Development Of Algorithm For Non-precipitation Meteorological Echoes Identification With Doppler Weather Radar

The ability to monitor and warn severe weather events has been greatly improved due to the deployment of China Doppler weather radar (CINRAD). With the capability of CINRAD to detect improved, when it is clear or there are clouds in the air but no precipitation on the ground, wide spread non-precipitation meteorological echoes are often observed near the radars, which have great effect on estimation of precipitation and assimilation of radar data. In order to discriminate these echoes efficiently on the basis of analysis of the characters of these echoes, a non-precipitation meteorological echo identification algorithm based on fuzzy logical and Storm Cell Identification and Tracking (SCIT) is developed with data observed by the SA radar in Beijing.The non-precipitation meteorological echoes (NPME) in this paper refer to the echoes that are observed by radar with no precipitation on the ground. When these echoes appear, in most cases, there is little cloud in the sky. These echoes are also likely to be detected when it is cloudy, and some of these echoes may be caused by insects and birds. NPME observed by the SA radar in Beijing are mostly clear air echoes (CAE), so the characters of CAE near Beijing and the relationship between CAE and weather conditions on the surface are analyzed in this paper. The results show that the CAE in Beijing has obvious diurnal and seasonal variation. In November, when the daily average surface temperature dips below0℃, the CAE is no longer observed. When the wind speed exceeds6m/s, the CAE is also disappeared.When the identification starts, echoes are assembled into pieces using SCIT firstly, and then if an echo piece meets one of some special conditions, the whole plan position indicator (PPI) would be recognized as precipitation echoes. If one PPI can not be recognized as precipitation echoes, the attributes value of echo pieces will be calculated with some membership functions. The threshold of echoes in the piece would be calculated with the attribute value of this echo piece. If the attribute value of one piece is greater than0.5or equal to0.5, the echoes in the piece would be recognized with the threshold0.5. If the attribute value of one piece is less than0.5, the echoes in the piece would be recognized with a threshold gotten by subtracting the attribute value of the piece from1. The echoes would be recognized as non-precipitation meteorological echoes if their attribute values are greater than or equal to the threshold. After using dynamic thresholds, the thresholds of most precipitation echoes would be greater than0.5. With this method, we can avoid the precipitation echoes being recognized as non-precipitation echoes efficiently. While, it should also be noticeable that the thresholds of some non-precipitation echoes would be greater than0.5, which would decrease the identifiable accuracy of non-precipitation echoes to some extent.The algorithm does well with most of non-precipitation meteorological echoes and precipitation echoes, but sometime it handles some weak precipitation echoes improperly. Compared with the insect clear air return detection algorithm (ICADA) used by National Center for Atmospheric Research (NCAR), identifiable accuracy for non-precipitation echoes can be improved remarkably after using dynamic thresholds, and the erroneous recognition for precipitation is also decreased obviously. Although the membership functions are obtained with the data observed by the SA radar in Beijing, it does well with the SA radar in Hangzhou.The method is tested in an operational mode in this paper. The runtime of one file is commonly below2s, which can meet the need of operational deployment. While, some shortcomings in this method are that:the parameters used in the calculation of the attribute values of echo pieces are not enough; it also should be improved on the identifications of weak precipitation echoes and strong non-precipitation meteorological echoes. To the radars in other districts, the parameters and membership functions may require further adjustment.