Verification of Multiple Doppler Radar Analysis to Retrieve Three-dimensional Wind Field with High-spatiotemporal Resolution Capabilities of X-band Radar Network in Japan

For retrieving three-dimensional wind field from multiple Doppler radar data using variational technique, there are two types of input data format; one is the data format of the plan position indicator (PPI), and the other is the data format of the constant altitude plan position indicator (CAPPI). For the former method (e.g. Gao et al., 1999), error due to the transformation from the polar coordinate system to the rectangular coordinate can be avoided. However, when the spatiotemporal resolution of PPI data is coarse, the former method can be very limited in the retrieval of the wind field because of its discontinuity in the volume scanning. For the latter method (e.g. Shimizu et al., 2007), error caused by the coordinate transform is severely worse than the former method. However, even when the spatiotemporal resolution of PPI data is coarse, the latter would provide more spatially continuous wind field and might provide better solution. The comparative advantages between two methods should be summarized for the various scan strategies (especially for high-spatiotemporal strategy of next generation radars). In this study, three-dimensional wind filed retrieved by multiple Doppler radar analysis from PPI data and CAPPI data using variational technique are verified. The suitable analysis method for high-spatiotemporal PPI data would be discussed.

This work is the case study for the meso-scale convective system around Nagoya, Japan at May 10, 2011. This case was captured by three Ministry of Land, Infrastructure, Transport and Tourism (MLIT) X-band multi-parameter (MP) Radars. The resolution of range and azimuth is 150 m and 1.2 degree, respectively, and the observation range is 80 km. The temporal resolution of three-dimensional volume scanning (twelve elevation angles) is 5 minutes (the lowest two elevation angles are scanned every one minute alternatively so that wind field at the lower levels could be retrieved every one minute). The variational technique used for the retrieval of three-dimensional wind field is based on Gao et al. (1999). The horizontal resolution of the analysis is 1 km and the vertical resolution of the analysis is 250 m. The data from the wind profiler at Nagoya is used for the verification of the horizontal components of the retrieved wind field. For the verification, the mean error (ME) and root mean square error (RMSE) are calculated for the horizontal wind filed (U, V) from 500 m AGL to 1500 m AGL at 10 minutes intervals.

From the time series of the observed and retrieved wind fields, it is found that the wind field retrieved from PPI data (hereafter PPI retrieval) is accurate at 500 m AGL, but the wind field retrieved from CAPPI data (hereafter CAPPI retrieval) is overestimated at the same level. On the other hand, CAPPI retrieval is accurate at 1500 m AGL, however PPI retrieval is underestimated at the same level. The analysis of RMSE shows that RMSE of PPI retrieval is gradually increased as height and RMSE of CAPPI retrieval is gradually decreased as it becomes the higher level (from 500 AGL to 1000 m AGL). The analysis of ME indicates that CAPPI retrieval is overestimated as it becomes the lower level and PPI retrieval is underestimated. For CAPPI retrieval, the stronger wind at high levels might influences the retrieved wind at the lower levels so that the retrieved wind is overestimated. For PPI retrieval, because PPI data might be coarse at the higher levels, the accuracy is poor at the altitude beyond 1000 m AGL. For future work, PPI data from the lowest two elevation angles scanned every one minute alternatively would be used to retrieve three-dimensional wind field every one minute.