The warm season across the northern Plains is a favorable time for thunderstorms. A significant portion of thunderstorm precipitation in this region is attributable to mesoscale convective systems (Fritsch et al. 1986). An exceptionally strong El Niño cycle which slowly faded by the Fall of 1998, brought about a larger than normal amount of rainfall to parts of the Plains and Midwest. This was suggested by Anderson and Arritt (1998) to be due to an above normal frequency of MCS's across parts of the nation. A noteworthy case was an MCS event that occurred over southwestern and south central North Dakota on August 21-22, 1998. It was a nocturnal episode that produced flash flooding across parts of south central North Dakota, breaking daily and monthly rainfall records for Bismarck, and flooding parts of the city and a few surrounding communities. This paper will attempt to review the dynamics behind the flash flooding and how it could have been forecasted.
Data for this case was collected for various parameters for the aforementioned dates, including 0000 and 1200 UTC SHARP soundings for Bismarck (BIS), North Dakota, Aberdeen (ABR) and Rapid City (UNR), South Dakota, and Glasgow (GGW), Montana. The SHARP data ultimately led to analysis of upward vertical velocity, Theta-e, relative vorticity, average 1000-500-mb relative humidity and precipitable water values. The K, Lifted and Showalter Indices were also of importance to show the stability prior to and during the event. Additionally, upper air data from Minneapolis and International Falls, Minnesota, Great Falls, Montana, Riverton, Wyoming, North Platte and Omaha, Nebraska were used to plot 850-mb, 500-mb and 250-mb charts for 0000 and 1200 UTC
A. Convective Outlook Discussion
Figure 1 shows the Storm Prediction Center's (SPC) Convective Outlook for 1451 UTC on 21 August. Clearly, the threat for severe convection was forecast over extreme eastern Montana into western North Dakota with general thunderstorms expected around this area. The potential was based on the fact that an Idaho-Wyoming shortwave at 500 mb was expected to generate large scale upward vertical motion and destablization to the threat area by evening (Figure 2). Surface heating along with a moist southeasterly flow was to continue across the region as well. A warm front was expected to lift northeastward to just south of North Dakota by late evening and would be the focus for nocturnal warm air advection (WAA) and the lift necessary for severe convection. At the time of the outlook, the potential existed for large hail and damaging winds.
B. 1200 UTC 21 August 1998 Parameters
A surface analysis showed that an area of high pressure was situated over southern Manitoba with basically a low-level southeasterly flow across North Dakota (Figure 3). A warm front also stretched southeastward from developing low pressure in western Montana across southeastern Montana into eastern Wyoming and north central Nebraska. Scattered showers and thundershowers were occurring along and ahead of this front. Some isolated showers and thundershowers were also observed by the Bismarck WSR-88D over parts of south central North Dakota during the morning. While much of this activity was weak, the significant weather threat waited until the evening hours across south central North Dakota. The evening is a favorable time for much of North Dakota's summer thunderstorms. A study by Czarnetzki (1998) has shown that many nocturnal thunderstorm events produce heavy rain which can lead to flash flooding. This turned out to be true in this event.
As can be seen in Table 1, the initial SWEAT Indices at GGW and UNR were 439 and 493, respectively. Along with Lifted and Showalter Indices around -5°, it appeared that the atmosphere was already quite unstable and would only destabilize further with afternoon surface heating. High K values (over 30) were also observed at GGW, UNR and BIS, signaling the potential for severe weather and/or heavy rainfall. It is interesting to note from the 1200 UTC data that precipitable water values were over one inch with average 1000-500-mb relative humidities at or above 50 percent from eastern Montana into western South Dakota and south central North Dakota. This high atmospheric moisture content should alert Forecasters to the potential for heavy rainfall. An 850-mb analysis for 1200 UTC showed good WAA was taking place across the threat region along with a developing low-level jet. These two factors are just a few suggested precursors to MCS development (Maddox 1983). Further research by Maddox shows that an east-west oriented frontal boundary along with rather weak, westerly upper level flow (~50 kts) are prime ingredients for MCS development north of the surface boundary. A moist and conditionally unstable environment in this region will also aid in the development of significant thunderstorms. Theta-e analysis at 1200 UTC (Figure 4) showed that strong ridging of theta-e was in place across the outlook area, which again suggested an unstable environment. This theta-e maximum along with a vorticity maximum was forecast by the NGM and ETA models to move into North Dakota by 0000 UTC 22 August. This coupled with a shortwave lifting northeastward from Idaho and Wyoming would only add to the potential for significant weather in North Dakota. A Mesoscale Convective Discussion (MCD) issued by the SPC at 1855 UTC, suggested that there was a threat for severe thunderstorms from Montana into Wyoming due to CAPE values around 2500 J/kg and a continued moist southeasterly surface flow over the area. As it turned out, no severe thunderstorm watches were issued for areas west and southwest of North Dakota during the afternoon.
Unmodified sounding data from SHARP for 1200 UTC 21 August 1998
|Precip. Water (in.)||1.17||1.23||0.85||1.22||0.86|
|1000-500 MB RH (%)||52||55||46||55||41|
|250 MB Winds (kts)||230/41||250/53||240/51||255/49||260/54|
|Theta-e (850 MB)||340.1||329.8||320.3||344.1||321.9|
|850 MB Winds (kts)||160/12||155/18||135/03||165/23||285/07|
|Max UVV (m/s)||52||0||0||38||0|
From a heavy rainfall perspective, the above mentioned high precipitable water values coupled with K values in the 30s and a moist, low-level jet (~25 kts), would alert most forecasters to a heavy rain potential. Funk (1991) states that these factors are necessary for producing heavy rainfall. Heavy rainfall usually occurs to the north of an east-west surface front as moist air is transported northward up and over this boundary, such as by a low-level jet, resulting in what is commonly called "overrunning". To further assess the potential for heavy rainfall, a Flash Flood Checklist was used. This checklist utilizes the K index and precipitable water values along with wind direction and speed shear for the upper levels. Upon tabulating the 1200 UTC parameters, "No Excessive Rainfall Expected" resulted. Though this table provides a first guess forecast, the reliability of a forecast of heavy rainfall will be highest when the total score is comprised of 'non-zero' values in four or more of the scoring areas. Flash Flood Guidance was not found for this case.
C. 0000 UTC 22 August 1998 Parameters
Surface analysis for this time showed the western Montana low pressure system had moved to north central Montana. A warm front had lifted northeast to southeastern Montana and west central South Dakota. The BIS, GGW, and UNR WSR-88D radars were all detecting an increase in shower and thunderstorm activity by 0000 UTC. New upper air data showed that the greatest instability had shifted to the Dakotas, with SWEAT values of 388 and 454 at BIS and UNR, respectively (Table 2). The Lifted Index had destabilized greatly at BIS from a +4 at 1200 UTC to -4 by 0000 UTC. Of greater interest was that CAPE values increased to 1567 J/kg at BIS from 0 J/kg at 1200 UTC. A look at the UNR SHARP data showed a CAPE value of 6070 J/kg, which was extremely unstable. Of the five sites in Table 2, the BIS sounding had moistened nearly another 0.50 inches from 1200 UTC to 1.68 inches at 0000 UTC. In fact, the average 1000-500-mb relative humidity reached 68 percent over BIS and was the maximum of all the surrounding sites. It appeared at this time that the heavy rainfall potential was real.
|Precip. Water (in.)||1.11||1.68||0.85||1.24||M|
|1000-500 MB RH (%)||24||68||44||50||M|
|250 MB Winds (kts)||215/56||265/54||265/57||235/64||M|
|Theta-e (850 MB)||333.1||340.8||330.5||348.1||M|
|850 MB Winds (kts)||175/03||135/22||150/19||195/17||M|
|Max UVV (m/s)||43||56||56||110||M|
A look at the evening NGM and ETA model output suggested that the best dynamics were over south central North Dakota with Theta-e ridging over the Dakotas. The 850-mb analysis at 0000 UTC showed a warm front at this level in nearly the same location as at the surface for this time (Figure 5). It can be seen that a band of nearly saturated air extended from around Bismarck east-southeast to near northern Iowa was depicting a moist axis, and that a southerly low-level jet (25 kts) was feeding WAA into the region. It was noted that the winds at BIS and ABR had 'backed' slightly to the southeast in line with this moist axis from that at 1200 UTC. With 250-mb winds around 55 kts from Wyoming into North Dakota and Minnesota, it appeared, as at 1200 UTC, that there was an absence of a strong upper jet. This relatively weak, upper level jet necessary for MCS development continued to be a prime ingredient, as stated earlier. Another factor was that the ETA and NGM models were forecasting 500-mb Potential Vorticity Advection (PVA) into North Dakota by 0000 UTC which would supply additional lift needed to maintain the thunderstorm activity. Another evaluation of the in-office Flash Flood Checklist with new upper air data resulted in a category that "Flash flood producing rainfall is a strong possibility.....a watch might be considered with 2.5 to 3.5 inch flash flood guidance totals". A later MCD (2029 UTC) from the SPC was concerned about CAPE values around 2500 J/Kg and strong WAA from Dickinson to Minot and Bismarck, and that this area was being monitored for a possible severe thunderstorm watch. However, as the evening progressed, no watch was issued.
By 0100 UTC, a cluster of thunderstorms over southwestern North Dakota was detected by the BIS WSR-88D. As the convection was moving eastward towards the city during the next several hours, back-building convection appeared to be occurring around the Dickinson area. This was observed on several Water Vapor and IR satellite images from 2202 UTC through 0102 UTC during the beginning stages of the MCS. As the convection approached the city, frequent to continuous in-cloud lightning was lighting up the sky. The flash rate, as detected by StormTracker (Boll 1996), was up to about 50 flashes per minute. It was at 0130 UTC that the storms began at the airport. As the rainfall became heavier around 0230 UTC, ASOS recorded 0.82 inches in about a 48 minute span. During the next 12 minutes, very heavy rainfall with almost continuous in-cloud lightning occurred. In addition, another 0.51 inches fell, bringing the total rainfall to 1.33 inches. The BIS WSR-88D at 0429 UTC showed that the area of thunderstorms, west of the city one hour ago, had taken on the appearance of a bow echo as they were moving into the area (Figure 6). However, as the storms passed through, only heavy rainfall was reported by spotters, resulting in localized street flooding. Vertically Integrated Liquid (VIL) values derived from the BIS WSR-88D at 0231 UTC were showing values near 80s kg/m2, but it appeared that the lower levels of the atmosphere were too warm (+17°C at 850-mb and +9°C at 700-mb) to support much, if any, hail formation. Surface winds were around 20 mph and from the North during the storms with the VAD Wind Product (VWP) showing stronger, westerly winds from 30 to 50 knots from 7,000 to 10,000 feet. From the period of 0300 UTC to 0400 UTC, an additional 2.46 inches of rain had fallen as the storms slowed their movement to around 15 mph. After 0400 UTC, the rainfall had diminished to only light rain as the bow echo moved east. By midnight (0500 UTC), only an additional 0.02 inches fell. The total from this convective event reached 3.79 inches, which combined with 0.83 inches earlier in the day, resulted in a grand total of 4.63 inches for the day. This set a daily rainfall record and pushed the August monthly total to 9.29 inches, which was also a record.. The heavy rainfall was responsible for flooding parts of the city and airport. Figure 7 shows the distribution of rainfall for the past 24 hours across North Dakota. A look at the WSR-88D Storm Total Precipitation product showed the rainfall pattern well at 0504 UTC. By this time, the rain had stopped, and the MCS was moving southeastward along the surface boundary towards southern Minnesota, as depicted in Figure 8.
Figure 6. WSR-88D image at 0.5° elevation from NWS Bismarck, North Dakota for 0429 UTC 22 August 1998.
This study has shown the complex nature of the atmosphere when trying to predict MCS development. In this particular event, strong frontogenetic forcing resulting from a warm frontal boundary up to 850 mb to the south of North Dakota, moisture convergence, and approaching low pressure, set up the scenario for an MCS. This forcing has been stated as a precursor to MCS development by Augustine and Caracena (1994). While the meteorological parameters appeared favorable for large hail and high winds over southwest and south central North Dakota earlier in the day, the lack of a strong upper level jet and mid-level drying, and too much moisture in the lower levels, resulted in a heavy rain event on August 21, 1998. Rainfall amounts from 1.50 to nearly 3.75 inches in about a 3-hour period fell from the MCS across the southern areas of North Dakota.
Anderson, C.J., and R.W. Arritt, 1998: MCC and Other Large, Long-Lived MCS Over the U.S. During 1998. Preprints, 19th Conference of Severe Local Storms, Minneapolis, AMS (Boston), 34-35.
Augustine, A.A., and F. Caracena, 1994: Lower-Tropospheric Precursors to Nocturnal MCS Development Over the Central U.S. Wea. Forecasting, 9, 116-135.
Boll, R., 1996: StormTracker Lightning System. Software Version 1.15, Boltek Corporation, Buffalo, NY.
Czarnetzki, A.C., 1998: Evaluation of a Forecast Strategy for Nocturnal Thunderstorms that Produce Heavy Rain. Preprints, 19th Conference of Severe Local Storms, Minneapolis, AMS (Boston), 453-454.
Fritsch, J.M., R.L. Kane, and C.R. Chelius, 1986: The Contribution of Mesoscale Convective Weather Systems To The Warm Season Precipitation Of The U.S. Mon. Wea. Rev., 25, 1333-1345.
Funk, T.W., 1991: Forecasting Techniques Utilized by the Forecast Branch of the National Meteorological Center During a Major Convective Rainfall Event. Wea. Forecasting, 6, 548-564.
Maddox, R.A, 1983: Large-Scale Meteorological Conditions associated with Mid-Latitude, Mesoscale Convective Complexes. Mon. Wea. Rev., 111, 1475-1493.