西北林学院学报2024，Vol.39Issue(2)：115-122,8.DOI:10.3969/j.issn.1001-7461.2024.02.15
基于高时空分辨率数据的上栗县植被NPP估算及分析
Estimation and Analysis of Vegetation NPP in Shangli County Based on Remote Sensing Data with High Spatial and Temporal Resolution
摘要
Abstract
In view of the lack of geospatial details in the pixel-level research results of vegetation net prima-ry productivity(NPP),this paper used the deep learning model to obtain accurate geo-like patches with the help of medium and high-score remote sensing images,and used the rasterization results as input parame-ters of the improved CASA model,and finally estimated the plot-level NPP values of different vegetation types in Shangli County of Jiangxi Province.The research results showed that 1)compared with the tradi-tional method of extracting only image spectral features,the ground type map obtained by deep learning technology was more accurate.Based on the forest map spot,combined with the middle division image to identify the forest land type,the classification accuracy of forest land was 91.313 4％,indicating that the combination of medium and high-resolution remote sensing images can better carry out the fine classifica-tion of vegetation at the district and county scales.2)Based on the theory of CASA model,the value of the maximum light-year utilization rate in the model was corrected.The rasterization results of the terrain pat-terns were used as the input parameters of the model,and the influence of non-vegetated areas such as built-up areas,roads,and bare land on the model calculation was eliminated,and the estimation results were compared and verified with other model estimation results,which proved the accuracy of the experimental results.3)The results has realized the refined expression of Shangli County NPP results in space,and the research results have good spatial detail characteristics,which not only meet the simple needs of area statis-tics and qualitative analysis,but also provide objective and quantitative data support for subsequent carbon cycle and carbon source/sink research.关键词
深度学习/改进的CASA模型/地类图斑/NPPKey words
deep learning/improved CASA model/land type patch/NPP分类
农业科技引用本文复制引用
罗露花,陈铭杰,杨树文,张新..基于高时空分辨率数据的上栗县植被NPP估算及分析[J].西北林学院学报,2024,39(2):115-122,8.基金项目
国家自然科学基金(42161069). （42161069）
