Spectroscopy and Spectral Analysis, Volume. 41, Issue 9, 2911(2021)

Leaf Nitrogen Concentration Retrieval Based on Polarization Reflectance Model and Random Forest Regression

Zi-han ZHANG1、*, Lei YAN1、1; 2;, Si-yuan LIU1、1;, Yu FU1、1;, Kai-wen JIANG1、1;, Bin YANG3、3;, Sui-hua LIU4、4;, and Fei-zhou ZHANG1、1; *;
Author Affiliations
  • 11. Beijing Key Lab of Spatial Information Integration and 3S Application, Institute of Remote Sensing and Geographic Information System, School of Earth and Space Science, Peking University, Beijing 100871, China
  • 33. College of Electrical and Information Engineering, Hunan University, Changsha 410082, China
  • 44. School of Geography and Environmental Science, Guizhou Normal University, Guiyang 550001, China
  • show less
    Figures & Tables(8)
    Spatial distribution of foliar nitrogen content retrieval plot
    Flowchart of data utilization
    Ratio of PBRF to BRFIn this figure, the × in the box represents the mean value, and the transverse line in the box represents the median value
    • Table 1. IGBP classes of dominant vegetation in Bavarian Forest National Park

      View table
      View in Article

      Table 1. IGBP classes of dominant vegetation in Bavarian Forest National Park

      树种IGBP类别类名
      欧洲云杉IGBP01常绿针叶林
      欧洲山毛榉IGBP04落叶阔叶林
      欧洲冷杉IGBP01常绿针叶林
      假挪威槭IGBP04落叶阔叶林
      欧洲花楸IGBP04落叶阔叶林
    • Table 2. Formulas of 5 BPDF models

      View table
      View in Article

      Table 2. Formulas of 5 BPDF models

      模型名称公式应用场景自由参数
      Nadal模型PBRF=A1-exp-BFp(γ,N)μS+μV森林、 灌木林和低植被地表覆盖类型A, B
      Waquet模型PBRF=AFp(γ,N)S(θS)S(θV) [其中,S(θ)为后向散射中最大值的遮蔽函数]森林、 作物种植地和城市地表覆盖类别A, B
      Maignan模型PBRF=Aexp(-tanαI)exp(-NDVI)Fp(γ,N)4(μS+μV)植被地表覆盖类型A
      Litvinov模型PBRF=AπFp(γ,N)4(μS+μV)cosθHS(γ,C)f(B,θH) [其中,$f(B,{θ}_{H}) $表征的是地表在空间中的高斯分布,S(γ,C)是遮蔽函数]植被和土壤地表覆盖类型A, B, C
      Diner模型PBRF=AFp(γ,N)8πμSμVcosθH草地地表覆盖类型A
    • Table 3. Accuracy assessment of 5 BPDF models on IGBP01/04/05

      View table
      View in Article

      Table 3. Accuracy assessment of 5 BPDF models on IGBP01/04/05

      类别Nadal模型Waquet模型Maignan模型Litvinov模型Diner模型
      RSQRMSERSQRMSERSQRMSERSQRMSERSQRMSE
      IGBP010.6970.0030.6900.0030.6950.0030.6880.0030.6550.003
      IGBP040.8750.0020.8570.0020.8710.0020.8780.0020.8210.002
      IGBP050.8120.0020.7890.0020.8180.0020.8080.0020.7590.002
    • Table 4. RF algorithm accuracy assessment of different estimator numbers

      View table
      View in Article

      Table 4. RF algorithm accuracy assessment of different estimator numbers

      决策树
      数量
      去偏前去偏后
      RSQRMSERSQRMSE
      50.7220.3070.7470.300
      100.7690.2900.8030.252
      500.7320.3030.7460.292
      1000.7200.3090.7290.303
    • Table 5. Accuracy assessment of 5 LNC retrieval algorithms

      View table
      View in Article

      Table 5. Accuracy assessment of 5 LNC retrieval algorithms

      算法名称去偏前去偏后
      RSQRMSERSQRMSE
      PLSR0.5950.3730.6240.359
      PCR0.6780.3290.6800.328
      SVR0.7320.3040.7350.302
      KNN0.7480.2910.7480.291
      RF0.7690.2900.8030.252
    Tools

    Get Citation

    Copy Citation Text

    Zi-han ZHANG, Lei YAN, Si-yuan LIU, Yu FU, Kai-wen JIANG, Bin YANG, Sui-hua LIU, Fei-zhou ZHANG. Leaf Nitrogen Concentration Retrieval Based on Polarization Reflectance Model and Random Forest Regression[J]. Spectroscopy and Spectral Analysis, 2021, 41(9): 2911

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Research Articles

    Received: Sep. 26, 2020

    Accepted: --

    Published Online: Oct. 29, 2021

    The Author Email: ZHANG Zi-han (zzh_cytus@pku.edu.cn)

    DOI:10.3964/j.issn.1000-0593(2021)09-2911-07

    Topics