Sensitivity Analysis of Short-duration Heavy Rainfall Related Diagnostic Parameters with Point-area Verification

The knowledge about the short-duration heavy rainfall related diagnostic parameters is very important for improving the accuracy, and it can help understand the possible mechanism of meso-scale system producing short-duration heavy rainfall. The data collections of basic datum stations (verification stations), automatic meteorological observation stations (AMOSs), and 6-hours NCEP final analysis data (FNL) diagnostic parameters from 2011 to 2012 during 1 June and 31 August are used. By considering characteristics of rain gauge distribution, the total precipitable water and the best lifted index obtained from NCEP FNL are firstly analyzed with the "point-area" verification method to reveal the sensitivities of short-duration heavy rainfall to the environment conditions. Values of diagnostic parameters for a specific basic datum stations (verification stations) is obtained by adopting bilinear interpolation method.Results show that the popularly used verification is just special cases of "point-area" verification: One could be reached by setting infinitesimal searching radius, the other can be reached by setting the record threshold infinite. Both the total precipitable water and best lifted index have optimum thresholds indicating short-duration heavy rainfall, and the short-duration heavy rainfall in 3 hours can only be directly affected by the moisture and instability within the radius of 140 km. A searching radius and a record threshold of 140 km and 2 are supposed, respectively, for 1°×1° NCEP dataset. A total precipitable water of 55 mm divides the threat score (T) into monotone increasing and monotone decreasing parts, indicating under-forecast and over-forecast, respectively. A best lifted index of -2 ℃ divides the threat score into over-forecast and under-forecast parts. It is found that the total precipitable water and K index are equal better while both got the same highest threat scores of 0.275 with the bias (B) desirable and the false alarm ratio (F) and the hit rate (H) in the reasonable range. Short-duration heavy rainfall is most sensitive to parameters concerning the environment water vapor, nine of the top ten diagnostic parameters are water vapor related parameters. Parameters indicating environment instability conditions are also influencing, but parameters used to represent dynamic conditions and vertical wind shear conditions are lower ranked.