نوع مقاله : مقاله پژوهشی
نویسندگان
1 گروه مهندسی آب، دانشگاه بین المللی امام خمینی (ره)، قزوین، ایران
2 استادیار گروه مهندسی آب/ دانشگاه بین المللی امام خمینی قزوین
3 پژوهشکده مطالعات و تحقیقات منابع آب، موسسه تحقیقات آب، وزارت نیرو، تهران، ایران
چکیده
کلیدواژه‌ها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
In recent years, numerous numerical models have been developed to simulate atmospheric variables such as precipitation. This study aims to assess the efficacy of the Weather Research and Forecasting (WRF) model and the Integrated Forecast System (IFS) numerical system in simulating precipitation within the Poldokhtar Basin. The findings revealed that the WRF model exhibited a stronger correlation with observed precipitation values in the 6-hour time step (The average CC of WRF for the events of 2016 and 2018 is equal to 0.49 and for the IFS system in 2016, 0.43, in 2018, 0.15), whereas the IFS system demonstrated a higher correlation with observational data over longer time steps (The average CC in the 24-hour time step in 2016 and 2018 for the WRF model is 0.72 and 0.60, respectively, and for the IFS system, it is 0.75 and 0.70, respectively). Based on the NRMSE error-index, the average NRMSE in time steps of 6, 12, and 24 hours for the WRF model is 0.98, 0.86, and 0.67 mm (2016), 0.97, 0.72, and 0.75 mm (2018), respectively and for IFS numerical system is 1.01, 0.80 and 0.66 mm (2016) and 1.20, 0.76 and 0.79 mm (2018) respectively. Additionally, in the 24-hour time step, the results from the IFS numerical system closely resembled those obtained from the WRF model. Thus, the model's daily predictions can be utilized with higher confidence levels. It is imperative to note that the implementation of bias correction techniques is essential for mitigating the output errors in numerical weather forecasting models.
کلیدواژه‌ها [English]
EXTENDED ABSTRACT
Numerical Weather Prediction (NWP) models play a crucial role in flood forecasting systems by providing dependable forecasts of precipitation. Numerous global, regional, and local models have been developed by various organizations and institutions to meet this need. These models primarily aim to predict climatic variables, particularly the precipitation. Based on physical processes, boundary conditions, initial conditions, and spatiotemporal resolutions, they typically fall into three categories: global, regional, and local models. Among these, the WRF numerical meteorological model is widely utilized in numerical weather forecasting and atmospheric simulations worldwide. However, its reliance on the Linux operating system and lack of a graphical user interface poses implementation challenges. However, recent advancements in research infrastructure have led to the development of systems that provide outputs from some numerical models. This study sought to compare the forecasting performance of the WRF local model and the IFS regional model in predicting rainfall at 6, 12, and 24-hour intervals within the Poldokhtar basin.
This study focuses on the Poldokhtar Basin, situated within the geographical coordinates of 47° 55 min to 48° 43 min east longitude and 33° 7 min to 33° 55 min north latitude. Precipitation data from seven synoptic stations within and around this basin were utilized to assess the performance of the WRF numerical meteorological model and IFS numerical system outputs. The WRF modeling system was employed for precipitation simulation, utilizing two nested domains with a scaling ratio of 1 - 3. The primary domain, covering a spatial resolution of 15 km, was supplemented by a subdomain of 5 km. The boundary and initial conditions for the WRF model were sourced from FNL data. Additionally, data from the IFS system, with a spatial resolution of approximately 8 km and a time step of 6 hours, were obtained from the https://rda.ucar.edu/datasets/ds113.1/. The non-parametric quantile mapping (QM) method was employed for bias-corrected precipitation in both models. Quantitative evaluation of predicted precipitation values compared to observed values was conducted using indices such as the correlation coefficient (CC), mean absolute error (MAE), normalized RMSE (NRMSE), false alarm rate (FAR), and probability of detection (POD).
These findings indicate that the IFS database exhibits a stronger correlation with the observed values over longer time steps. Specifically, in the 6-hour time step, for IFS and WRF the average correlation coefficients for the 2016 event were 0.43 and 0.49, respectively, and 0.15 and 0.49 for the 2018 event. For the 12-hour time step, the correlation levels for the same events (2016 and 2018) were 0.62 and 0.65 for the WRF model, and 0.61 and 0.68 for the IFS database. Furthermore, both the WRF model and the IFS database demonstrated relatively higher errors in simulating 6-hour rainfall among the studied time steps. However, there was a noticeable trend of error reduction with increasing time steps, suggesting that the predictions from these models can be used with greater confidence in daily time steps. Evaluation of the efficiency of the WRF model and the IFS numerical database in distinguishing between rainy and non-rainy periods, based on POD and FAR tabular indices across different time steps, indicated acceptable performance in all three-time intervals. Analysis of spatial precipitation changes in the 2018 event at 6 and 12-hour time steps revealed that the simulations of the WRF numerical model closely resembled the observed values in terms of both the magnitude and spatial distribution of precipitation within the study area.
The comparison of performance between local and regional meteorological forecasting numerical models is of significant importance in the estimation of meteorological variables, particularly precipitation. As the purpose of meteorological modeling is to predict precipitation at sub-daily time intervals, utilization of the WRF numerical model is recommended. The adaptability and open-source nature of this model enable it to be tailored to specific regions and variables, thereby enhancing its ability to accurately predict precipitation at smaller time intervals. Conversely, for simulating precipitation variations within a 24-hour time step, the IFS numerical system presents a viable solution for many research endeavors.
Conceptualization, Koohi S. and Azizian A.; methodology, Koohi S. and Azizian A. and Najafi MS.; software, Koohi S. and Najafi MS.; Koohi S. and Azizian A. and Najafi MS.; validation, Koohi S. and Azizian A. and Najafi MS.; writing—original draft preparation, Koohi S.; writing—review and editing, Koohi S. and Azizian A. All authors have read and agreed to the published version of the manuscript.
Data is available on reasonable request from the authors.
The authors would like to thank the reviewers and editor for their critical comments that helped to improve the paper.
The authors avoided data fabrication, falsification, plagiarism, and misconduct.
The author declares no conflict of interest.
)