Research on Climate Drivers of Ecosystem Services’ Value Loss Offset in the Qinghai–Tibet Plateau Based on Explainable Deep Learning
As one of the highest and most ecologically vulnerable regions in the world, the Qinghai–Tibet Plateau (QTP) presents significant challenges for the application of existing ecosystem service value (ESV) assessment models due to its extreme climate changes and unique plateau environment. Current mode...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-12-01
|
| Series: | Land |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2073-445X/13/12/2141 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1846104039108902912 |
|---|---|
| author | Wenshu Liu Chang You Jingbiao Yang |
| author_facet | Wenshu Liu Chang You Jingbiao Yang |
| author_sort | Wenshu Liu |
| collection | DOAJ |
| description | As one of the highest and most ecologically vulnerable regions in the world, the Qinghai–Tibet Plateau (QTP) presents significant challenges for the application of existing ecosystem service value (ESV) assessment models due to its extreme climate changes and unique plateau environment. Current models often fail to adequately account for the complex climate variability and topographical features of the QTP, making accurate assessments of ESV loss deviations difficult. To address these challenges, this study focuses on the QTP and employs a modified ESV loss deviation model, integrated with explainable deep learning techniques (LSTM-SHAP), to quantify and analyze ESV loss deviations and their climate drivers from 1990 to 2030. The results show that (1) between 1990 and 2020, the offset index in the eastern QTP consistently remained low, indicating significant deviations. Since 2010, low-value clusters in the western region have significantly increased, reflecting a widening range of ecological damage caused by ESV losses, with no marked improvement from 2020 to 2030. (2) SHAP value analysis identified key climate drivers, including temperature seasonality, diurnal temperature variation, and precipitation patterns, which exhibit nonlinear impacts and threshold effects on ESV loss deviation. (3) In the analysis of nonlinear relationships among key climate drivers, the interaction between diurnal temperature range and precipitation in wet seasons demonstrated significant effects, indicating that the synergistic action of temperature variation and precipitation patterns is critical to ecosystem stability. Furthermore, the complex nonlinear interactions between climate factors exacerbated the volatility of ESV loss deviations, particularly under extreme climate conditions. The 2030 forecast highlights that wet season precipitation and annual rainfall will become key factors driving changes in ESV loss deviation. By combining explainable deep learning methods, this study advances the understanding of the relationship between climate drivers and ecosystem service losses, providing scientific insights for ecosystem protection and sustainable management in the Qinghai–Tibet Plateau. |
| format | Article |
| id | doaj-art-da59dc1b826c4b36b911c29f53bfbc51 |
| institution | Kabale University |
| issn | 2073-445X |
| language | English |
| publishDate | 2024-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Land |
| spelling | doaj-art-da59dc1b826c4b36b911c29f53bfbc512024-12-27T14:35:15ZengMDPI AGLand2073-445X2024-12-011312214110.3390/land13122141Research on Climate Drivers of Ecosystem Services’ Value Loss Offset in the Qinghai–Tibet Plateau Based on Explainable Deep LearningWenshu Liu0Chang You1Jingbiao Yang2School of Ethnology and Sociology, Minzu University of China, Beijing 100081, ChinaSchool of Ethnology and Sociology, Minzu University of China, Beijing 100081, ChinaCollege of Life and Environmental Sciences, Minzu University of China, Beijing 100081, ChinaAs one of the highest and most ecologically vulnerable regions in the world, the Qinghai–Tibet Plateau (QTP) presents significant challenges for the application of existing ecosystem service value (ESV) assessment models due to its extreme climate changes and unique plateau environment. Current models often fail to adequately account for the complex climate variability and topographical features of the QTP, making accurate assessments of ESV loss deviations difficult. To address these challenges, this study focuses on the QTP and employs a modified ESV loss deviation model, integrated with explainable deep learning techniques (LSTM-SHAP), to quantify and analyze ESV loss deviations and their climate drivers from 1990 to 2030. The results show that (1) between 1990 and 2020, the offset index in the eastern QTP consistently remained low, indicating significant deviations. Since 2010, low-value clusters in the western region have significantly increased, reflecting a widening range of ecological damage caused by ESV losses, with no marked improvement from 2020 to 2030. (2) SHAP value analysis identified key climate drivers, including temperature seasonality, diurnal temperature variation, and precipitation patterns, which exhibit nonlinear impacts and threshold effects on ESV loss deviation. (3) In the analysis of nonlinear relationships among key climate drivers, the interaction between diurnal temperature range and precipitation in wet seasons demonstrated significant effects, indicating that the synergistic action of temperature variation and precipitation patterns is critical to ecosystem stability. Furthermore, the complex nonlinear interactions between climate factors exacerbated the volatility of ESV loss deviations, particularly under extreme climate conditions. The 2030 forecast highlights that wet season precipitation and annual rainfall will become key factors driving changes in ESV loss deviation. By combining explainable deep learning methods, this study advances the understanding of the relationship between climate drivers and ecosystem service losses, providing scientific insights for ecosystem protection and sustainable management in the Qinghai–Tibet Plateau.https://www.mdpi.com/2073-445X/13/12/2141deep learningQinghai–Tibet Plateauecosystem service value lossCLIMATE variables |
| spellingShingle | Wenshu Liu Chang You Jingbiao Yang Research on Climate Drivers of Ecosystem Services’ Value Loss Offset in the Qinghai–Tibet Plateau Based on Explainable Deep Learning Land deep learning Qinghai–Tibet Plateau ecosystem service value loss CLIMATE variables |
| title | Research on Climate Drivers of Ecosystem Services’ Value Loss Offset in the Qinghai–Tibet Plateau Based on Explainable Deep Learning |
| title_full | Research on Climate Drivers of Ecosystem Services’ Value Loss Offset in the Qinghai–Tibet Plateau Based on Explainable Deep Learning |
| title_fullStr | Research on Climate Drivers of Ecosystem Services’ Value Loss Offset in the Qinghai–Tibet Plateau Based on Explainable Deep Learning |
| title_full_unstemmed | Research on Climate Drivers of Ecosystem Services’ Value Loss Offset in the Qinghai–Tibet Plateau Based on Explainable Deep Learning |
| title_short | Research on Climate Drivers of Ecosystem Services’ Value Loss Offset in the Qinghai–Tibet Plateau Based on Explainable Deep Learning |
| title_sort | research on climate drivers of ecosystem services value loss offset in the qinghai tibet plateau based on explainable deep learning |
| topic | deep learning Qinghai–Tibet Plateau ecosystem service value loss CLIMATE variables |
| url | https://www.mdpi.com/2073-445X/13/12/2141 |
| work_keys_str_mv | AT wenshuliu researchonclimatedriversofecosystemservicesvaluelossoffsetintheqinghaitibetplateaubasedonexplainabledeeplearning AT changyou researchonclimatedriversofecosystemservicesvaluelossoffsetintheqinghaitibetplateaubasedonexplainabledeeplearning AT jingbiaoyang researchonclimatedriversofecosystemservicesvaluelossoffsetintheqinghaitibetplateaubasedonexplainabledeeplearning |