A Study On The Dynamic Change Of Bamboo Forest In Jiangxi Province Based On Landsat Images And Sample Migration

Bamboo forests play an important role in mitigating climate change,providing wildlife habitat and creating benefits,and understanding the dynamic changes in the spatial and temporal distribution of bamboo forests is an important prerequisite for promoting the realization of their ecological values.While traditional field assessment methods are inefficient and costly in terms of human and material resources,remote sensing technology has the advantages of wide coverage,high efficiency,low cost and less restriction by ground conditions,thus it can overcome the problems faced by traditional field measurement methods and achieve scientific mapping and monitoring of this valuable resource of bamboo forests.The reference samples collected in the field determine the accuracy of supervised classification results,and the sample migration method provides a possible solution to the challenge of the difficulty of collecting spatio-temporally consistent reference samples.This study uses Landsat image data for the period 1990-2020 to achieve dynamic mapping of bamboo forests in Jiangxi Province for long time series based on random forest algorithm and spectral similarity algorithm using GEE cloud computing platform,and also quantifies the effects of climatic and topographic factors on the spatial and temporal distribution of bamboo forests using geographic probes.This study demonstrates the potential of Landsat long time series imagery and sample migration methods in the study of spatial and temporal changes in bamboo forests,and the generated results are useful for sustainable management and ecological assessment of bamboo forests in Jiangxi Province.The main results of this study are as follows:(1)The bamboo forest mapping at 30m resolution in Jiangxi Province in 2017 had high accuracy,with an overall precision of 96.19%and the kappa coefficient of 0.92.Vegetation indices among the taxonomic features play an important role in bamboo forest information extraction,with NDMI in summer making the greatest contribution to bamboo forest extraction.(2)The overall precision of the six target years(1990,1995,2000,2005,2010,and2020)was 92.46%,91.68%,92.86%,94.75%,95.85%,and 95.71%,respectively,and the kappa coefficients ranged from 0.81 to 0.91,with F1 scores all greater than 0.85,and the estimated of the areas were in strong agreement with statistics from other sources,proving the feasibility and reliability of the sample migration method.(3)The bamboo forests in Jiangxi Province showed a rapid expansion trend from1990 to 2020.Spatially,bamboo forests are mainly concentrated in the northwestern,northeastern and southwestern mountainous areas of Jiangxi Province,showing an overall expansion from northeast to southwest.In terms of area change,the area of bamboo forests in Jiangxi Province increased from 5755.42 km~2 in 1990 to 12049.47km~2 in 2020,a total increase of 6294.05 km~2 and a 109%increase in the area of bamboo forests.The increase of bamboo forest mainly comes from non-bamboo forest land,and the new bamboo forest usually occurs in low elevation areas and middle and gentle slope areas.(4)The elevation factor in the single factor plays a dominant role in the spatio-temporal distribution of bamboo forests,and the factor interaction shows a two-factor enhancement and a non-linear enhancement,and the interaction of the two factors can strengthen the explanatory power of the spatio-temporal variation of bamboo forests.