Journal of Geo-information Science, Volume. 22, Issue 9, 1799(2020)

Estimating Soil Organic Matter based on Machine Learning Under Sparse Sample

Mingjie LIU1,2, Zhuokui XU1,3, Yunbing GAO2,4、*, Jing YANG2,4, Yuchun PAN2,4, Bingbo GAO5, Yanbing ZHOU2,4, Wanpeng ZHOU2,6, and Ling WANG7
Author Affiliations
  • 1School of Traffic and Transportation Engineering, Changsha University of Science and Technology, Changsha 410114, China
  • 2Beijing Research Center for Information Technology in Agriculture, Beijing 100097, China
  • 3Engineering Laboratory of Spatial Information Technology of Highway Geological Disaster Early Warning in Hunan Province (Changsha University of Science & Technology),Changsha 410114, China
  • 4National Engineering Research Center for Information Technology in Agriculture, Beijing 100097, China
  • 5China Agricultural University, Beijing 100083, China
  • 6Henan Polytechnic University, Jiaozuo 454003, China
  • 7Institute of Agricultural Resources and Environment, Hebei Academy of Agriculture and Forestry Sciences, Shijiazhuang 050051, China
  • show less
    Figures & Tables(14)
    Research Roadmap
    The structure chart of General Regression Neural Network
    The structurechart of Random Forest
    The overview of Daxing district of Bejing
    Variograms founction and related parameters of all experimental groups
    Comparison of the results predicted by the three methods in D2703, D676, D169 and D43
    The change of RMSE of GRNN, RF and Ordinary Kriging when the number of sampling points decreases
    • Table 1. Variance analysis of soil organic matter

      View table
      View in Article

      Table 1. Variance analysis of soil organic matter

      方差来源偏差平方和自由度Df均方FP
      用地类型组间351.750487.9386.8710.000
      组内1023.8688012.796
      总体1375.61884
      土壤质地组间356.4053118.8029.4420.000
      组内1019.2138112.583
      总体1375.61884
      畜禽粪便利用强度组间241.351548.2703.3620.008
      组内1134.2677914.358
      总体1375.61884
      土壤类型组间0.94520.4720.0280.972
      组内1374.6748216.764
      总体1375.61884
      植被指数组间17.59228.7960.5310.590
      组内1358.0278216.561
      总体1375.61884
    • Table 2. Descriptive statistics of soil organic matter in all experimental groups in the study area

      View table
      View in Article

      Table 2. Descriptive statistics of soil organic matter in all experimental groups in the study area

      实验组极大值/(g/kg)极小值/(g/kg)平均值/(g/kg)标准差/(g/kg)变异系数偏度峰度K-S双侧显著性
      D270324.731.2010.493.9137.280.147-0.0980.302
      D135224.731.6510.533.9237.230.159-0.0030.632
      D67624.731.6510.584.1138.870.3180.0360.640
      D33924.731.8110.553.9437.320.2230.2920.946
      D16922.981.9810.203.9939.080.2900.1120.964
      D8518.052.0310.934.0537.01-0.181-0.8040.934
      D43_117.974.3511.073.4631.24-0.187-0.6160.972
      D43_217.732.9810.074.1741.390.122-0.8930.951
      D43_317.843.5311.103.5331.80-0.091-0.9000.855
      D43_419.672.0210.334.0439.130.128-0.0640.906
      D43_517.682.1910.394.4642.90-0.329-0.9980.445
      D22_118.283.5310.704.3040.130.293-0.9880.611
      D22_217.613.9111.283.8434.03-0.037-0.6340.940
      D22_316.403.7210.493.6034.29-0.380-0.6920.783
      D22_415.344.2910.563.2430.72-0.320-0.8250.912
      D22_517.862.369.354.3646.600.260-0.5400.999
    • Table 3. Soil organic matter content grading standard

      View table
      View in Article

      Table 3. Soil organic matter content grading standard

      土壤属性丰富较丰富中等较缺乏缺乏极缺乏
      有机质/(g/kg)>4030~4020~3010~206~10< 6
    • Table 4. Prediction accuracy of GRNN, RF and Ordinary Krigingin all experimental groups

      View table
      View in Article

      Table 4. Prediction accuracy of GRNN, RF and Ordinary Krigingin all experimental groups

      实验组RMSEMRE/%MAE
      KrigeGRNNRFKrigeGRNNRFKrigeGRNNRF
      D27032.823.112.9926.8929.5928.922.212.432.35
      D13523.023.173.0129.5430.2829.503.022.462.36
      D6763.453.343.2333.0031.9430.502.742.652.57
      D3393.383.173.1735.6030.6431.192.822.472.53
      D1693.553.413.3135.9236.6434.392.832.752.61
      D854.103.172.9643.7633.3730.453.472.722.46
      D43_13.362.842.7230.4025.6323.732.742.312.18
      D43_24.133.193.1144.5834.5233.703.342.702.65
      D43_33.683.143.3135.5330.2631.933.172.672.91
      D43_43.952.763.1844.8633.5937.363.212.362.69
      D43_54.693.423.6167.4438.4647.364.062.703.01
      D22_14.343.223.4243.6035.0633.763.772.862.91
      D22_24.312.432.6265.6424.6026.553.832.132.28
      D22_33.902.893.0777.1127.6333.043.542.272.64
      D22_43.532.562.8658.3823.2327.423.032.182.45
      D22_54.312.982.98146.3535.8341.133.792.442.55
    • Table 5. Spatial correlation analysis of all experimental groupsamples

      View table
      View in Article

      Table 5. Spatial correlation analysis of all experimental groupsamples

      实验组平均最短距离/mMoran's IZ得分P实验组平均最短距离/mMoran's IZ得分P
      D2703371.450.3556.780.00D43_3*2615.740.151.150.25
      D1352425.630.3330.830.00D43_4*3109.010.151.300.19
      D676567.680.3314.990.00D43_5*2909.110.080.830.41
      D339810.790.208.240.00D22_1*4237.90-0.32-1.390.17
      D1691285.910.175.680.00D22_2*4384.49-0.15-0.530.60
      D851914.450.193.060.00D22_3*3662.95-0.10-0.190.85
      D43_12792.950.222.410.02D22_4*4342.870.100.740.46
      D43_23110.800.141.520.13D22_5*4723.02-0.11-0.480.63
    • Table 6. Detected result of three influence factor of soil organic matter

      View table
      View in Article

      Table 6. Detected result of three influence factor of soil organic matter

      实验组土地利用类型土壤质地畜禽粪便影响强度
      qpqpqp
      D27030.1820.0000.1990.0000.1480.000
      D13520.2090.0000.1900.0000.1390.000
      D6760.1980.0000.2330.0000.1710.000
      D3390.2120.0000.1890.0000.1380.000
      D1690.2100.0000.2290.0000.2020.000
      D850.2560.0670.2590.0160.1750.035
      D430.5000.0040.1770.0710.3450.050
      D220.4960.0160.4770.0490.7420.034
    • Table 7. The correlation analysis between the observed values and predicted values of GRNN, RF and Ordinary Kriging

      View table
      View in Article

      Table 7. The correlation analysis between the observed values and predicted values of GRNN, RF and Ordinary Kriging

      实验组KrigeGRNNRF实验组KrigeGRNNRF
      D27030.686**0.608**0.642**D43_30.1550.388*0.392**
      D13520.635**0.590**0.637**D43_40.1530.669**0.576**
      D6760.543**0.581**0.616**D43_5-0.0670.591**0.577**
      D3390.450**0.580**0.578**D22_1-0.1370.572**0.590**
      D1690.433**0.500**0.545**D22_2-0.3430.709**0.630**
      D850.1750.599**0.660**D22_3-0.0940.584**0.440*
      D43_10.2100.532**0.594**D22_4-0.1020.547**0.398*
      D43_20.1290.559**0.627**D22_50.0540.635**0.762**
    Tools

    Get Citation

    Copy Citation Text

    Mingjie LIU, Zhuokui XU, Yunbing GAO, Jing YANG, Yuchun PAN, Bingbo GAO, Yanbing ZHOU, Wanpeng ZHOU, Ling WANG. Estimating Soil Organic Matter based on Machine Learning Under Sparse Sample[J]. Journal of Geo-information Science, 2020, 22(9): 1799

    Download Citation

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

    Received: Aug. 13, 2019

    Accepted: --

    Published Online: Apr. 23, 2021

    The Author Email: GAO Yunbing (gybgis@163.com)

    DOI:10.12082/dqxxkx.2020.190441

    Topics