P2.1
The impact of the assimilation of precipitation data and Radar reflectivity with a pre-operational 4DVAR for the JMA nonhydrostatic model

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 1 February 2006
The impact of the assimilation of precipitation data and Radar reflectivity with a pre-operational 4DVAR for the JMA nonhydrostatic model
Exhibit Hall A2 (Georgia World Congress Center)
Yuki Honda, Japan Meteorological Agency, Tokyo, Japan; and K. Koizumi

The Japan Meteorological Agency has been operating a mesoscale numerical weather prediction (NWP) system for disaster prevention caused by severe weather. The system consists of a forecast model and an assimilation system. Currently the JMA nonhydrostatic model (JMA-NHM) is used as a forecast model (Saito et al. 2005). The assimilation system is a four dimensional variational data assimilation system for a hydrostatic model (Meso 4D-Var) (Ishikawa et al. 2005). As a successor system, a new pre-operational "JMA Nonhydrostatic model"-based variational data assimilation system (JNoVA) is being developed (Honda et al. 2005).

One of natural disasters in Japan is heavy rainfall. Accurate quantitative precipitation forecast (QPF) is one of the most important missions issued to the mesoscale NWP system. For this purpose, the assimilation of precipitation data is a vital factor to improve the short-range forecast. Koizumi et al. (2005) showed that the direct assimilation of precipitation data improved the QPF and ameliorated the spin-up problem of model precipitation. Therefore, the direct assimilation of precipitation data is the imperative issue assigned to the JNoVA.

The moist physics of the nonlinear model of the current JNoVA is simplified one: Large-scale condensation and moist convective adjustment (MCA) scheme. The predicted moist variable is only the mixing ratio of the water vapor. As shown in Honda et al.(2005), the current JNoVA succeeded to assimilate the total precipitable water data from SSM/I and the relative humidity data from radiosonde observation. However, there are two problems in the JNoVA with simplified moist processes. One is model-predicted precipitation area. It is important that predicted precipitation area is close to the observation because precipitation data can be assimilated only in the predicted precipitation area. But the JMA-NHM with simplified moist physics cannot produce well-qualified precipitation area compared with that by the JMA-NHM with microphysics. The other problem is the kind of analyzed moist variables. More and more observation data are available these days and some of them require hydromete! ors to be analyzed to describe the relationship between observation and model variables, like Radar reflectivity data. For these reasons, we upgraded the moist processes of the JNoVA.

Although the operational JMA-NHM uses 3-ice bulk microphysics, we decided to adopt 2-ice bulk microphysics considering the simplicity and efficiency of the microphysics. Besides the water vapor, another four hydrometeors of cloud water, cloud ice, rain and snow, are analyzed. The prediction of graupel is, of course, crucial to generate intense heavy rainfall. However, the life time of this kind of convection is too short compared with the supposed assimilation window, which is about 3 hours. So we omitted graupel from the microphysics of the JNoVA. A comparison of the simulations of the heavy rainfall event using the JMA-NHM with two different microphysics shows that there are no large differences in the predicted precipitation amount. As for the cumulus convective scheme, the Grell scheme is adopted in the JNoVA to replace with the current MCA. This is because the Grell scheme works well with resolution of ten to several tens kilometers and is less complicated than the Kain-Fritsch scheme which is used in the operational JMA-NHM.

Two kinds of precipitation data are assimilated with the JNoVA. Both of them are currently assimilated with the Meso 4D-Var in the operational system. One is the objectively analyzed precipitation data that are estimated precipitation data by the weather Radar calibrated with the in-situ precipitation data from the Automated Meteorological Data Acquisition System (AMeDAS). The AMeDAS is the high-density surface observation network covering Japan. Precipitation is observed at all AMeDAS stations, the average resolution of which is about 17km. These data are called Rader-AMeDAS precipitation analysis data. The other is the rainfall intensity data retrieved from satellite microwave imagers.

We will show results of data assimilation experiments of two kinds of precipitation data with the pre-operational JNoVA and would like to discuss its impact to the QPF by the JMA-NHM comparing with the QPF from the analysis of the Meso-4DVar, which is the operational one.

As described above, there is an inevitable problem that precipitation data can be assimilated only in the model-predicted precipitation area. To mitigate this problem, the assimilation of three-dimensional Radar reflectivity might be helpful since the Radar reflectivity is related with hydrometeors by the empirical relationship. The upgrade of moist processes in the JNoVA from the large-scale condensation to the bulk microphysics enables us to do this assimilation experiment. The effect given to the efficiency of the assimilation of precipitation data will be investigated and also reported in the presentation.

[Reference]

Honda et al., 2005: A pre-operational variational data assimilation system for a nonhydrostatic model at Japan Meteorological Agency: Formulation and preliminary results. (submitted to Quart. J. Roy. Meteor. Soc.)

Ishikawa et al., 2005: The JMA mesoscale four-dimensional variational data assimilation system. (in preparation)

Koizumi et al., 2005: Assimilation of precipitation data to JMA mesoscale modle with a four-dimensinal variational method and its impact on precipitation forecasts. Scientific Online Letters on the Atmosphere, 1, 45-48.

Saito et al., 2005: The operational JMA nonhydrostatic mesoscale model. (submitted to Mon.Wea.Rev.)