常用植被物候提取方法 (TIMESATE/R语言/Python)-4.0
文章内容仅用于自己知识学习和分享,如有侵权,还请联系并删除 :)
- 常用植被物候提取方法 (TIMESATE/R语言/Python)-1.0见 link
- 常用植被物候提取方法 (TIMESATE/R语言/Python)-2.0见 link
- 常用植被物候提取方法 (TIMESATE/R语言/Python)-3.0见 link
两篇文章一篇用于sentinel2,一篇用于Planetscope,使用了同一种物候提取算法。
- 下面首先介绍两篇论文
- 然后介绍物候提取算法的主要流程
- 最后介绍论文中可以借鉴的物候提取code和code逻辑
----------------------------------------------持续更新分线-----------------------------------------------------------------------------------------------------
1. 论文
【1】 Bolton D K, Gray J M, Melaas E K, et al. Continental-scale land surface phenology from harmonized Landsat 8 and Sentinel-2 imagery[J]. Remote Sensing of Environment, 2020, 240: 111685.
【2】Moon M, Richardson A D, Milliman T, et al. A high spatial resolution land surface phenology dataset for AmeriFlux and NEON sites[J]. Scientific Data, 2022, 9(1): 448.
2. 算法流程
step1: 影像预处理。主要指去除掉有云覆盖等异常的像素
step2: 创建daily EVI2时间序列
step2-1: Outlier elimination
step2-2: Generating daily time series of EVI2 (见下图)
step3: Identifying phenological cycles (见下图a)
step4: Retrieving LSP metrics (见下图b)
3. R代码
1. 代码链接:Python and R source code to download and process the PlanetScope imagery and generate the product can be obtained through a public repository at: link
2. 代码简介:该代码主要以Planetscope为例,输入的数据是Planetscope data,输出的数据格式是netcdf。
- 详细介绍见参考文献 Moon M, Richardson A D, Milliman T, et al. A high spatial resolution land surface phenology dataset for AmeriFlux and NEON sites[J]. Scientific Data, 2022, 9(1): 448.
(1)00和01:用来进行Planetscope数据的下载和预处理
(2)02: 创建chunck提高运行效率,A script for PlanetScope image process; save mosaiced images into chunks
(3)03: 提取物候矩阵,A script for estimating phenometrics
- 这部分是主函数,分块提取提取物矩阵
for (i in 1:numPix){pheno_mat[i,] <- DoPhenologyPlanet(band1[i,],band2[i,],band3[i,],band4[i,],dates,phenYrs,params,waterMask[i])if(i%%10000==0) print(i)
}
- DoPhenologyPlanet函数的主体在PLSP_Functions.R中
- 输入输出路径和变量名称需要在PLSP_Parameters.json和PLSP_Layers.csv中修改
(4)04和05:用来创建geotiff文件和生成netcdf文件
- 04: A script for saving data layers into GeoTiff format
- 05: A script for saving data layers into netCDF format。
- 输出的每个netcdf有24个图层,每个图层的介绍详细见参考文献[2]