Southern United States Heavy rain and flood event 6-8 April 2014 By Richard H. Grumm National Weather Service State College, PA 1. Overview Heavy rainfall (Fig. 1) affected the southern United States from 6-8 April 2014. The rain produced flooding which produced deaths in both Mississippi and Georgia (AP 2014a, AP2014b, Climate Progress). The heavy rains were associated with thunderstorms, isolated severe thunderstorms, and produced reports of over 350 mm (7 inches) of rain over the two day period. The locally heavy rains and flooding displaced thousands of people from flooded homes and mobile home parks as indicated in the new stories and imagery. The storms and flood waters downed power lines leaving thousands without power. The rain was associated with a deep trough (Fig. 2) which moved southwestern United States and across the central United States from 6-9 April 2014. East of this trough there was a ridge over the southwestern Atlantic associated with +1σ height anomalies. As the trough moved eastward the anomalies in the trough approached -4σ below normal (Fig. 2c). The flow between the trough and ridge caused a surge of moisture into the southern United States (Fig. 3) with precipitable water (PW) values ranging from 40 to 50mm with +2 to +3σ anomalies. The surge of high PW was associated with a strong low-level jet. As the deepening trough approached from the west, the 850 hpa winds increased (Fig. 4). Between 0000 UTC 7 to 1800 UTC 7 April strong southerly flow (Fig. 4b-e) in the high PW plume (Fig.3b-e) this resulted in high values of 850 hpa moisture flux (Fig. 5). The net result was nearly an ideal conceptual model supporting heavy rainfall and embedded convection (Bodner et al 2011; Stuart and Grumm 2009; Durkee et al 2012). The patterns associated with heavy rainfall are well known and relatively well studied in the literature. The key is in identifying these patterns in forecast systems and leveraging the information in these systems to produce forecasts. Due to sensitivity in numerical models and imprecise data used to initialize these models, each weather pattern and event has a unique predictability horizon (Palmer 2005; Lorenz 1969). Lorenz (1969) demonstrated the upscale error propagation in a forecast simulation and developed the concept of chaos theory (Lorenz 1993). Thus, the model or ensemble forecast system may correctly produce the larger synoptic pattern but be incapable of predicting the location of the high impact weather phenomena, which typically occur on a relatively mesoscale region. Ensemble forecasts have been shown to be the primary tools from which forecast uncertainty information can be extracted (Toth et al 2001)
This paper will document the heavy rainfall event in the southern United States on 6-8 April 2014 and the associated pattern. The ability of ensemble prediction systems to forecast this event is presented from a pattern and QPF probability reference frame. 2. Data and Methods The large scale pattern was reconstructed using the 00-hour forecast of the NCEP Global Forecast System as first guess at the verifying pattern. The standardized anomalies were computed in Hart and Grumm (2001). All data were displayed using GrADS (Doty and Kinter 1995). The focus on the larger scale pattern was to show the standardized anomalies and how they may help identify potential high impact weather events (HIWE). The NCEP GFS, GEFS, and SREF were used to examine the forecasts of this event. As the pattern was generally well predicted the focus is on the quantitative precipitation forecast (QPF) produced by these systems. The stage-iv data were used to get a first guess at where, when, and how much precipitation was observed. 3. Forecasts The overall pattern was generally well predicted by the NCEP forecast suites with lead-times related to a pattern favoring heavy rainfall and QPFs suggesting heavy rainfall implying at least a 5-day lead-time. The relatively course GEFS is used to illustrate the pattern. The GEFS 500 hpa forecast from 6 GEFS cycles (Fig. 6) show the GEFS correctly predicted a strong trough to cross the United States and be over the lower Mississippi Valley around 1200 UTC 7 April 2014. Due to uncertainty issues (not shown) these forecasts showed -1 to -2σ height anomalies in the trough. The spread about the mean indicated some timing and intensity issues were associated with the trough. It is interesting to note how well and how persistent the strong ridge over the West Coast was forecast. The forecast of the deep trough and ridge over the Atlantic resulted in a surge of high PW air (Fig. 7) and a strong low-level jet (Fig. 8) into the southern United States. The location of the low-level jet and moisture plume showed some variation over time. Though all forecasts indicated a pattern conducive for heavy rain, though the pattern was also forecast to be somewhat progressive implying the event would not endure in most areas for more than 24-36 hours. These GEFS forecasts were reflected in the GFS (not shown) which also produced similar features. This can be seen in the GFS and GEFS QPFs (Fig. 9) which indicted that GFS QPFs from 0000 UTC 3 April (Fig. 9a) forecast areas of 100 mm of QPF and the GEFS initialized at 0000 and 0600 UTC indicated the probability of areas of in excess 75 mm at the 40% level. Though not shown areas impacted by 50mm of QPF or more were in the 90% confidence level.
Shorter range forecasts from 04 April showed similar extreme rainfall in the GFS (Fig. 4d) and higher threats for 75mm or more of QPF from both the 0000 and 0600 UTC GEFS. Shorter range GFS, SREF and GEFS forecast from 0000, 0300, and 0000 UTC respectively on 5 and 6 April 2014 (Fig. 10) showed a continued threat for heavy rainfall with a more focused narrow axis in the GFS from the 5 April forecasts and areas of a high probability of over 50 mm in both the SREF (Fig. 10b) and GEFS (Fig. 10c). The forecasts for 06 April showed a similar agreement amongst forecast systems. Clearly the GFS and GEFS, sharing the same model core focused the heavy rainfall in the broader western band over the same general region. The eastern area of rainfall was larger in the GFS (Fig. 10d) relative to the focused area near the Florida panhandle in the GEFS (Fig. 10f). Similar forecasts but for the probability of 100mm of QPF (Fig. 11) showed considerable uncertainty as to where the heaviest rain would fall. The region ranged from Louisiana to Georgia. Mesoscale details were not as exact as the synoptic pattern or region. The GFS, GEFS, and SREF 1200 UTC and 0900 UTC cycles (Fig. 12) showed similar patterns and a similar region to be affected by heavy rainfall. Note the amounts in the GFS varied fromrun-to-run. The consistent feature or pattern was the threat for heavy rain over the same general region from forecasts over 4 consecutive days which likely provided confidence in the concept of heavy rainfall with some issues related to the details. 4. Summary A HIWE event, in the form of heavy rain and associated flooding affected the southern United States from 6-8 April 2014. This event occurred in a pattern known to be conducive for heavy rainfall which included a deep trough to the west and a ridge to the east. As the trough moved eastward the strong flow between the trough and ridge produced a surge of deep moisture and strong low-level flow. The result was a widespread heavy rainfall event. The pattern was relatively well predicted by the NCEP GEFS (Fig. 6-8). Correctly predicting the synoptic pattern, the GEFS and GFS correctly predicted the potential for a heavy rainfall event in the southern United States with a 4-6 days lead-time. The exact areas to received heavy rainfall varied from run-to-run (Fig. 9-11). At shorter ranges, the SREF showed a similar focus for heavy rainfall in the southern United States. The comparative GFS-SREF-GEFS forecasts indicate that the GEFS often follows the general pattern of the GFS. This is not surprising a version of the GFS run at 55km is the core model of the NCEP GEFS. The primary differences between the GFS (27km) and the GEFS (55km) are model resolution (horizontal and vertical resolution) and varied initial conditions. The finer scale GFS should theoretical produce higher QPF amounts than the coarser GEFS members.
The SREF (16km), with a mix of model cores, model physics, and varied initial conditions is a more diverse system than the GEFS (55km). The SREF is actually run at a finer resolution, 16km than the 27 km GFS. The mean QPF from the SREF was not shown here. The focus was on the probabilistic data. Strength of the SREF in this event was showing a wider region and slightly different region which might experience heavy rainfall relative to the GFS-GEFS family of forecasts. All three systems correctly predicted heavy rain in the southern United States where heavy rain was observed. This case appears to imply that this event had a relatively long predictability horizon in the GFS/GEFS and SREF. These data also show how closely, when forecasts are converging, the GEFS follows the general GFS forecasts. The lack of varied model-core and model physics may be a general weakness in the GEFS relative to the SREF. Clearly, a super-ensemble of these 4 systems would be of value in the forecast process. Though not shown, the CME-Ensemble forecast system performed similarly and may imply that this was a relatively predictable event. The data here showed 48 hour QPF accumulations. Shorter range windows showed more focused areas. But these data also showed timing issues which at times lowered the probabilities. The larger window was selected to show the overall larger scale success. A more interesting time window was the 0000 UTC 7-8 April window (Fig 13) which was the period of more intense rainfall. There is data to ferret out of these systems, the internal model climate (M-Climate) for each system would be of considerable value to know when each systems is predicting a record or near record 12,24 and in this case 48 your QPF event. These data would be of value in assessing HIWE when the pattern also supports heavy rainfall. 5. Acknowledgements Jun Du (NCEP/EMC) for data exchange and software iterations to present and improve displays of EFS data. The Pennsylvania State University for data access in real-time. 6. References Bodner, M. J., N. W. Junker, R. H. Grumm, and R. S. Schumacher, 2011: Comparison of atmospheric circulation patterns during the 2008 and 1993 historic Midwest floods. Natl. Wea. Dig., 35, 103-119. Doswell,C.A.,III, H.E Brooks and R.A. Maddox, 1996: Flash flood forecasting: ingredients based approach. Wea. Forecasting, 11, 560-581.
Doty, B.E. and J.L. Kinter III, 1995: Geophysical Data Analysis and Visualization using GrADS. Visualization Techniques in Space and Atmospheric Sciences, eds. E.P. Szuszczewicz and J.H. Bredekamp, NASA, Washington, D.C., 209-219. Durkee, JD., L Campbell, K Berry, D Jordan, G Goodrich, R Mahmood, S Foster, 2012: A Synoptic Perspective of the Record 1-2 May 2010 Mid-South Heavy Precipitation Event. Bull. Amer. Meteor. Soc., 93, 611 620. doi: http://dx.doi.org/10.1175/bams-d-11-00076.1 Junker, N. W., R. S. Schneider and S. L. Fauver, 1999: Study of heavy rainfall events during the Great Midwest Flood of 1993. Wea. Forecasting, 14, 701-712. Lorenz, E. N., 1963: Deterministic nonperiodic flow. J. Atmos. Sci., 20, 130 141. Lorenz, E.N, 1969: The predictability of a flow that possesses many scales of motion. Tellus, 21, 289 307. Lorenz,E.N, 1993: The Essence of Chaos. University of Washington Press, 227pp. Palmer, T. N., 2005: Quantum Reality, Complex Numbers, and the Meteorological Butterfly Effect. Bull. Amer. Meteor. Soc., 86, 519 530.doi: http://dx.doi.org/10.1175/bams-86-4-519 Stuart, N., Grumm, R. (2009) The Use of Ensemble and Anomaly Data to Anticipate Extreme Flood Events in the Northeastern United States. Nat. Wea. Digest 33: 185-202. Toth, Z, Y.Zhu,and T Marchok, 2001: On the ability of ensembles to distinguish between forecasts with small and large uncertainty. Weather and Forecasting, 16,436-477
Figure 1. Stage-IV rain estimated for the 24 hour periods ending at a) 1200 UTC 7 April and b) 1200 UTC 8 April 2014. Return to text.
Figure 2. GFS 00-hour forecasts of 500 hpa heights (m) and height anomalies in 24 hour increments from a) 1200 UTC 6 April through f) 1200 UTC 11 April 2014. Return to text.
Figure3. As in Figure 2 except for precipitable water (mm) and precipitable water anomalies in 6-hour increments from a) 1800 UTC 06 April through f) 0000 UTC 8 April, the time of the heaviest rainfall. Return to text.
Figure 4. As in Figure 3 except for 850 hpa winds (kts) and v-wind anomalies. Return to text.
Figure 5. As in Figure 4 except for 850 hpa moisture flux and moisture flux anomalies. Return to text.
Figure 6. GEFS ensemble mean forecasts of 500 hpa heights (m) and height anomalies valid at 1200 UTC 7 April 2014 from GEFS initialized at a) 0000 UTC 2 April, b) 0000 UTC 3 April, c) 0000 UTC 4 April, d) 0000 UTC 5 April, e) 1200 UTC 5 April, and f) 00000 UTC 6 April 2014. Return to text.
Figure 7. As in Figure 6 except for precipitable water (mm) zoomed over the southern United States.. Return to text.
Figure 8. As in Figure 6 except for 850 hpa winds and v-wind anomalies zoomed over the southern United States.. Return to text.
Figure 9. GFS QPF (mm) and GEFS probability of 75mm or more QPF valid for the 48 hour period ending at 1200 UTC 8 April 2014 from a) GFS initialized at 0000 UTC 3 April, b) GEFS initialized at 0000 UTC 3 April, c) GEFS initialized at 0600 UTC 3 April, d) GFS initialized at 0000 UTC 4 April, e) GEFS initialized at 0000 UTC 4 April, f) GEFS initialized at 0600 UTC 3 April Return to text.
Figure 10. As in Figure 9 except for GFS QPF and SREF and GEFS probability of 50 mm or more QPF valid for the 48 hour period ending at 1200 UTC 8 April 2014 from a) GFS initialized at 0000 UTC 5 April, b)sref initialized at 0300 UTC 5 April, c) GEFS initialized at 0600 UTC 5 April, d) GFS initialized at 0000 UTC 6 April, e) SREF initialized at 0000 UTC 6 April, f) GEFS initialized at 0000 UTC 3 April Return to text.
Figure 11. As in Figure 10 except probabilities are for 100 mm or more QPF. Return to text.
Figure 12. As in Figure 11 except for GEFS at 1200 and SREF at 0900 UTC 5 and 6 April 2014. Return to text.
Figure 13. As in Figure 1 except for a) event total to match the QPFs shown in the forecast section and b) the period of heaviest rainfall from 0000 7-8 April 2014. Return to text.