and rural livelihoods.It goes without saying that accurate spatial distribution information of rubber plantations is crucial for formulating effective forest management strategies.In recent years,the burgeoning rubber plantations have undergone some new changes,but their dynamic characteristics are still unclear.In particular,rubber plantations in Indochina are mainly planted by smallholder farmers,with small and scattered plot sizes,especially in mountainous areas.Satellite data with relatively low spatial resolution,such as30-m Landsat,commonly used for rubber plantations remote sensing monitoring in the past,have not been able to comprehensively and objectively reveal their detailed spatial distribution information,and fine-grained dynamic monitoring is insufficient.This study selected the region with frequent rubber plantations expansion and cross-border development-the China-Laos-Myanmar border region-as the study area,based on Sentinel-2 images and the unique phenological characteristics of rubber plantations in tropical regions,improved and perfected a set of phenological algorithms for rubber plantations remote sensing monitoring,extracted the spatial distribution information of rubber plantations in the China-Laos-Myanmar border region from 2017 to 2022,clarified the latest dynamic characteristics of rubber plantations,and analyzed the interrelationships between rubber plantations distribution and altitude,slope,and aspect.In addition,this study systematically investigated the distribution patterns and spatiotemporal change characteristics of rubber plantations along roads,rivers,and border lines(3 “A”),and compared and analyzed the national differences in the dynamic changes of rubber plantations in the China-Laos-Myanmar border region.The main conclusions are as follows:(1)Among the 20 spectral indices selected,most vegetation-moisture indices show potential for distinguishing between rubber plantations and natural forests.Among them,the pure red-edge spectral index,RESI,has the strongest ability to identify rubber plantations,while partial red-edge indices,such as NDVIre1 n,have the best recognition performance.Among non-red-edge indices,RGNDI and NBR can effectively identify rubber plantations.Combining the extracted rubber plantations distribution information with accuracy validation,RGNDI is determined as the optimal index for identifying rubber plantations.Based on this,a new Re NDI phenological algorithm is constructed based on multiple time window for remote sensing monitoring of rubber plantations in the study area.(2)From 2017 to 2022,the overall trend of rubber plantations area has slightly increased in the China-Laos-Myanmar border region,but there are significant national differences.The rubber plantations area in Mengla County and Jinghong City in China has decreased more significantly,while the rubber plantations in the border regions of Laos and Myanmar,especially those near China,such as the western part of Phongsaly Province,the northern part of Luang Namtha Province and Oudomxay Province,and the eastern part of Mongphat County,are expanding.It is worth noting that rubber plantations in the protected areas on the China side are continuously decreasing,while rubber plantations in the protected areas in Laos are expanding significantly.(3)Elevation and slope have important impacts on the planting and expansion of rubber plantations.Rubber plantations are mainly distributed in areas with elevations between 600 and 1000 m(about 80%)and slopes less than 25°(over 90%).Compared with Laos and Myanmar,rubber plantations on the China side are distributed in areas with higher elevations and steeper slopes,but rubber plantations in Laos and Myanmar are also expanding to steep slopes.Slope aspect has not had a significant impact on rubber plantation planting.The rubber plantations area on south-facing slopes,southeast-facing slopes,and southwest-facing slopes is slightly higher than that on north-facing slopes and other slope aspects.(4)The areas near roads and rivers,especially within 2 km along their courses,are the main regions for rubber plantation distribution.Rubber plantations are preferably planted in areas close to roads,while areas near rivers are more suitable for rubber plantations due to their high soil moisture content and ease of irrigation.Additionally,the expansion of rubber plantations in Laos and Myanmar exhibits an evident edge effect and is largely influenced by China,which tends to favor planting rubber plantations in areas close to the China-Laos and China-Myanmar borders,particularly in the case of Myanmar.This study developed a phenology algorithm based on Sentinel-2 images,which enriched the remote sensing mapping method of rubber plantations at the regional level and generated precise and detailed information on rubber plantations.This not only provides new insights into the dynamic changes of rubber plantations(e.g.,decrease),but also helps with the rational planning of future rubber plantations,forest management,and biodiversity conservation,particularly in the tropics.