Simulation of CO2 flux in floating-leaf vegetation zones of Lake Taihu based on machine learning models.

Λεπτομέρειες βιβλιογραφικής εγγραφής
Τίτλος:	Simulation of CO2 flux in floating-leaf vegetation zones of Lake Taihu based on machine learning models.
Μεταγλωττισμένος Τίτλος:	基于机器学习模型的太湖浮叶植物区CO2通量模拟.
Συγγραφείς:	Luo SJ; NUIST Center on Atmospheric Environment, Nanjing University of Information Science and Technology, Nanjing 210044, China.; School of Applied Meteorology, Nanjing University of Information Science and Technology, Nanjing 210044, China., Zhang M; NUIST Center on Atmospheric Environment, Nanjing University of Information Science and Technology, Nanjing 210044, China.; School of Applied Meteorology, Nanjing University of Information Science and Technology, Nanjing 210044, China., Jia L; Jiangsu Province Meteorological Observation Center, Nanjing 210044, China., Xiao W; Jiangsu Province Meteorological Observation Center, Nanjing 210044, China., Qiao H; NUIST Center on Atmospheric Environment, Nanjing University of Information Science and Technology, Nanjing 210044, China.; School of Applied Meteorology, Nanjing University of Information Science and Technology, Nanjing 210044, China., Zhang SB; NUIST Center on Atmospheric Environment, Nanjing University of Information Science and Technology, Nanjing 210044, China.; School of Applied Meteorology, Nanjing University of Information Science and Technology, Nanjing 210044, China., Shi J; NUIST Center on Atmospheric Environment, Nanjing University of Information Science and Technology, Nanjing 210044, China.; School of Applied Meteorology, Nanjing University of Information Science and Technology, Nanjing 210044, China., Ge P; NUIST Center on Atmospheric Environment, Nanjing University of Information Science and Technology, Nanjing 210044, China.; School of Applied Meteorology, Nanjing University of Information Science and Technology, Nanjing 210044, China., Yang FY; NUIST Center on Atmospheric Environment, Nanjing University of Information Science and Technology, Nanjing 210044, China.; School of Applied Meteorology, Nanjing University of Information Science and Technology, Nanjing 210044, China., He Y; Liaoning Ecological Meteorology and Satellite Remote Sensing Center, Shenyang 110166, China.
Πηγή:	Ying yong sheng tai xue bao = The journal of applied ecology [Ying Yong Sheng Tai Xue Bao] 2026 May; Vol. 37 (5), pp. 1651-1664.
Τύπος έκδοσης:	Journal Article
Γλώσσα:	English
Στοιχεία περιοδικού:	Publisher: Ying yong sheng tai xue bao bian ji wei yuan hui Country of Publication: China NLM ID: 9425159 Publication Model: Print Cited Medium: Print ISSN: 1001-9332 (Print) Linking ISSN: 10019332 NLM ISO Abbreviation: Ying Yong Sheng Tai Xue Bao Subsets: MEDLINE
Imprint Name(s):	Publication: Shenyang Shi : Ying yong sheng tai xue bao bian ji wei yuan hui Original Publication: Shenyang Shi : Ying yong sheng tai xue bao bian ji wei yuan hui, 1990-
Ιατρικοί όροι (MeSH):	Carbon Dioxide/analysis , Plant Leaves/metabolism , Lakes* , Machine Learning*, Environmental Monitoring/methods ; Random Forest ; China ; Long Short Term Memory ; Seasons ; Computer Simulation ; Support Vector Machine ; Neural Networks, Computer ; Models, Theoretical ; Predictive Learning Models ; Ecosystem
Περίληψη:	As an important component of inland waters, shallow lakes are hotspots for CO2 emissions. Due to the influence of eutrophication and aquatic macrophyte, CO2 fluxes at the water-air interface of shallow lakes exhibit complex variability, posing challenges for high-accuracy simulation. To compare the performance of different machine learning models in simulating CO2 fluxes in shallow lakes, we focused on a floating-leaved vegetation zone in eastern Lake Taihu. Based on CO2 flux observations from an eddy covariance system, combined with meteorological, water quality, and vegetation variables, we developed four machine learning models, random forest (RF), support vector machine (SVM), backpropagation neural network (BPNN), and long short-term memory network (LSTM). Then, we evaluated the performance under three modeling scenarios, including growing season, non-growing season, and whole-season. Among the three modeling scenarios, the whole-season modeling approach achieved the best overall performance, with test-set metrics consistently outperforming those of the seasonal models. The RF model exhibited the highest accuracy and robustness under all the three scenarios. In the whole-season mode-ling scenario, the RF model achieved a coefficient of determination (R²) of 0.72 and a root mean square error (RMSE) of 0.57 μmol·m^-2·s^-1. For the growing-season model, the RF performance yielded an R² of 0.64 and an RMSE of 0.88 μmol·m^-2·s^-1, while in the non-growing-season model, the R² and RMSE were 0.61 and 0.43 μmol·m^-2·s^-1, respectively. The SVM and BPNN models showed comparable but inferior performance, whereas the LSTM model performed relatively poorly. Furthermore, we used recursive feature elimination (RFE) to identify the optimal combination of driving factors for the RF model under the whole-season scenario. The selected feature set included: surface water temperature (Tw_20), sediment temperature (Ts), dissolved oxygen (DO), air tempera-ture (Ta), incoming shortwave radiation (Rs_in), wind speed (WS), total nitrogen (TN), water pH, friction velocity (u^*), and normalized difference vegetation index (NDVI). This feature set further improved simulation accuracy (R²=0.76, RMSE=0.55 μmol·m^-2·s^-1) and effectively reduced model complexity. The SHAP analysis showed the significant influences of water temperature, radiation, dissolved oxygen, and vegetation index on CO2 fluxes. The results would provide a useful methodological reference for CO2 flux modeling and carbon cycle studies in shallow lakes.
Contributed Indexing:	Keywords: CO2 flux; Lake Taihu; floating-leaf vegetation; machine learning; random forest (RF) Local Abstract: [Publisher, Chinese] 浅水湖泊作为内陆水体的重要组成部分,是全球CO₂排放研究的热点之一。受富营养化和水生植物生长影响,浅水湖泊水-气界面CO₂通量变化复杂,其准确模拟仍面临挑战。为比较不同机器学习模型对浅水湖泊CO₂通量的模拟能力,本研究以太湖东部浮叶植物区为研究区域,基于涡度相关观测获取的CO₂通量实测数据,结合气象、水质和植被因子,构建了随机森林(RF)、支持向量机(SVM)、反向传播神经网络(BPNN)和长短期记忆网络(LSTM)4种模型,并在生长季、非生长季和全年3种建模情境下比较其拟合与预测性能。结果表明:3种建模情境下,全年建模的整体效果最佳,其测试集性能普遍优于分季建模结果。RF模型在3种建模情境下均表现最优,其在全年建模情境中测试集决定系数(R²)达0.72,模拟的CO₂通量均方根误差(RMSE)为0.57 μmol·m^-2·s^-1;生长季建模R²=0.64,RMSE=0.88 μmol·m^-2·s^-1;非生长季建模R²=0.61,RMSE=0.43 μmol·m^-2·s^-1。SVM和BPNN模型次之,LSTM模型模拟效果欠佳。进一步通过递归特征剔除,确定了全年建模情境下RF模型的最优特征组合,即表层水温(T_{w_20})、底泥温度(T_s)、溶解氧(DO)、气温(T_a)、入射短波辐射(R_{s_in})、风速(WS)、总氮(TN)、pH、摩擦风速(u^*)和归一化植被指数(NDVI)的组合,该组合在提升模拟准确性(R²=0.76,RMSE=0.55 μmol·m^-2·s^-1)的同时有效降低了模型复杂程度。SHAP分析进一步揭示了水温、光照、溶解氧和植被指数对CO₂通量的显著影响,研究结果可为浅水湖泊CO₂通量建模及相关碳循环研究提供借鉴方法。.
Substance Nomenclature:	142M471B3J (Carbon Dioxide)
Entry Date(s):	Date Created: 20260625 Date Completed: 20260626 Latest Revision: 20260625
Update Code:	20260627
DOI:	10.13287/j.1001-9332.202605.035
PMID:	42350140
Βάση Δεδομένων:	MEDLINE

Περιγραφή
ISSN:	1001-9332
DOI:	10.13287/j.1001-9332.202605.035