Scholarlink

<p>Climate change presents a substantial risk to water quality in agricultural reservoirs. Nevertheless, the application of machine learning (ML) to project future changes is frequently constrained by the scarcity of long-term observational data. This study develops a robust framework to project future Total Organic Carbon (TOC) concentrations by combining a Wasserstein Generative Adversarial Network (WGAN) for data augmentation with high-performance ML models. We applied this framework to two contrasting South Korean reservoirs (one pristine, one polluted). Historical data (<i>n</i> = 60) for each were augmented using WGAN and evaluated using Kolmogorov-Smirnov (KS), Anderson-Darling (AD), and Energy Distance (ED) tests, showing overall distributional similarity, with marginally significant KS/AD differences for TOC in Sacheon Reservoir (<i>p</i> &lt; 0.05) but a non-significant ED. Nine ML models were then optimized and evaluated. The best-performing model, M5 Model Tree with Stochastic Gradient Boosting (M5-SGB), was then used to project TOC from 2025 to 2100 using a 12 Global Climate Model (GCM) ensemble from CMIP6 under SSP2-4.5 and SSP5-8.5 scenarios. Notably, the effectiveness of WGAN augmentation was model-dependent, significantly boosting the performance of certain algorithms while having a negligible effect on others. Future projections revealed a statistically significant increase in TOC for both reservoirs. However, the response was site-specific: the pristine yet temporally unstable reservoir demonstrated vulnerability even under the moderate-emissions scenario. In contrast, the polluted yet stable reservoir demonstrated a significant trend exclusively under the high-emissions scenario. This research presents a practical framework for predicting water-quality context under limited data availability. The findings demonstrate that a reservoir’s vulnerability to climate change is critically linked to its dynamic variability, not just to its static water-quality grade. This offers crucial insights for designing more effective, risk-based water management and monitoring strategies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Projecting Future TOC in Data-Scarce Agricultural Reservoirs: a WGAN-Enhanced Framework Revealing the Importance of Dynamic Variability

  • 期刊论文

  • 发表日期：

    2026年2月21日

查看全文
  • Christian Joseph Siose,
  • Jeongho Han,
  • Kyung-Sook Choi,
  • Byoung Han Choi,
  • Kyoung Jae Lim

摘要

Climate change presents a substantial risk to water quality in agricultural reservoirs. Nevertheless, the application of machine learning (ML) to project future changes is frequently constrained by the scarcity of long-term observational data. This study develops a robust framework to project future Total Organic Carbon (TOC) concentrations by combining a Wasserstein Generative Adversarial Network (WGAN) for data augmentation with high-performance ML models. We applied this framework to two contrasting South Korean reservoirs (one pristine, one polluted). Historical data (n = 60) for each were augmented using WGAN and evaluated using Kolmogorov-Smirnov (KS), Anderson-Darling (AD), and Energy Distance (ED) tests, showing overall distributional similarity, with marginally significant KS/AD differences for TOC in Sacheon Reservoir (p < 0.05) but a non-significant ED. Nine ML models were then optimized and evaluated. The best-performing model, M5 Model Tree with Stochastic Gradient Boosting (M5-SGB), was then used to project TOC from 2025 to 2100 using a 12 Global Climate Model (GCM) ensemble from CMIP6 under SSP2-4.5 and SSP5-8.5 scenarios. Notably, the effectiveness of WGAN augmentation was model-dependent, significantly boosting the performance of certain algorithms while having a negligible effect on others. Future projections revealed a statistically significant increase in TOC for both reservoirs. However, the response was site-specific: the pristine yet temporally unstable reservoir demonstrated vulnerability even under the moderate-emissions scenario. In contrast, the polluted yet stable reservoir demonstrated a significant trend exclusively under the high-emissions scenario. This research presents a practical framework for predicting water-quality context under limited data availability. The findings demonstrate that a reservoir’s vulnerability to climate change is critically linked to its dynamic variability, not just to its static water-quality grade. This offers crucial insights for designing more effective, risk-based water management and monitoring strategies.