Journal of Integrative Agriculture ›› 2022, Vol. 21 ›› Issue (8): 2422-2434.DOI: 10.1016/S2095-3119(21)63692-4

所属专题： 农业生态环境-遥感合辑Agro-ecosystem & Environment—Romote sensing

• • 上一篇    下一篇

利用机器学习和环境关联对大面积复杂地区土体深度进行空间预测

  


  • 收稿日期:2021-03-23 接受日期:2021-04-23 出版日期:2022-08-01 发布日期:2021-04-23

Predicting soil depth in a large and complex area using machine learning and environmental correlations

LIU Feng1, 2, YANG Fei1, ZHAO Yu-guo1, 2, ZHANG Gan-lin1, 2, 3, LI De-cheng1    

  1. 1 State Key Laboratory of Soil and Sustainable Agriculture, Institute of Soil Science, Chinese Academy of Sciences, Nanjing 210008, P.R.China

    2 University of the Chinese Academy of Sciences, Beijing 100049, P.R.China

    3 Key Laboratory of Watershed Geographic Science, Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, Nanjing 210008, P.R.China

  • Received:2021-03-23 Accepted:2021-04-23 Online:2022-08-01 Published:2021-04-23
  • About author:LIU Feng, E-mail: fliu@issas.ac.cn; Correspondence ZHANG Gan-lin, Tel: +86-25-86881279, E-mail: glzhang@issas.ac.cn
  • Supported by:
    This study was supported by the National Natural Science Foundation of China (41130530, 91325301 and 42071072).

PDF

摘要：

本研究构建了一个可直接估计空间不确定性的集合式机器学习模型，即分位数回归森林，定量土体深度与环境条件之间的关系。将该模型与丰富的环境协同变量结合，预测了位于我国西北地区、面积为14万km2的黑河流域的土体深度空间分布，估算了制图结果的空间不确定性。使用了275个土体深度观测样本和26个环境协同变量数据。结果显示，模型预测精度R2为0.587，RMSE为2.98 cm（平方根尺度），可解释近60%的土体深度变异。土体深度图清晰地展示了土体深度的区域分布模式和局部细节。谷底、平原等低平低洼景观部位土体深度较大，而山坡、山脊、台地等高陡景观部位土体深度较小；绿洲内土体深度明显大于绿洲之外的荒漠地区，冲积平原中部土体深度明显大于边缘地带，而湖泊平原中部土体深度明显小于边缘地带。高的预测不确定性主要出现在可达性差、缺少样本的区域。分析发现，土壤发生过程和地貌过程共同塑造了该流域土体深度的空间模式，但地貌过程起主导作用。这一点可能也适用于世界上其它寒旱地区类似的“高寒山地-平原绿洲-荒漠戈壁”流域。


Abstract:

Soil depth is critical for eco-hydrological modeling, carbon storage calculation and land evaluation.  However, its spatial variation is poorly understood and rarely mapped.  With a limited number of sparse samples, how to predict soil depth in a large area of complex landscapes is still an issue.  This study constructed an ensemble machine learning model, i.e., quantile regression forest, to quantify the relationship between soil depth and environmental conditions.  The model was then combined with a rich set of environmental covariates to predict spatial variation of soil depth and straightforwardly estimate the associated predictive uncertainty in the 140 000 km2 Heihe River basin of northwestern China.  A total of 275 soil depth observation points and 26 covariates were used.  The results showed a model predictive accuracy with coefficient of determination (R2) of 0.587 and root mean square error (RMSE) of 2.98 cm (square root scale), i.e., almost 60% of soil depth variation explained.  The resulting soil depth map clearly exhibited regional patterns as well as local details.  Relatively deep soils occurred in low lying landscape positions such as valley bottoms and plains while shallow soils occurred in high and steep landscape positions such as hillslopes, ridges and terraces.  The oases had much deeper soils than outside semi-desert areas, the middle of an alluvial plain had deeper soils than its margins, and the middle of a lacustrine plain had shallower soils than its margins.  Large predictive uncertainty mainly occurred in areas with a lack of soil survey points.  Both pedogenic and geomorphic processes contributed to the shaping of soil depth pattern of this basin but the latter was dominant.  This findings may be applicable to other similar basins in cold and arid regions around the world.

Key words: digital soil mapping , spatial variation , uncertainty ,  machine learning ,  soil-landscape model ,  soil depth