Comment on hess-2024-212 DOI Creative Commons

Опубликована: Авг. 21, 2024

Abstract. Hydrometeorological forecasting is crucial for managing water resources and mitigating the impacts of extreme hydrologic events. At sub-seasonal scales, readily available hydrometeorological forecast products often exhibit large uncertainties insufficient accuracies to support decision making. We propose a deep learning based modelling framework joint precipitation streamflow forecasts lead time up 30 days. This achieved by coupling (1) convolutional neural network (CNN) architecture with ResNet blocks statistically downscaling ECMWF raw (2) hybrid model integrating conceptual Xin’anjiang (XAJ) long-short term memory (LSTM) forecasting. The CNN incorporates specialized loss function that combines continuous form threat score mean absolute error. Applying modeling source region Yangtze River Basin, results indicate CNN-based exhibits ~13 % ~10 less RMSE than quantile mapping (QM) forecasts, respectively, averaged over 30-day time. Similarly, achieves ~2 ~5 lower QM events above 90th percentile historic daily precipitation. Using these as meteorological drivers XAJ-LSTM model, we found forecasted flood peaks driven have 18 %–32 relative errors 13 %–22 compared those forecasts. However, standalone XAJ shows marginal improvements, or in some cases, no improvement at all, same enhanced highlights importance understanding effectiveness part chain. Our study expected provide implications leveraging advanced AI techniques enhance accuracy operational efficiency effective management disaster preparedness.

Язык: Английский

Dust cycle, soiling effect and optimum cleaning schedule for PV modules in Iran: A long-term multi-criteria analysis DOI
Seyyed Shahabaddin Hosseini Dehshiri, Bahar Firoozabadi

Energy Conversion and Management, Год журнала: 2023, Номер 286, С. 117084 - 117084

Опубликована: Апрель 27, 2023

Язык: Английский

Процитировано

35
A novel four-stage integrated GIS based fuzzy SWARA approach for solar site suitability with hydrogen storage system DOI
Seyyed Shahabaddin Hosseini Dehshiri, Bahar Firoozabadi

Energy, Год журнала: 2023, Номер 278, С. 127927 - 127927

Опубликована: Май 23, 2023

Язык: Английский

Процитировано

33
A multidisciplinary approach to select wind turbines for power-hydrogen production: Energy, exergy, economic, environmental under uncertainty prediction by artificial intelligence DOI
Seyyed Shahabaddin Hosseini Dehshiri, Bahar Firoozabadi

Energy Conversion and Management, Год журнала: 2024, Номер 310, С. 118489 - 118489

Опубликована: Май 4, 2024

Язык: Английский

Процитировано

12
Comparison, evaluation and prioritization of solar photovoltaic tracking systems using multi criteria decision making methods DOI Open Access
Seyyed Shahabaddin Hosseini Dehshiri, Bahar Firoozabadi

Sustainable Energy Technologies and Assessments, Год журнала: 2022, Номер 55, С. 102989 - 102989

Опубликована: Дек. 27, 2022

Язык: Английский

Процитировано

32
Photovoltaic plant site selection considering dust soiling effects: A novel hybrid framework based on uncertainty and reliability with optimum cleaning schedule DOI
Seyyed Shahabaddin Hosseini Dehshiri, Bahar Firoozabadi

Applied Energy, Год журнала: 2025, Номер 382, С. 125252 - 125252

Опубликована: Янв. 5, 2025

Язык: Английский

Процитировано

1
An integrated feature selection and machine learning framework for PM10 concentration prediction DOI
Elham Kalantari, Hamid Gholami, Hossein Malakooti

и другие.

Atmospheric Pollution Research, Год журнала: 2025, Номер unknown, С. 102456 - 102456

Опубликована: Фев. 1, 2025

Язык: Английский

Процитировано

1
Aircraft observations of aerosol and BC during the East Asian dust storm event: Vertical profiles, size distribution and mixing state DOI
Xingguang Liu, Delong Zhao,

Zhong-qing Niu

и другие.

Atmospheric Environment, Год журнала: 2024, Номер 327, С. 120492 - 120492

Опубликована: Апрель 2, 2024

Язык: Английский

Процитировано

5
A compromise solution for comparison photovoltaic tracking systems: A 7E and uncertainty analysis assisted by machine learning algorithm DOI
Seyyed Shahabaddin Hosseini Dehshiri, Bahar Firoozabadi

Energy Conversion and Management, Год журнала: 2024, Номер 323, С. 119242 - 119242

Опубликована: Ноя. 19, 2024

Язык: Английский

Процитировано

5
Improving the Asian dust storm prediction using WRF-Chem through combinational optimization of physical parameterization schemes DOI Creative Commons
Ji Won Yoon,

Ebony Lee,

Seon Ki Park

и другие.

Atmospheric Environment, Год журнала: 2024, Номер 326, С. 120461 - 120461

Опубликована: Март 18, 2024

This study aims to enhance the accuracy of Weather Research and Forecasting model coupled with Chemistry (WRF-Chem) in forecasting Asian dust storms (ADSs) by using micro-Genetic Algorithm (μGA). We developed an optimization system---the WRF-Chem-μGA system---to seek optimal combination planetary boundary layer (PBL) land surface parameterization schemes, which are crucial for numerical forecast storms. The was conducted concerning meteorological air quality variables, i.e., aerosol optical depth, PBL height, 2 m temperature, relative humidity, 10 wind speed, simultaneously three ADS cases over domain, including South Korea. Among a total 32 available combinations physical scheme options (8 from 4 schemes), optimized set through system consists Asymmetrical Convective Model version (ACM2) Noah Multiple Parameterization (Noah-MP) scheme. showed improvement ratio up 22.5 % terms normalized RMSE all compared various non-optimized sets schemes two additional cases. proposed this can be used comprehensively forecasts problems East region, WRF-Chem model.

Язык: Английский

Процитировано

4
Deep-learning-based sub-seasonal precipitation and streamflow ensemble forecasting over the source region of the Yangtze River DOI Creative Commons
Ningpeng Dong,

Haoran Hao,

Mingxiang Yang

и другие.

Hydrology and earth system sciences, Год журнала: 2025, Номер 29(8), С. 2023 - 2042

Опубликована: Апрель 22, 2025

Abstract. Hydrometeorological forecasting is crucial for managing water resources and mitigating the impacts of hydrological extremes. At sub-seasonal scales, readily available hydrometeorological forecast products often exhibit large uncertainties insufficient accuracies to support decision-making. We propose a deep-learning-based modelling framework joint precipitation streamflow ensemble forecasts lead time up 30 d. This achieved by coupling (1) an enhanced convolutional neural network (CNN) models with ResNet blocks specialized loss function statistically downscaling European Centre Medium-Range Forecasts (ECMWF) (2) hybrid hydrologic model integrating conceptual Xin'anjiang (XAJ) long short-term memory (LSTM) (XAJ-LSTM). Applying source region Yangtze River Basin, results indicate that CNN-based exhibits ∼34 % ∼26 less root mean squared error (RMSE) than raw ECMWF quantile mapping (QM) forecasts, respectively, averaged over d time. Similarly, CNN achieves approximately 6 10 lower RMSE QM heavy events. Using these as meteorological forcing XAJ-LSTM model, we found forecasted flood peaks driven have 16 %–33 relative errors 20 %–31 compared those forecasts. However, standalone XAJ shows only marginal improvements same highlights importance understanding effectiveness part chain. Our study expected provide implications leveraging advanced AI techniques enhance accuracy operational efficiency effective management disaster preparedness.

Язык: Английский

Процитировано

0