{"title":"基于可解释人工智能和 LSTM 的土壤温度预测","authors":"Qingtian Geng, Leilei Wang, Qingliang Li","doi":"10.3389/fenvs.2024.1426942","DOIUrl":null,"url":null,"abstract":"Soil temperature is a key parameter in many disciplines, and its research has important practical significance. In recent years, the prediction of soil temperature by deep learning has achieved good results. However, deep learning is difficult to popularize in practical use because of its opacity. This study aims to interpret and analyze the Long Short Term Memory Network (LSTM) model for global soil temperature prediction using SHapley Additive exPlanation (SHAP), Permutation Importance (PI) and Partial Dependence Plot (PDP). The results show that Temperature of air at 2 m above the surface of land has the greatest influence on the prediction of soil temperature, and its SHAP and PI characteristic values have significant seasonality. Meanwhile, radiation also has a certain influence on the prediction results. There was a significant positive correlation between the temperature of 2 m and the soil temperature. The explanatory insights provided in this paper enhance the transparency and confidence of the model, which promotes the applicability of soil temperature prediction models in relevant fields.","PeriodicalId":12460,"journal":{"name":"Frontiers in Environmental Science","volume":"66 1","pages":""},"PeriodicalIF":3.3000,"publicationDate":"2024-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Soil temperature prediction based on explainable artificial intelligence and LSTM\",\"authors\":\"Qingtian Geng, Leilei Wang, Qingliang Li\",\"doi\":\"10.3389/fenvs.2024.1426942\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Soil temperature is a key parameter in many disciplines, and its research has important practical significance. In recent years, the prediction of soil temperature by deep learning has achieved good results. However, deep learning is difficult to popularize in practical use because of its opacity. This study aims to interpret and analyze the Long Short Term Memory Network (LSTM) model for global soil temperature prediction using SHapley Additive exPlanation (SHAP), Permutation Importance (PI) and Partial Dependence Plot (PDP). The results show that Temperature of air at 2 m above the surface of land has the greatest influence on the prediction of soil temperature, and its SHAP and PI characteristic values have significant seasonality. Meanwhile, radiation also has a certain influence on the prediction results. There was a significant positive correlation between the temperature of 2 m and the soil temperature. The explanatory insights provided in this paper enhance the transparency and confidence of the model, which promotes the applicability of soil temperature prediction models in relevant fields.\",\"PeriodicalId\":12460,\"journal\":{\"name\":\"Frontiers in Environmental Science\",\"volume\":\"66 1\",\"pages\":\"\"},\"PeriodicalIF\":3.3000,\"publicationDate\":\"2024-07-31\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Frontiers in Environmental Science\",\"FirstCategoryId\":\"93\",\"ListUrlMain\":\"https://doi.org/10.3389/fenvs.2024.1426942\",\"RegionNum\":3,\"RegionCategory\":\"环境科学与生态学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"ENVIRONMENTAL SCIENCES\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Frontiers in Environmental Science","FirstCategoryId":"93","ListUrlMain":"https://doi.org/10.3389/fenvs.2024.1426942","RegionNum":3,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENVIRONMENTAL SCIENCES","Score":null,"Total":0}
引用次数: 0
摘要
土壤温度是许多学科的关键参数,其研究具有重要的现实意义。近年来,利用深度学习预测土壤温度取得了不错的效果。然而,深度学习因其不透明性,在实际应用中难以普及。本研究旨在利用SHAPLE Additive exPlanation(SHAP)、Permutation Importance(PI)和Partial Dependence Plot(PDP)对全球土壤温度预测的长短期记忆网络(LSTM)模型进行解释和分析。结果表明,地表以上 2 米处的空气温度对土壤温度的预测影响最大,其 SHAP 和 PI 特性值具有明显的季节性。同时,辐射对预测结果也有一定影响。2 m 温度与土壤温度之间存在明显的正相关。本文提供的解释性见解增强了模型的透明度和可信度,促进了土壤温度预测模型在相关领域的应用。
Soil temperature prediction based on explainable artificial intelligence and LSTM
Soil temperature is a key parameter in many disciplines, and its research has important practical significance. In recent years, the prediction of soil temperature by deep learning has achieved good results. However, deep learning is difficult to popularize in practical use because of its opacity. This study aims to interpret and analyze the Long Short Term Memory Network (LSTM) model for global soil temperature prediction using SHapley Additive exPlanation (SHAP), Permutation Importance (PI) and Partial Dependence Plot (PDP). The results show that Temperature of air at 2 m above the surface of land has the greatest influence on the prediction of soil temperature, and its SHAP and PI characteristic values have significant seasonality. Meanwhile, radiation also has a certain influence on the prediction results. There was a significant positive correlation between the temperature of 2 m and the soil temperature. The explanatory insights provided in this paper enhance the transparency and confidence of the model, which promotes the applicability of soil temperature prediction models in relevant fields.
期刊介绍:
Our natural world is experiencing a state of rapid change unprecedented in the presence of humans. The changes affect virtually all physical, chemical and biological systems on Earth. The interaction of these systems leads to tipping points, feedbacks and amplification of effects. In virtually all cases, the causes of environmental change can be traced to human activity through either direct interventions as a consequence of pollution, or through global warming from greenhouse case emissions. Well-formulated and internationally-relevant policies to mitigate the change, or adapt to the consequences, that will ensure our ability to thrive in the coming decades are badly needed. Without proper understanding of the processes involved, and deep understanding of the likely impacts of bad decisions or inaction, the security of food, water and energy is a risk. Left unchecked shortages of these basic commodities will lead to migration, global geopolitical tension and conflict. This represents the major challenge of our time. We are the first generation to appreciate the problem and we will be judged in future by our ability to determine and take the action necessary. Appropriate knowledge of the condition of our natural world, appreciation of the changes occurring, and predictions of how the future will develop are requisite to the definition and implementation of solutions.
Frontiers in Environmental Science publishes research at the cutting edge of knowledge of our natural world and its various intersections with society. It bridges between the identification and measurement of change, comprehension of the processes responsible, and the measures needed to reduce their impact. Its aim is to assist the formulation of policies, by offering sound scientific evidence on environmental science, that will lead to a more inhabitable and sustainable world for the generations to come.