Scholarlink

<p>Quantifying and understanding the water cycle and water storage components often preclude amalgamating all interconnected aspects, leading to misinterpretations. Here, GRACE-derived terrestrial water storage (TWS) and auxiliary data from multiple sources are leveraged, leading to an ensemble of 36 probabilistic water budgets. A comprehensive framework for quantifying water storage variations in the earthquake-prone region of Japan is provided. A series of procedures are performed, including seismic correction, data gap filling by artificial neural network (ANN) and long short-term memory (LSTM) models, water balance closure, temporal variability disaggregation, and future projections centered on policymaking aspects. The seismic correction accounts for an increase of 313&#xa0;mm of TWS during March 2011 and results in a 98% increase in TWSA from April 2002 to May 2021. Both ANN and LSTM performed reasonably well (<i>r</i> &gt; 0.90, NSE &gt; 0.85) for data gap filling. Two different closure techniques provide physically consistent water budget components. Results demonstrate (a) mixed biases in precipitation and evapotranspiration, (b) continuous wet bias (overestimation) in runoff and TWS change, and (c) reduced ensemble spread in corrected/enforced components. Further, temporal disaggregation shows minimal contribution (fraction of variance) from long-term trends but a dominant contribution from seasonal (highest of ~ 97% for evapotranspiration) and/or sub-seasonal (as high as ~ 62% for soil moisture storage) variability in all the water cycle and water budget components. The currently proposed framework for a secular and holistic variability of water resources can be applied to various earthquake-prone regions, and may support improved freshwater management decisions.</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

Earthquake correction enabled holistic assessment of water storage variations in Japan

  • 期刊论文

  • 发表日期：

    2026年5月7日

查看全文
  • Abhishek,
  • Jinghua Xiong,
  • Wenpeng Zhao,
  • Jin Li,
  • Tsuyoshi Kinouchi

摘要

Quantifying and understanding the water cycle and water storage components often preclude amalgamating all interconnected aspects, leading to misinterpretations. Here, GRACE-derived terrestrial water storage (TWS) and auxiliary data from multiple sources are leveraged, leading to an ensemble of 36 probabilistic water budgets. A comprehensive framework for quantifying water storage variations in the earthquake-prone region of Japan is provided. A series of procedures are performed, including seismic correction, data gap filling by artificial neural network (ANN) and long short-term memory (LSTM) models, water balance closure, temporal variability disaggregation, and future projections centered on policymaking aspects. The seismic correction accounts for an increase of 313 mm of TWS during March 2011 and results in a 98% increase in TWSA from April 2002 to May 2021. Both ANN and LSTM performed reasonably well (r > 0.90, NSE > 0.85) for data gap filling. Two different closure techniques provide physically consistent water budget components. Results demonstrate (a) mixed biases in precipitation and evapotranspiration, (b) continuous wet bias (overestimation) in runoff and TWS change, and (c) reduced ensemble spread in corrected/enforced components. Further, temporal disaggregation shows minimal contribution (fraction of variance) from long-term trends but a dominant contribution from seasonal (highest of ~ 97% for evapotranspiration) and/or sub-seasonal (as high as ~ 62% for soil moisture storage) variability in all the water cycle and water budget components. The currently proposed framework for a secular and holistic variability of water resources can be applied to various earthquake-prone regions, and may support improved freshwater management decisions.