Scholarlink

<p>We present a monthly data-driven explanatory modelling framework for regulated basins, applied to the Jucar Hydrographic Confederation (JHC), a Mediterranean system strongly shaped by human interventions. The framework is built from over 20,000 monthly records (1988–2023), integrating meteorological variables, sectoral water demands, and hydrological observations from 94 control points within a consistent 40 × 40&#xa0;km spatial grid. A key methodological element is the inclusion of a first-order autoregressive term (<i>Value_t–1</i>) to represent hydrological memory. An autoregressive exogenous formulation (ARX) is adopted as the primary explanatory model, and Random Forest is reported as a benchmark. Model performance is evaluated using blocked time cross-validation with R², RMSE, NSE and KGE. The framework achieves high reconstruction accuracy in storage-dominated resources, reflecting strong temporal persistence in these systems, and moderate skill in rivers while avoiding the calibration of physical parameters or reliance on external climate projections. The results show that a single integrated model can reproduce basin dynamics with accuracy comparable to subsystem-specific models. Peak-flow dynamics remain more challenging at monthly resolution. Overall, the framework provides a lightweight and operational decision-support tool. Its direct reliance on observed data ensures traceability and transparency, making it particularly relevant for water authorities and decision-makers engaged in drought risk management, adaptive planning, and sustainable water governance under conditions of climatic variability.</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

A data-driven framework for monthly hydrological modelling in regulated basins: application to the Jucar Hydrographic Confederation

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月18日

查看全文
  • A. Garcia-Monteagudo,
  • M. Arnaldos,
  • M. A. Pardo

摘要

We present a monthly data-driven explanatory modelling framework for regulated basins, applied to the Jucar Hydrographic Confederation (JHC), a Mediterranean system strongly shaped by human interventions. The framework is built from over 20,000 monthly records (1988–2023), integrating meteorological variables, sectoral water demands, and hydrological observations from 94 control points within a consistent 40 × 40 km spatial grid. A key methodological element is the inclusion of a first-order autoregressive term (Value_t–1) to represent hydrological memory. An autoregressive exogenous formulation (ARX) is adopted as the primary explanatory model, and Random Forest is reported as a benchmark. Model performance is evaluated using blocked time cross-validation with R², RMSE, NSE and KGE. The framework achieves high reconstruction accuracy in storage-dominated resources, reflecting strong temporal persistence in these systems, and moderate skill in rivers while avoiding the calibration of physical parameters or reliance on external climate projections. The results show that a single integrated model can reproduce basin dynamics with accuracy comparable to subsystem-specific models. Peak-flow dynamics remain more challenging at monthly resolution. Overall, the framework provides a lightweight and operational decision-support tool. Its direct reliance on observed data ensures traceability and transparency, making it particularly relevant for water authorities and decision-makers engaged in drought risk management, adaptive planning, and sustainable water governance under conditions of climatic variability.