NORTHEAST SEVERE WEATHER DISTRIBUTION AS A FUNCTION OF FLOW REGIME

Alicia C. Cacciola*, Lance F. Bosart, Sheryl F. Honikman, Thomas J. Galarneau, University at Albany/SUNY, Albany, New York

Kenneth D. LaPenta and John S. Quinlan, National Weather Service, Albany, New York

1. Introduction

It has been noticed that severe weather occurences in eastern New York and western New England are affected by the underlying terrain as a function of the prevailing large-scale flow patterns. The Hudson-Mowhawk river valleys and surrounding mountains can cause significant modifications to the low- and mid-level large scale flow. This study is a climatology of severe weather events in a box which encompasses the Hudson-Mohawk valley, and parts of western New England (41.5?N to 43.5?N, 72.5?W to 75.5?W). (See Fig. 1). The period of the climatology is from 1955 to 1998. It is hoped that the results of this study will aid forecasters in eastern New York and western New England in anticipating preferred areas for severe weather, particularly tornadoes, based upon the direction of the low- and mid- level flow.

2. Data and Methodology

All wind data used in this study is obtained from the "Radiosonde Data of North America" CD-ROM, provided by the National Climatic Data Center. Dates and times of severe weather events in the previously defined box were obtained from the Storm Data database, and plotted using the PLOTSVR program, written by Glenn Wiley of the National Weather Service Forecast Office in Albany, New York. Population density data was obtained from the 1990 census CD-ROMs. The previously defined latitude/longitude box was divided into 0.5? by 0.5? grid boxes, overlapping every 0.25?. Based on the block-group level data, area average population density was computed for each box.

For each available severe weather report, the wind direction at 700 hPa, 500 hPa, and 850 hPa was obtained by using the closest sounding time to the event. Events which occured prior to 1400 local time were defined to be closest to the 12Z sounding, and events which occured after 1400 local time were defined to be closest to the 00Z sounding. There is a potential for errors to occur if a frontal passage occurs between the 12Z morning sounding and the 00Z evening sounding. The events which are closer to the 00Z sounding could still potentially have been in the pre-frontal passage environment. Tests are being done on subsets of the dataset which are close to the sounding time to see if the subset is representative of the whole set.

Wind rose plots were created for 850 hPa, 700 hPa, and 500 hPa flow. Wind directions were divided up into thirty six 10? bins, from 0? to 360?. The percent of high wind, hail and tornado events which occured in each bin was totalled and plotted on wind roses. Each day on which severe weather occured in the geographic area in this study was defined as a severe weather day. For each 0.5? by 0.5? box, the number of events per severe weather day was totalled for northwest flow days (greater than 280?), west flow days (260? to 280?),and southwest flow days (less than 260?) at 700 hPa. The 700 hPa level was chosen with the idea that it is high enough to reflect some of the large scale flow pattern, yet low enough to feel the influence of the topography. In addition, the ratio of events per southwest flow day to events per northwest flow day was calculated in order to identify which areas have a preference for severe weather to occur for different 700 hPa flow directions.

A population density correction was performed on the Storm Data following the method used in Paruk and Blackwell (1994). The number of events was plotted against population density, and a linear regression analysis was performed. Based on the regression equation, a correction factor was developed which, when multiplied by the actual number of events, produced an 'adjusted' number of events. This corrected value assumes that all the events in the box with the highest population density are reported, and adjusts each of the other boxes to the number of events which would be reported if the population density was equal to that in the highest box.

3. Results

When the database of severe weather events was stratified by 700 hPa flow direction, some interesting results appear (see Figures 2 and 3) There is a distinct preference for severe weather to occur north of 42.5?N (in the Adirondack and southern Green mountains) on a day with southwest flow at mid-levels. On days with northwest flow, the preferential areas for severe weather to occur are clearly south of 42.5?N, into the Catskills and Berkshires. An interesting result is the pocket of preference for northwest flow days in south central Vermont, extending southeast into Massachussets. This preference for severe weather to occur on northwest flow days may be a result of the Housatonic river valley, which runs north-south through the Berkshires. Low-level terrain-channeled southerly flow in the valley beneath northwest flow aloft may help to create a favorable shear flow for severe weather in this region (Bracken et al., 1998). This maxima is approximately centered in and just downwind of the valley. Low-level terrain-channeled southerly flow in the valley beneath northwest flow aloft may help to create a favorable shear flow for severe weather in this region (Bracken et al., 1999). Interestingly, while the population corrected plots did change the overall shape of the severe weather distribution by eliminating the maxima associated with population centers, the ratio of events per southwest flow severe weather day to events per northwest flow severe weather day changed very little. (see Figures 2 and 3).

Figures 4 to 6 are wind rose plots for 850 hPa, 700 hPa, and 500 hPa. Few or no severe weather events occur when the flow at any level has an easterly component. For each type of severe weather event (high wind, hail, and tornado), there is a high percentage of events which occur in southwest flow at 500 hPa, indicating that an approaching trough is preferential to other types of flow patterns. At 700 hPa, over 20 percent of tornado events occured when the flow was out of the southwest, suggesting that there is a strong preference for southwest flow. However, another small peak is apparent at 290?, perhaps suggesting that a preference exists for tornadoes to occur when there is an approaching shortwave trough in west-northwesterly flow. The distribution of hail events is also bimodal at 700 hPa, although the peaks are more equal in magnitude. The high wind event plot has a much more rounded appearance than the hail or tornado events plots at 700 hPa. This may indicate that as long as the flow has a westerly component, there is not a preferential direction for wind events to occur. At 850 hPa, each type of event shows a peak at due west (270?), but hail events show a peak at northerly flow, while wind events show a clear peak in the south-southwesterly direction. Tornado events show a peak in the northwesterly direction, suggesting that perhaps the Mohawk Valley has some importance in modifying the flow direction at low- and mid-levels so as to be more conducive to tornado development in a favorable environment. It is interesting that a certain percentage of hail events seem to show a preference for north and north-northwesterly flow at 850 hPa. The intrusion of colder, drier air may actually help the formation of hail in an environment where storms are already occuring.

4. Conclusion

The Hudson-Mohawk valley and surrounding mountains appear to have a quantifiable effect on the occurrence of severe weather in upstate New York and western New England. Funneling of the low-level flow can create more favorable shear profiles for tornadoes to occur in certain areas. The preferential occurrence of severe weather based on the 700 hPa flow direction also appears to be of an orographic nature. The Adirondack, Green, Catskill, and Berkshire mountains have a significant effect in initiating convection when the wind is in a favorable direction for upslope flow to occur.

5. Acknowledgements

This research was supported in part by COMET grant # 09915806.

REFERENCES

Bracken, W.E., Lance F. Bosart, Anton Seimon, Kenneth D. LaPenta, John S. Quinlan, John W. Cannon, 1998: Supercells and tornadogenesis over complex terrain: The Great Barrington (Massachusetts) Memorial Day (1995) tornado. Preprints from the 19th Conf. on Severe Local Storms, Am. Met. Soc., 18-21.

King, Patrick, 1997: On the absence of population bias in the tornado climatology of southwestern Ontario. Weather and Forecasting,12, 939-946.

Paruk, B.J., and S.R. Blackwell, 1994: A severe thunderstorm climatology for Alberta. National Weather Digest, 19, 27-33.

Sterner, R., 1995. Color landform atlas of the United States. http://fermi.jhuapl.edu/states/states.html.

Whiteman, C. David, and J.C. Doran, 1993. The relationship between overlying synoptic-scale flows and winds within a valley. Journal of Applied Meteorology, 32, 1669-1682.