This article is a re-edited version of an excerpt from "Verification of the accuracy of air density calculated by meso-meteorological models," presented at the 2021rd Wind Energy Utilization Symposium held in November 11.

table of contents
  1. 1. In the beginning
  2. XNUMX.Method
  3. 3. Results and Discussion
  4. XNUMX.Summary

1. In the beginning

In the wind power field, air density is a meteorological element necessary for power generation evaluation and wind turbine design. For example, in the "Wind Farm Certification: Onshore Wind Power Plants" by Nippon Kaiji Kyokai,1)In order to design a wind turbine, it is necessary to calculate the air density at the turbine location when it is generating power and when it is on standby for a storm.

To obtain air density easily and over a wide area, the meso-scale meteorological model WRF (Weather Research and Forecasting model) is used.2) Simulation results by WRF are useful (Figure 1). However, it is important to verify the accuracy of such simulation results by referring to observed values, and to understand and improve the errors. Therefore, in this study, we verified the accuracy of air density calculated by WRF using observed values ​​and considered the calculation errors.

Figure 1. Average annual distribution of air density in the New European Wind Atlas (NEWA)3)

XNUMX.Method

In this study, we used the offshore wind condition map NeoWins4)Among the WRF simulation results used in the development of the model, we used hourly values ​​from a 30m grid calculation targeting a distance of 500km offshore (Table 1 and Figure 1). To verify the WRF calculation results, we used air density calculated from the ground temperature and pressure at 2 meteorological stations in Japan (Figure 142) from the Japan Meteorological Agency's AMeDAS data as on-site observation values.

Table 1 WRF calculation conditions4)

Figure 2 WRF computational domain4)

Figure 3 Locations of 142 meteorological stations

3. Results and Discussion

Figure 4 shows box plots of bias (average error) and R2 (coefficient of determination) for WRF-calculated temperature, pressure, and air density at meteorological stations. The absolute value of the bias is small, the average coefficient of determination is 0.97 to 0.99, and the majority of the samples have small variations. These results show that WRF calculates temperature, pressure, and air density with high accuracy.

Figure 4 Box plots of (1) bias and (2) R2 for (a) temperature, (b) pressure, and (c) air density calculated by WRF at meteorological stations.
*The numbers at the bottom of the figure are the average of each statistic.

To examine the error trends of WRF air density in more detail, the biases in Figure 4 are divided by sea level pressure, and the results are shown in Figure 5. When the sea level pressure is greater than 980 [hPa], the biases are consistent with the positive and negative trends of all samples mentioned above (Figure 4). On the other hand, when the sea level pressure is less than 980 [hPa], the biases have smaller positive values ​​for temperature, larger positive values ​​for pressure, and slightly positive values ​​for air density compared to other samples. This result suggests that when weather conditions cause a decrease in pressure (when a low pressure system or typhoon passes), WRF tends to overestimate pressure and therefore slightly overestimate air density.

In order to improve the accuracy of WRF air density, it is necessary to improve the errors in the temperature and pressure that are the basis of the WRF air density. RTI plans to continue research into these meteorological elements, aiming to improve the accuracy of WRF simulations.

Figure 5 Box plots of biases for (a) temperature, (b) pressure, and (c) air density calculated by WRF for different sea level pressures at meteorological stations.

*Sea level pressure measurements taken by meteorological stations are divided into bins of 10 hPa to calculate each statistic, and statistics with fewer than 30 samples are excluded in advance. The figures at the bottom of the figure are the averages of each statistic.

XNUMX.Summary

In this study, we verified the accuracy of the air density calculations calculated by the mesoscale weather model WRF using the Japan Meteorological Agency's AMeDAS observation data. The main conclusions are summarized below.

  • Verification using AMeDAS observations revealed that WRF is able to reproduce air density with high accuracy.
  • Comparing WRF calculations by sea level pressure, it was suggested that when low pressure systems or typhoons pass through, WRF tends to slightly overestimate air density by overestimating pressure.

(Written by Takeyuki Misaki)

References

  1. Nippon Kaiji Kyokai, Wind Farm Certification: Onshore Wind Power Plants, 2021, p.41, URL: https://www.classnk.or.jp/hp/pdf/authentication/renewableenergy/ja/windfarm/NKRE-GL-WFC01_July2021_Jpn_Corrected_October2021.pdf(Accessed: February 2022, 2).
  2. Skamarock, W., Klemp, J., Dudhia, J., Gill, D., Barker, D., Duda, M., Huang, X., Wang, W., Powers, J. A description of the advanced research WRF version 3. NCAR Tech. Note NCAR/TN-475+ STR. 2008.
  3. The NEWA project, URL: https://www.neweuropeanwindatlas.eu/(Accessed: February 2022, 1).
  4. NEDO, Offshore Wind Condition Map NeoWins, URL: http://app10.infoc.nedo.go.jp/Nedo_Webgis/top.html
    (Accessed: February 2022, 2).