Heavy rains and precipitable water anomalies 17-19 August 2010 By Richard H. Grumm And Jason Krekeler National Weather Service State College, PA 16803 1. INTRODUCTION Heavy rain fell over the Gulf States, into the Tennessee Valley and in the Mid-Atlantic region from 17-19 August 2010 (Fig. 1). All three areas of heavy rainfall were associated with above normal precipitable water (PW). The rainfall over the Gulf and Ohio Valley was associated with a rejuvenated circulation of former Tropical Depression #5. The circulation was clearly evident as the system moved onshore late 16-7 August 2010 (Fig. 2). The heavy rainfall with the remnants of TD#5 fell mainly west of the track of the 850 hpa low but east of the surface cyclone over eastern Louisiana and western Mississippi (Fig. 1). Higher resolution data revealed a cyclonic pattern to the rainfall. The surge of high PW produced the second largest all-time PW (2.88 inches) value at Slidell, Louisiana on 1200 UTC 17 August 2010 and Jackson, MS set a new record high PW value (2.76 in) at 0000 UTC 18 August 2010. The old record was 2.57 inches set on 12 August 1987. Interesting all 10 of the record high PW values in Jackson were set in August (6) and July (4). The rainfall in the Tennessee Valley was also associated with the remnants of TD#5 as it interacted with a frontal system. The same general frontal system, with high PW air in it also produced the heavy rainfall in the Mid-Atlantic region. This paper will document the rainfall and the associated high PW which produced it. The goal here is to demonstrate the value of anomalies in forecasting heavy rainfall events. 2. METHOD The pattern was reconstructed used the NCEP GFS and NAM and were possible the JMA 1.25x1.25 data (Onogi et al. 2007). All data were plotted in GrADS (Doty and Kinter 1995). The severe weather data was overlaid on the JRA data. The higher resolution NCEP NAM is used to show the conditions during the event. The anomalies were computed from the NCEP/NCAR re-analysis data (Kalnay et al 1996) as describe by Hart and Grumm 2001 and Grumm and Hart 2001. Unless otherwise stated, the base data was the NAM and the means and standard deviations were computed by comparing the NAM to the NCEP/NCAR 30-year climatological values. For brevity times are referred to in the format of 18/1800 for 18 August 2010 1800 UTC. 3. RESULTS i. Large scale pattern Figure 3 shows the 250 hpa winds (kts) and wind anomalies over the United States from 17/0000 through 19/1200 UTC. A strong jet stream was present over the Mid-West and Great Lakes early in the event with weak flow over the Gulf States. A weak trough over the United States and a strong
Figure 1. Total precipitation from Stage-IV data from 0000 UTC 17 to 1200 UTC 19 August 2010. Data begin at 8 m as indicated by the color scale on the right. subtropical ridge was present over the southeastern United States. The strong jet was in the tight gradient associated with this feature (Fig. 4). ii. Regional Patterns-South Figure 6 shows the mean-sea level pressure over the southern United States from 16/1800 through 18/1800 UTC. The circulation of TD#5 is clearly visible as is the attendant pressure anomaly as the system drifted westward across the Gulf and into eastern Texas before re-curving to the northeast. These data implied a weak surface cyclone. Despite the weak cyclone, the system was associated with PW values in the 60 to 70 mm range (Fig. 7) which produced +3 to +5SD PW anomalies over the region. PW values over 65 mm were associated with the +4 to +5SD PW anomalies. This lead to
Figure 2. GFS 00-hour forecasts of 850 hpa winds (kts) and total wind anomalies (standard deviations) valid at a) 1800 UTC 16 August 2010, b) 0000 UTC 17 August, c) 0600 UTC 17 August, d) 1200 UTC 17 August, e) 1800 UTC 17 August, f) 0000 UTC 18 August, g) 0600 UTC 18 August, h) 1200 UTC 18 August, i) 1800 UTC 18 August 2010. several Rawinsonde sites setting all-time or second all-time highest PW values. On a broader scale, the PW field in Figure 7 showed a sharp frontal boundary from Arkansas eastward across the Tennessee Valley which set up the second area of heavy rainfall as the high PW air eventually surged northward into the Tennessee Valley (Fig. 7f-i). Figure 2 showed the 850 hpa winds and wind anomalies. These data combined with the PW data show that the initial high PW air was associated with the cyclone. The circulation intensified and the southerly winds increased (Fig 2d-i) and surged northward. The winds peaked at +5SDs above normal with mainly southerly. This flow then interacted with the frontal boundary producing the heavy rainfall in the Tennessee Valley (Fig. 1).
iii. Regional Pattern-East Figure 8 shows the PW and anomalies over the eastern United States. The surge of high PW air into the Tennessee Valley is evident, peaking near 60mm and 3SDs above normal at 18/1200 UTC (Fig. 8d). At about this time the PW value over Maryland also peaked at 60 mm. The high PW air with 55 to 60 mm values remained over Tennessee from 18/1200 UTC through 19/0600 UTC. Drier air had moved over most of Kentucky after 19/0000 UTC. The 850 hpa winds and anomalies are shown in Figure 9. Clearly, the 3-4SD wind anomalies stayed just south of the heavy rainfall area in Tennessee but the PW plume did penetrate the region. With the exception of weak easterly wind anomalies, the winds were not exceptional over the Mid-Atlantic region. The 850 hpa moisture flux (MFLUX) and MFLUX anomalies (Fig. 11) showed some large anomalies over Mississippi (Fig. 11a-d) which moved over Alabama and into the Tennessee Valley. The period of high MFLUX over the Mid- Atlantic was around 18/1200 UTC. iv. Rainfall The overall rainfall pattern over the eastern United States was shown in Figure 1. These data showed 4 areas of heavy rainfall. The discussion here is limited to that rainfall associated with the remnants of TD#5 over the Gulf Coast and the Tennessee Valley and in the Mid-Atlantic region. The discussion is based on 6-hourly rainfall, the images in Figures 2-11 and radar images (not shown). The 6-hourly total rainfall at select times is shown in Figure 12 along with regional totals in Figure 13. In the Gulf States, the heavy rains began on the morning of the around 17/1200 UTC, primarily in the lower Mississippi River Valley. The remnants of TD#5 moved over the LA/MS border and moved slowly through the region (Fig. 4). This system and its attendant near record high PW values produced heavy rain over a 48 period. Between 25 to 50 mm of rain fell on the 17 th along the central LA/MS border. Overnight the intensity picked up and a large area of 25 to 50 mm fell with an area of 150 + mm just east of the LA/MS border. During the day on the 18 th the circulation remained stationary dropping heavy rains in the same areas as the previous day. A widespread area of 25 to 50 mm fell along the Mississippi river with areas of 150 mm. In total a maximum of 500 mm fell just west of the LA/MS border with widespread 50 to 100 mm totals across LA/ southwestern MS. In the Tennessee Valley, mainly over Tennessee and Kentucky, isolated heavy rains began on the morning of after 17/0000 UTC across southern TN and moved into northern TN and southern KY by the evening. Overnight a large band of 25 mm rainfall amounts fell across central TN with areas of north central TN with 75 to 150 mm. After 18/0600 UTC the heaviest rains moved into southern KY and north Central TN. Widespread 25 to 50 mm rain fell with isolated 150+ mm in north central TN. The rain began to wind down on the after 19/0600 UTC but another 25 to 50 mm fell across north central TN and southern KY. Rainfall reports suggest widespread area of 75 to 150 mm of rainfall over the region with a few isolated 250 mm amounts. The eastern edge of the stationary boundary which contributed to the heavy rainfall in the Tennessee Valley also produced heavy rainfall in the Mid-Atlantic region from the
Figure 3. As in Figure 2 except for 250 hpa winds and wind anomalies. 17 th to the evening on the 18 th. Heavy rains fell along this boundary along a line from south-central VA through southern MD and DE across southern NJ. On the 17 th a few isolated cells developed along the boundary and remained nearly stationary during the afternoon hours. Overnight a short wave through moved northeast along the boundary producing a steady heavy rain event. Total rainfall amounts of 25 to 75 mm fell across extreme northern NC and south central VA. The Washington, DC Metro area received 25 to 75 mm of rainfall and southern central DE received 50 to 125 mm of rainfall. v. Forecasts Three SREF QPF and probabilities are shown in Figure 14. Though not shown, the SREF and other NCEP systems predicted the surge of high PW air and the boundary shown in Figures 7 & 8. Longer ranges forecasts did not handle the boundary so well pushing it too far to the south.
Figure 4. As in Figure 2 except for 50 0 hpa heights and height anomalies. The QPF data suggests that the SREF did reasonably well with the QPF and high QPF threat over the Mid-Atlantic region. However, other than the pattern the SREF under predicted the rainfall and rainfall potential over the Gulf States and the Tennessee Valley. As shown in Figure 15, the GEFS faired little better. Similar to the SREF it got the pattern right after 15 August and seemed to have a signal over the Mid-Atlantic region. The higher resolution GFS forecasts from 9 forecast runs are shown in Figure 16 over the south-central United States. These data indicate that several GFS runs showed potential for heavy rainfall in the Gulf States. There was considerable run-to-run variation between models forecast cycles. Shorter range forecasts appeared to show the potential for heavy rainfall over the Tennessee Valley too.
Figure 17 shows the accumulated GFS rainfall over the Mid-Atlantic region. These forecasts showed that the GFS was forecasting 50 to 100 mm at times over the Mid-Atlantic region. These values, on such a scale were high though some areas did see near 100 mm of rainfall. Not forecasts of PW or winds are shown. These precipitation data suggest that the GFS had high PW values and large anomalies and thus the data outcome of the high PW air, the large QPF is a sufficient proxy. 4. Discussion/Conclusions The remnants of TD#5 drifted into Louisiana and Mississippi producing heavy rains from 17-19 August 2010. As this system and it s moisture plume interacted with a mid-latitude trough, the moisture surged into the Tennessee Valley producing heavy rainfall in that region. The same general boundary between warm moist tropical air and above normal PW air produced heavy rains in the Mid-Atlantic region. It should be stated that the Mid- Atlantic region received less rainfall than the other two locations. The pattern indicated high PW air over the southern Gulf States. Models and ensembles also predicated high values of PW and thus high PW anomalies. Rawinsonde data verified that indeed there was high PW air associated with this system. Record PW values were set for the date at several locations and many locations had a top 10 all-time PW value record in rawinsonde data to included: Slidell, LA reached a PW of 2.88 on the 17/1200 UTC which is the 2 nd highest all-time value. Jackson, MS reached a PW of 2.76 on the 18/0000 UTC which set a new alltime record value. Nashville, TN reached a PW of 2.56 on the 18/0000 UTC which is 3 rd all-time. Lake Charles, LA reached a PW of 2.81 on the 18/1200 UTC which is 2 nd alltime Surge in LLJ (Fig. 2) and PW (Fig. 5) may relate to the tropical system interacting with the jet entrance to the north. Clearly, the surge of high PW air and areas of large PW anomalies aligned with the regions of heavy rainfall. Despite a relatively well predicted pattern, the QPF was not so well predicted by the EFS s. As shown in Figures 14 & 15 there was a high probability of 2 inches of rainfall in the Mid-Atlantic region. The overall QPF pattern was well predicted. However, the GEFS and SREF under predicted the heavy rainfall over the southern Gulf States and the Tennessee Valley. These poor forecasts suggest sub-grid scale processes and thus the importance of pattern recognition. The high PW air and near record PW anomalies indicated the potential for extremely heavy rainfall. The models and EFS predicted this but did not produce the requisite QPF. Higher resolution deterministic models such as the NAM and GFS (Fig. 16-17) did show areas of heavy rainfall over Louisiana and the Tennessee Valley. There was considerable run-to-run variation with the heavy rainfall over both Louisiana and Mississippi. At times as much as 256 mm of QPF was predicted. Other forecasts were less robust. This suggests that there were convective parameterization and model uncertainty issues.
The GFS predicted 50 to 75mm of rainfall over the Tennessee Valley (Fig. 16). This is a significant amount of QPF in the correct region. Lower than many locally observed maximum. Similar to along the Gulf Coast, the QPF was low. The GFS did better, in terms of consistency with the rainfall over the Mid-Atlantic region. However, it appeared to over predict the high end amounts and coverage. Why did the GFS perform more consistently over the Mid-Atlantic region? The high PW values and strong winds in the Gulf and Tennessee Valley were clues to heavy rainfall. In the East the high PW as a good signal for heavy rain. In each area, the models suggested rainfall. However, lacking model climatology, it is hard to know if the models were forecast record or near record rainfall, this would be an important piece of additional forecast information. Though not shown, these models also predicted much above normal PW values and the potential for near record high values. Events such as this, with high PW values and strong winds are indicators of the potential for heavy rainfall. When convective processes may play a significant role, higher resolution ensembles may under predicted the heavy rainfall potential. Higher resolution models may offer better clues as to the potential for heavy rainfall. Optimally, an ensemble of high resolution regional ensembles may be the best forecast tool in situations such as these.
Figure 6. As in Figure 6 except for mean sea-level pressure and anomalies valid at a) 1800 UTC 16 August-i) 1800 UTC 18 August 2010.
Figure 7. As in Figure 6 except for PW and PW anomalies.
Figure 8. GFS PW and PW anomalies over the eastern United States valid at a) 1800 UTC 17 August 2010, b) 0000 UTC 18 August, c) 0600 UTC 18 August, d) 1200 UTC 18 August, e) 1800 UTC 18 August, f) 0000 UTC 19 August, g) 0600 UTC 19 August, h) 1200 UTC 19 August, i) 1800 UTC 19 August 2010.
Figure 10. As in Figure 9 except 850 hpa winds and wind anomalies.
Figure 11. As in Figure 9 except 850 hpa moisture flux and anomalies.
Figure 12. Stage-IV QPE (mm) for the 6 hour periods ending at a) 1800 UTC 17 August, b) 0000 18 August, d) 0600 UTC 18 August and d) 1200 UTC 18 August 2010.
Figure 13. Total estimated rainfall from the Stage-IV data valid for the period of 0000 UTC 17 through 1200 UTC 19 August 2010. Values (mm) as in color code. Images are regionalized. Figure 13-b offers an other anlaysis.
Figure 13-b. National Mosaic QPE focused over the southern US and Tennessee Valleys. Experimental rainfall products.
J Figure 14. SREF forecasts of 2.00 inches of QPF for 36 hour periods valid at the time indicated from forecasts initialized at a) 0900 UTC 15 August, b) 0900 UTC 16 August and c) 0900 17 August 2010. Upper panels show the probability of 2 inches in the 36 hour period and lower panels show the ensemble mean QPF and each member 2 inch contour.
Figure 15. As in Figure 14 except for GEFS forecasts of 2 inches of QPF in 24 hours ending at 0600 UTC 19 August 2010 from forecasts initialized at a) 0000 UTC 15 August, b) 0000 UTC 16 August and c) 0000 UTC 17 August.
Figure 16. GFS accumulated QPF from the start of the model cycle ending at 1200 UTC 19 August 2010 from GFS forecasts initialized at a) 0000 UTC 15 August, b) 1200 UTC 15 August, c) 0000 UTC 16 August, d) 0600 UTC 16 August, e) 1200 UTC 16 August, f) 1800 UTC 16 August 2010, g) 0000 UTC 17 August, h) 0600 UTC 17 August and i) 1200 UTC 17 August 2010. Values in mm as per color bar to right.
Figure 17. As in Figure 16 except over the Mid-Atlantic region.