Effect of microphysical schemes on simulation of a rainfall process in the central parts of the Democratic People's Republic of Korea


Chairman Kim Jong Il said:

"The hydrometeorological sector should forecast flood and weather accurately and enhance the level of promptness and scientific character of the hydrometeorological data by furnishing itself with modern weather observation equipment."

It is very important to simulate microphysical processes properly to improve the accuracy of precipitation forecasting by numerical forecasting models.

In the present study, we analyzed the impact of the microphysical schemes of the worldwide widely used weather study and forecasting model, WRF version 3.6, on the simulation of a rainfall processes in the central part of DPRK.

The microphysical schemes used in the study are the WDM6 scheme, the Thompson scheme and the SBU-YLin scheme, and each simulation is for 45 h from 0000 LST (local standard time: UTC + 9 h) on 11 to 2100 LST on 12 August 2009.

The initial and lateral boundary conditions were obtained from the National Centers for Environmental Prediction (NCEP 2.5° × 2.5°) reanalysis data

The model was configured with three interactive nested domains (Fig. 1a) with 34 vertical levels. The outer domain was configured with 27km resolution. The second and third domains had a horizontal resolution of 9 and 3 km, respectively.

We use Noah land surface model, shortwave radiation, and the Rapid Radiative Transfer Model (RRTM) longwave parameterizations in all domains. The Kain-Fritsch cumulus parameterization scheme was used for the outer domains (27 and 9 km). In the inner domain 3, only microphysics was employed and no convective parameterization scheme was used. The YSU planetary boundary layer scheme is used in each domain. Other schemes are similarly taken and three simulations are performed by choosing different microphysical parameterization schemes.

The observed data used in the study are 45 h cumulative precipitation observed at 29 stations in the central part of DPRK from 0000LST on 11 August 2009 to 2100LST on 12 August 2009.

The observation data and simulated data were evaluated to determine the statistical skills including probability of detection (POD), false alarm ratio (FAR), and BIAS.

Three nested model domains
Fig. 1 (a): Three nested model domains at 27, 9 and 3 km resolution. (b): the corresponding model orography (unit: m) and the location of the land stations (red dots) in the inmost domain and the location of the land stations (yellow dots) over the study region

The analysis results are as follows:

-The SBU-YLin scheme simulates the amount of rainfall remarkably similar to the observations compared to the Thompson and WDM6 schemes. For the 29 weather stations in the central part of DPRK considered, the SBU-YLin scheme has the highest POD and the lowest FAR for all rainfall thresholds.

accumulated precipitation
Fig. 2 45-h accumulated precipitation (mm) for the whole simulations from 0000 LST on 11 to 2100 LST on 12 on 11 August 2009: (a) observation, (b) WDM6, (c) Thompson, (d) SBU-YLIN

-The SBU-YLin and Thompson schemes simulate the spatio-temporal distribution of hydrometeors well compared to the WDM6 scheme.

The results of this case study show that the amount of rainfall simulated by WRF model are different from three microphysics schemes, and the SBU-YLIN scheme produces a more reasonable description of precipitation over the study area.

Our study results were published in the journal "Meteorology and Atmospheric Physics" (2020, 133(4): 1159-1175) under the title of "Effect of microphysical schemes on simulation of a rainfall process in the central parts of the Democratic People's Republic of Korea" (