Statistical Analysis of the Mesoscale Convective Systems Propagation East of the Rocky Mountains


 In this work, we characterized the occurrences and propagation speeds of Mesoscale Convective Systems (MCSs) east of the Rocky Mountains, using 15 years of radar data. The central United States has a complex topography. The region also has atmospheric environments that initiate and maintain MCSs at multiple scales. The diurnal and regional variability of MCSs based on their longevities was obtained using high-resolution observation data (Stage IV) and an object tracking algorithm MODE-Time Domain (MTD). MTD-determined MCSs in spring and summer were divided into daytime (initiated from 12 to 23 UTC, MCS12) and nighttime MCSs (formed between 00 and 11 UTC, MCS00) and into short lived (less than the 75th percentile) and long lived MCSs (greater or equal to the 75th percentile). Propagation speeds of MCSs were calculated using distances between MCSs’ centroids at each time step. We suggest a novel way to obtain a Hovmoller diagram to indicate average propagation speeds. There were two key results: 1) Spatial and temporal features of propagation speeds vary at each location and time and, 2.) heavy rainfall (rain rates ≥ 5.0 mmhr-1 ) contributed more than lighter rainfall to overall precipitation. In the east during spring, long-lived MCSs occurred more frequently in the spring than in summer. Short-lived daytime MCSs in spring and summer exhibited similar spatial distributions. In summer alone, short-lived nighttime MCSs occurred more frequently that they did in spring. To the east, the average propagation speeds of short-lived MCSs increased in spring and summer, whereas long-lived MCSs indicated decreasing trends.

Historically, Stage IV data were incorporated into data assimilation. However, the 126 advantages of high temporal and spatial resolution of data enabled their further 127 use to accurately measure precipitation and to compare these measurements to 128 ground station-, radar-, and satellite-based algorithms (Nelson et al., 2016). Some 129 studies have compared the precipitation systems from observational data and Re-  2) defining objects with a user-defined threshold (conv thresh=5.0 mmhr −1 ).

142
3) identifying and numbering contiguous objects in the time domain, considering 143 the minimum three-dimensional object size (i.e., latitude, longitude, and time), 144 while the minimum size is predefined by a user (min volume=2000).

145
The precipitation systems determined by MTD are sensitive to the set-up val-146 ues in a configuration as "conv radius", "conv thresh", and "min volume". When 147 the smoothing radius and threshold are set to larger values, only a few objects 148 are remained by smoothing and thresholds (Prein et al., 2017). In this study, the 149 configuration is designed to detect MCSs related to extreme precipitation systems.

151
A Hovmoller diagram is a method to plot meteorological data to highlight the role 152 of wave-like features. The data are shown as axes of longitude or latitude (abscissa 153 or x-axis) and time (ordinate or y-axis). The values are indicated as shadings to emphasize features propagating in the east-west or south-north direction with 155 respect to time. In this study, the propagation of a precipitation system is repre-156 sented as a Hovmoller diagram to indicate longitudinal movement with time. The 157 method to obtain representative occurrences in the longitude uses equations as: t =time in UTC (00 to 23) i =1 ... m, index of Longitude (i.e., west to east) j =1 ... n, index of Latitude (i.e., south to north).
(1) while RR is the rain rate in mmhr −1 in three dimensions in time, latitude, 159 and longitude. Using this method, (OCC in (1)), we obtained the occurrences 160 of rain rates over 5.0 mmhr −1 at each location. In this study, we obtained two- (2) Note that Spd(t, i, j) is an average propagation speed in time, longitude, and       (Fig. 5g, h). Note that the highest values of occurrences 313 between 99 and 96 • W were in eastern Texas to Oklahoma in MAM (Fig. 4g) and 314 in Oklahoma to Kansas in JJA (Fig. 4h) for MCS12.

Statistics of MTD determined MCSs
however, there were fewer than 2 yr −1 (i.e., less frequently occurrences). MCS00l 361 in JJA exhibited increasing propagation speeds in the north, whereas MCSs near 362 the coastal area showed propagation speeds of less than 8 ms −1 (Fig. 9d). Similar 363 features were seen in MCS12l in JJA, which showed lower propagation speeds in 364 the coastal areas and higher speeds in Iowa and to the northeast, as seen in Fig.   365 9h. The propagation speeds of MCS00l in MAM did not appear to be related to the east (lower than 8 ms −1 (in Fig. 10h). bell-shaped occurrences and peaks at 96 • W (Fig. 11f) Fig. 9, but represented in Hovmoller diagrams. Note that y-axes of a) to d) start from 00 UTC and e) to h) from 12 UTC. Occurrences of MCSs are superimposed as black contours from 50 to 150 counts