• Journal of Infrared and Millimeter Waves
  • Vol. 39, Issue 1, 111 (2020)
Xue-Ling LI1, Ying-Ying DONG2、3、*, Yi-Ning ZHU1, and Wen-Jiang HUANG2、3
Author Affiliations
  • 1School of Mathematical Sciences, Capital Normal University, Beijing00048, China
  • 2Key laboratory of Digital Earth Science, Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing100094, China
  • 3Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing100094, China
  • show less
    DOI: 10.11972/j.issn.1001-9014.2020.01.015 Cite this Article
    Xue-Ling LI, Ying-Ying DONG, Yi-Ning ZHU, Wen-Jiang HUANG. Leaf area index estimation with EnMAP hyperspectral data based on deep neural network[J]. Journal of Infrared and Millimeter Waves, 2020, 39(1): 111 Copy Citation Text show less
    研究区EnMAP遥感数据及产品 (a) 主要作物分布图 (b) LAI产品
    Fig. 1. 研究区EnMAP遥感数据及产品 (a) 主要作物分布图 (b) LAI产品
    深度神经网络基本结构(a)卷积,(b)池化,(c)全连接
    Fig. 2. 深度神经网络基本结构(a)卷积,(b)池化,(c)全连接
    基于SSLLAI-Net的LAI反演流程图
    Fig. 3. 基于SSLLAI-Net的LAI反演流程图
    基于SSLLAI-Net的叶面积指数定量反演散点密度图(a)谷物SSLLAI-Net5000 (b)玉米SSLLAI-Net5000 (c)油菜SSLLAI-Net5000 (d)其他作物SSLLAI-Net5000(e)谷物SSLLAI-Net300 (f)玉米SSLLAI-Net300 (g)油菜SSLLAI-Net300 (h)其他作物SSLLAI-Net300(i)谷物SSLLAI-Net50 (j)玉米SSLLAI-Net50 (k)油菜SSLLAI-Net50 (l)其他作物SSLLAI-Net50
    Fig. 4. 基于SSLLAI-Net的叶面积指数定量反演散点密度图(a)谷物SSLLAI-Net5000 (b)玉米SSLLAI-Net5000 (c)油菜SSLLAI-Net5000 (d)其他作物SSLLAI-Net5000(e)谷物SSLLAI-Net300 (f)玉米SSLLAI-Net300 (g)油菜SSLLAI-Net300 (h)其他作物SSLLAI-Net300(i)谷物SSLLAI-Net50 (j)玉米SSLLAI-Net50 (k)油菜SSLLAI-Net50 (l)其他作物SSLLAI-Net50
    SSLLAI-Net LAI反演精度
    Fig. 5. SSLLAI-Net LAI反演精度
    基于SSLLAI-Net的研究区LAI反演结果(a) SSLLAI-Net5000 (b) SSLLAI-Net 50 (c) LAI差值对比(d) SSLLAI-Net5000局部 (e) SSLLAI-Net 50局部 (f) SSLLAI-Net5000局部 (g) SSLLAI-Net 50局部
    Fig. 6. 基于SSLLAI-Net的研究区LAI反演结果(a) SSLLAI-Net5000 (b) SSLLAI-Net 50 (c) LAI差值对比(d) SSLLAI-Net5000局部 (e) SSLLAI-Net 50局部 (f) SSLLAI-Net5000局部 (g) SSLLAI-Net 50局部
    基于SSLLAI-Net LAI的研究区LAI反演结果(带噪声)(a) SSLLAI-Net5000 (b) SSLLAI-Net 50 (c) LAI差值对比(d) SSLLAI-Net5000局部 (e) SSLLAI-Net 50局部 (f) SSLLAI-Net5000局部 (g) SSLLAI-Net 50局部
    Fig. 7. 基于SSLLAI-Net LAI的研究区LAI反演结果(带噪声)(a) SSLLAI-Net5000 (b) SSLLAI-Net 50 (c) LAI差值对比(d) SSLLAI-Net5000局部 (e) SSLLAI-Net 50局部 (f) SSLLAI-Net5000局部 (g) SSLLAI-Net 50局部
    模型

    谷物

    总量108187

    玉米

    总量25334

    油菜

    总量28870

    其他作物

    总量27071

    R2RMSER2RMSER2RMSER2RMSE
    SSLLAI-Net50000.990.060.990.030.990.090.990.06
    SSLLAI-Net10000.990.110.990.050.990.100.980.10
    SSLLAI-Net5000.990.150.990.060.990.130.970.14
    SSLLAI-Net3000.990.150.990.070.980.160.970.15
    SSLLAI-Net1000.980.230.990.090.980.200.950.18
    SSLLAI-Net500.950.340.990.120.980.210.900.28
    Table 1. 基于SSLLAI-Net的植被叶面积指数定量反演精度
    模型谷物玉米油菜其他作物
    R2RMSER2RMSER2RMSER2RMSE
    Model150000.631.000.990.110.540.980.770.42
    Model110000.621.010.990.110.530.980.760.43
    Model15000.621.010.990.110.530.980.760.43
    Model13000.631.000.990.110.530.990.760.43
    Model11000.621.010.990.120.541.000.760.44
    Model1500.611.020.990.120.531.000.760.44
    Model250000.621.010.890.420.481.040.720.46
    Model210000.621.010.890.420.481.040.710.47
    Model25000.621.020.890.420.481.040.710.47
    Model23000.621.020.890.420.481.050.710.47
    Model21000.611.050.890.420.481.080.710.48
    Model2500.611.030.890.430.471.080.710.48
    SSLLAI-Net50000.990.060.990.030.990.090.990.06
    SSLLAI-Net10000.990.110.990.050.990.100.980.10
    SSLLAI-Net5000.990.150.990.060.990.130.970.14
    SSLLAI-Net3000.990.150.990.070.980.160.970.15
    SSLLAI-Net1000.980.230.990.090.980.200.950.18
    SSLLAI-Net500.950.340.990.120.980.210.900.28
    Table 2. 基于SSLLAI-Net模型和统计类模型的LAI定量反演精度对比
    模型

    谷物

    总量108187

    玉米

    总量25334

    油菜

    总量28870

    其他作物

    总量27071

    R2RMSER2RMSER2RMSER2RMSE
    SSLLAI-Net50000.990.110.990.050.990.170.990.07
    SSLLAI-Net10000.990.120.990.050.990.170.980.09
    SSLLAI-Net5000.990.160.990.090.980.170.970.15
    SSLLAI-Net3000.980.170.990.090.980.180.970.15
    SSLLAI-Net1000.970.250.990.130.970.220.930.23
    SSLLAI-Net500.950.330.980.180.960.270.890.29
    Table 3. 基于SSLLAI-Net的植被叶面积指数定量反演精度(带噪声)
    Xue-Ling LI, Ying-Ying DONG, Yi-Ning ZHU, Wen-Jiang HUANG. Leaf area index estimation with EnMAP hyperspectral data based on deep neural network[J]. Journal of Infrared and Millimeter Waves, 2020, 39(1): 111
    Download Citation