Document Type : Full length article
Physical Geography Department, Faculty of Geography, University of Tehran
Physical geography department, Geography faculty, Tehran university, Tehran, Iran
Physical Geography department, Faculty of geography, University of Tehran
Due to man-made heat generation and the release of numerous gaseous pollutants and particulate matter, cities generally experience significant ambient heat loads and air pollution. Researchers are considering alternative ways for evaluating the influence of wind on climate quality and natural ventilation in metropolitan spaces. Wind simulation in numerical climate models like the Weather Research and Forecasting (WRF) is always fraught with unknowns. The impact of ECMWF (ERA5) and NCEP (FNL) boundary condition data, as well as seven distinct physical schemas, was investigated in this study. The goal of this study is to find out the following: Is it possible to use the WRF model to evaluate wind speed and direction simulations? The results revealed that the wind direction simulated by the WRF model differs significantly from observational data. However, when it comes to wind speed, this disparity is less variable. The efficiency of the physical schema and the selected planetary boundary layer in the outcomes are determined by how well the schemas perform in proportion to the components of changeable direction and velocity. As a result, they performed best in the Exp speed (1& 2& 6), as well as the Exp direction (7& 3). The original data of boundary conditions was also discovered to have a substantial impact on the result simulation. The findings show that ERA5 boundary conditions perform better for modeling the wind direction component of Tehran, whereas FNL performs better for speed.
Keywords: Initial conditions, Wind direction and speed, Simulation, WRF, Tehran.
The wind has always been considered as an energy source from two perspectives: pattern and behavior in urban contexts, and potential in suburban environments. There are usually two major strategies for this purpose: one based on observational data and the other providing simulation data with the creation of climate models at various numerical scales (Han et al., 2014: 17). Numerical models are used in most studies to evaluate regional winds nowadays (Haman et al., 2010: 954; Shimada et al., 2011: 21). Simulated weather research and forecasting (WRF) has been used to conduct studies on this topic (Liu et al., 582: 2018; Salvaso et al., 276: 2018; Matar et al., 22: 2016; Charabi et al., 1: 2019; Tokhtenhagen et al., 119: 2020). The sensitivity and performance of the WRF model to initial and boundary conditions, as well as its impact on wind simulation, are investigated in this study. A planetary boundary layer scheme is also chosen to simulate the wind field in the city of Tehran.
Data and methods
The Meteorological Organization provided observational data on wind direction and speed for Mehrabad, Chitgar, Geophysical, and North Tehran (Shemiran) synoptic stations from 2018 on a three-hour time scale (Table 1). Data analysis time series from two databases, the National Environmental Forecasting Center (NCEP-FNL) and the European Center for Medium-Term Weather Forecasting (ECMWF-) ERA5), were used as the initial and boundary conditions to achieve the frequency and distribution of wind direction and velocity for January, May, July, and October. The WRF model, version 4.1.2, was used to simulate the components of wind speed and direction using boundary condition data in this investigation. The RRTM longwave radiation model, the Goddard shortwave radiation design, the Noah surface model, the WSM6 microphysical schema, the two-dimensional Cumulus Betts-Miller-Janjic schema, and the three-dimensional Grell – Freitas schema were all employed in the study. The MRF Medium-Range Prediction Model, the Younesi University YSU Scheme, the MYJ Scheme, the second ACM2 Asymmetric Convection Scheme, the QNSE Normal Gaussian Scale, and the second and third MYNN Turbulence Scale are all used to test the performance sensitivity of the planetary boundary layer schemas.
Result and discussion
By examining the characteristics of observation stations according to Table 1, all selected stations have an average height difference of at least 110 meters and the difference between the lowest (Mehrabad) and highest (Shemiran) stations is 360 meters and naturally, this characteristic can be in the amount of The accuracy of the simulations is influenced by the weather forecasting research model. Among these 4 selected stations, the geophysical station has the weakest outputs due to its complex spatial topography and more surface friction.
By examining the skewness statistics in the wind speed component in both types of starting and border conditions, it appears that there is a growing trend with increasing skew height in January and October, but no specific pattern in May and July, although there is a trend in the speed component. With rising station height and skew, the trend is almost climbing in the majority of situations. The stations demonstrate a reduction in the degree of correlation with increasing altitude during January when looking at the degree of correlation of the wind direction component. Given the location of the stations in Tehran, it is feasible to appreciate the research area's complexity in comparison to smooth and uniform surfaces, which has a substantial impact on the simulations' accuracy, particularly the components of direction and wind speed that are dependent. It will feature a steep slope and a strong topography. The basis for error and increased uncertainty is provided in the simulations due to the model's demand for high-resolution geographic data (topography, vegetation, land use, soil type, soil moisture, albedo, etc.) and the lack of this data in the research model of weather forecasting.
The results of simulating the wind speed component with wind direction were significantly more satisfactory. Looking at the average output of the data in Table 9, the FNL model boundary conditions in January and May performed better for the wind speed component. The ERA5 demonstrated increased capability in July. Both models had very identical circumstances in October. However, the FNL boundary conditions only performed well in January for the wind direction component; and in general, the ERA5 model has a numerical advantage in simulation. The best schemes in different months, according to Table 8, for the wind direction component, were MRF, QNSE, and MYNN, respectively. However, for the wind speed component, the YSU, MYJ, and ACM2 schemas performed best, as evidenced by Santos et al. (2013) and Gholami et al. (2013) research (1397).