1. Introduction The enduring Louisiana rain and flooding of August 2016 by Richard H. Grumm National Weather Service State College, PA 16803 Historic flooding impacted Louisiana on 12-15 August 2016 (TWC 2016, USA Today 2016) causing over a billion dollars in estimated damages. This event was the ninth billion-dollar weather related disaster in the United States of 2016. At least 13 deaths were attributed to the flooding (AP 2016b). The maximum reported rainfall was 31.39 inches at Watson, Louisiana. The range of extreme rainfall at observing sites in Florida and Louisiana ranged from 14.43 inches in Panama City, Florida to 27.47 in Brownfields Louisiana (NOAA/NCEP/WPC). Rivers and streams in Louisiana crested above flood stage with 11 gauges setting new record crests. During the event several locations had exceeded all-time record stages and forecasts indicated flood stages that had not been observed before on Comite and Amite rivers. Most of the record crests were archived on 13 and 14 th of August. As observed in 6-hour increments, the heavy rain moved over Louisiana around 0000 UTC 12 August 2016 (Fig. 1a). Intense rainbands were evident over the next 30 hours indicating flow about a cyclonic circulation (Fig. 1a-f). The maximum 6-hour rainfall in the Stage-IV rainfall data was over south-central Louisiana where 6-hour rainfall rates were over 300 mm (12 inches: Fig 2). Higher rates were indicated offshore. The maximum 12 hour rates were slightly higher than 400 mm in the same region. The 6-hour rates imply some areas had 2 inch an hour rates for prolonged periods. Many extreme rainfall events are often associated with extremely high rainfall rates (Brooks and Stensrud 2000). They showed that for the 1978 Johnstown flood that rain fell over about a 9 hour period with 3 hours of rainfall rates on the order of 1.5 to 2.0 inches per hour (see Fig. 2: Brooks and Stensrud 2000). Using hourly rates of 1 to 1.5 inches per hour they showed that extreme rainfall rates peak in the summer reaching a maximum in July, along the Gulf Coast from eastern Texas to western Florida. Their indicated that rainfall rates above 4 inches are relatively rare occurring about once per year (see their Fig. 4) and the climatological peak month is July. The maximum 24 hour rainfall (Fig. 3) and total rainfall for the event (Fig. 3) show an axis of heavy rainfall from the Gulf Coast northwestward across southern Louisiana. A broad region of the southern and eastern Louisiana had over 200 mm of rainfall. A few areas had 4-day totals in excess of 600 mm (24 inches) of rainfall. The 4-day total implies something about the enduring nature of the rainfall while the maximum 6-,12-, and 24-hour data show the periods of intense rainfall. A slow moving surface low moved over the region and was provided as an explanation for the rainfall. However, at 850 hpa a more intense cyclone was evident (Fig. 4). The 850 hpa cyclone had strong flow about the cyclone with -2 to -3s u-wind anomalies on the north side of the cyclone and +3 to +4s u-wind anomalies on the south side of the cyclone as it lumbered across Louisiana. Most of the rain fell south and west of the cyclone track.
This paper will present the pattern and GEFS quantitative precipitation forecasts associated with the historic and devastating Louisiana Floods of 12-14 August 2016. This paper will also examine the forecast issues in the NCEP models related to predict both an intense and enduring rainfall event. 2. Methods and data The climate forecast system re-analysis (CFSR) data was used to reconstruct the pattern and the standardized anomalies associated with the event. The CFSR is used to show the pattern, which forecasters often use to gain confidence in a potential significant weather event. These same patterns, when forecast may produce high end QPF which may reinforce confidence in the forecast. The Stage-IV rainfall data (Seo 1998) was used to estimate the rainfall over several 6-,12-,24- hour periods from 10 to 15 August 2016. Data from 10 to 14 August were primarily used in the document. Unless otherwise indicated the 6-hour Stage-IV data were used. Analysis was conducted on 1-hour data over the same 6 days. Python was used to extract maximum rainfall rates and amounts over 1-,6-,12-, and 24-hour periods. The software retained the highest observed amount of for the specified duration at each grid point. Data was produced locally using archived GRIB files from the CFSRV2, GFS, GEFS, and HRRR. The focus here was on the GEFS ability to predict an intense and enduring rainfall event. The Average Recurrence interval images for the GFS were produced using NOAA14 ARI data in GRIB format. 3. Results a. The pattern The large scale pattern at 500 hpa (Fig. 5) showed large subtropical ridge over the western Atlantic and eastern United States. Most of the region had 500 hpa heights in the +1 to +2σ range. There was a weak trough over the intermountain West. In these 12 hour data, a weak negative 500 hpa height anomaly was present over Louisiana at 0000 UTC 13-14 August 2016. The weak 500 hpa disturbance was associated with a strong 850 hpa circulation (Fig. 4) and a blob of deep moisture as indicated by the precipitable water (PW) fields (Fig. 6 & 7) which showed very high PW values and PW anomalies over +3σ above normal. A frontal boundary with lower PW values and below normal PW values over Oklahoma eventually interacted wth the PW blob on 13 August (Fig. 7d-f). The high PW air, with values over 60 mm on the south side of the 850 hpa low were relatively well aligned with the heavy rainfall areas (Fig.2). The surface pressure field (Fig 8) showed a weak cyclone with pressure anomalies on the order of -1 to -2σ as it moved across Louisiana on 12-13 August 2016. The 850 hpa circulation and the strong winds about this circulation likely played a critical role in the rainfall (Figs. 4 & 9). The
850 hpa v-winds anomalies showed a stronger signal than the 850 hpa u-wind anomalies. Clearly, there was a strong circulation at 850 hpa. b. Rainfall Plots of the rainfall were shown in the Introduction showing the 6-hour rainfall from 0000 UTC 12 to 0600 UTC 13 August 2016 (Fig. 1). The heaviest 6-hour rainfall rates were observed on the 13 th as shown in Figure 10. The rainfall and rainfall rates dropped of significantly after 0600 UTC 14 August 2016. Though not shown the 6-hour and 12-hour (Fig. 11) maximum rainfall amounts in southern Louisiana were observed on the 13 th. The maximum 24 hour period was dominated by the 24- hour period ending at 1800 UTC 13 August 2016 (Fig. 11). The data in Figure 11 show extremely high rainfall for the 12- and 24-hour periods ending at 1800 UTC 13 August values which were not present at 1200 UTC. These data can be gleaned from the 6-hour plots of the QPE ending at each time period. c. GEFS Forecasts Planview images of GEFS QPFs were examined to see how the GEFS forecast the event. Six GEFS forecast cycles valid for the 48 hour period ending at 1200 UTC 13 August and the 72 hour period ending at 1200 UTC 14 August are presented. The 48 hour mean QPF (Fig. 12) with any members 200 mm contour indicated that the GEFS was able to predict a widespread heavy rainfall event in the central Gulf. Longer range forecasts (Fig. 12f-d) had the heavy rainfall and the axis of heavy rainfall too far east. Only shorter range forecasts were able to produced 200 mm of QPF. Thus, the probability of 100 mm or more QPF from these same GEFS cycles are shown in Figure 13. The 72 hour mean QPF for these 6 cycles and the 150 mm contour from each member is shown in Figure 14. As in the 48-hour accumulation window, earlier forecasts of heavy rainfall were too far east. But as the forecast length decreased the mean QPF came up, shifted over Louisiana and there were several member that forecast in excess of 150 mm of QPF. The GEFS accumulated QPF is shown for Baton Rouge for two GEFS forecast cycles (Fig. 15). The GEFS initialized at 0000 UTC 12 August showed 75 to over 400 mm of QPF. The mean QPF was closer to 200 mm. The GEFS was able to forecast heavy rain and an enduring heavy rainfall event at Baton Rouge. The 1200 UTC 11 August GEFS had a similar forecast with a larger spread and lower mean. Spatial and temporal issues were evident in longer range forecasts. The forecast from all GEFS runs from 6 to 142 hours out showing the QPF for the 24 hour period near Baton Rouge are shown in Figure 16. These data show longer range forecasts had a light rain event. The GEFS did not converge on the 24 hour period of heavy rainfall until about 36 to 48 hours before the event and like the plumes, showed a large spread. The lower QPF at 6, 12, 18 hours before the event reflect that shorter range forecasts did not extend long enough and are artificially lower due to spin-up and shorter accumulation periods. The main message is there were clues for heavy rainfall with about 2 days lead time using point data.
d. Return periods Not accomplished yet 4. Conclusions A slow moving tropical wave and a plume of deep moisture brought heavy rain and historic flooding to Louisiana on 12-14 August 2016. At 850 hpa a closed cyclone with strong winds around the cyclone lumbered across the State. The 850 hpa u- and v-wind anomalies were in excess of -3 and +4σ respectively with this system. The strong 850 hpa cyclone and the above normal precipitable water likely contributed to the conditions and moisture to produce an enduring and intense rainfall event. Despite some spatial and temporal issues, the NCEP GEFS forecast some incredible rainfall amounts in the Gulf States with at least 5 days of lead time and got the general potential for extreme rainfall for locations in Louisiana with about 2-days leadtime. In addition to the forecast amounts, the GEFS forecast an intense and enduring rainfall event. The 6-hour Stage-IV data were used here to examine the extreme rainfall. The 6-,12-, and 24- hour maximum rainfall rates were impressive. May locations saw several periods of in excess of 100 mm of rainfall and 24-hour rates exceeded 300 mm. The 1-hour data could be used to find the periods of maximum 1-hour rates and the times of these extreme rates. They key finding examining the 6-,12-, and 24-hour rainfall rates was the intensity and the enduring nature of the event. The GEFS correctly forecast a heavy rainfall event in the central Gulf States. Longer range forecasts had difficulty with the location of the heavy rain with the axis initially located too far east. The axis slowly shifted over Louisiana and the GEFS attempted to show the potential for 100 to 200 mm of QPF over portions of Louisiana. It under forecast the extreme values and it could not get the location correct. However, the model provided excellent signals for a potential high end rainfall event and an enduring rainfall event. The GEFS point data showed that the GEFS was able to show the potential for extremely heavy rainfall for a point such as Baton Rouge. But due to temporal and spatial errors the signal was useful with about 1.5 to 2 days of lead time. This was best illustrated in Figure 16. The plan view data provided additional lead time to the potential for an extreme rainfall event in the general region but these data were unable to outline the exact regions. The rainfall forecast by the GEFS was quite large, with member forecasting over 200 mm of QPF. Good knowledge of the GEFS QPF climatology would go a long way to put these data into a more useful context. Clearly, when the model forecasts a record or near record event they can and often do occur in the real atmosphere. 5. Acknowledgements The Pennsylvania State University for real-time data access. 6. References 7.
The Weather Channel: 2016: Historic 2016 Flooding in Louisiana after 2+ Feet of Rain Sends Rivers to Record Levels. TWC USA Today 2016: How much rain fell in Louisiana? USA Today. Associated Press 2016: Louisiana's Historic August Flooding Likely the State's Second Billion- Dollar Disaster of 2016. And similar stories AP 2016. Associated Press 2016b: Louisiana flood death toll rises to 13. And similar stories AP 2016. Benjamin, S.G, and contributors: 2016: A North American Hourly Assimilation and Model Forecast Cycle: The Rapid Refresh. MWR, 144, 1669-1693. Brooks, HE and D.J. Stensrud: 200: Climatology of Heavy Rain Events in the United States from Hourly Precipitation Observations. MWR, 128, 1194-1201. Ebert, E. E., 2001: Ability of a poor man s ensemble to predict the probability and distribution of precipitation. Mon. Wea. Rev., 129, 2461 2480. Juanzhen Sun, Ming Xue, James W. Wilson, Isztar Zawadzki, Sue P. Ballard, Jeanette Onvlee- Hooimeyer, Paul Joe, Dale M. Barker, Ping-Wah Li, Brian Golding, Mei Xu, and James Pinto, 2014: Use of NWP for Nowcasting Convective Precipitation: Recent Progress and Challenges. Bull. Amer. Meteor. Soc., 95, 409 426. Maddox,R. A,, C. F. Chappell, and L. R. Hoxit, 1979: Synoptic and Meso-α Scale Aspects of Flash Flood Events, Bull Amer. Meteor. Soc., 60, 115 123, DOI: http://dx.doi.org/10.1175/1520-0477-60.2.115 Doswell, C.A,, H. E. Brooks, and R.A. Maddox, 1996: Flash Flood Forecasting: An Ingredients-Based Methodology, Wea. Forecasting, 11, 560 581:DOI: http://dx.doi.org/10.1175/1520-0434(1996)011<0560:fffaib>2.0.co;2 WP 2016: Two dead after severe flash flood in Maryland. Washington Post 31 July 2016. WP 2016: This is how and off-the-charts flood ravaged Ellicott City. Washington Post 1 August 2016. Weaver, S. C., and S. Nigam, 2008: Variability of the Great Plains low level jet: Large scale circulation context and hydroclimate impacts. J. Climate,21,1532 1551.
Figure 1. Stage-IV estimated precipitation (QPE) in 6-hour increments from a) 0000 UTC 12 August 2016 through f) 0600 UTC 13 August 2016. Return to text.
Figure 2. Maximum observed rainfall in the Stage-IV data during the event showing the maximum 6- hour and maximum 12-hour rainfall. Shading as in color bar. Contours every 100 mm beginning at 200 mm. Return to text.
Figure 3. As in Figure 2 except for the maximum 24-hour rainfall in any 24 hour period and the total rainfall. Return to text.
Figure 4. CFSR 850 hpa winds and u-wind anomalies in 12 hour increments from a) 1200 UTC 11 through f) 0000 UTC 14 August 2016. Winds in ms-1. Anomalies are in the color bar. Return to text.
Figure 5. As in Figure 4 except for 500 hpa heights and anomalies every 12 hours from a) 1200 UTC 11 to f) 0000 UTC 14 August 2016. Return to text.
Figure 6. As in Figure 4 except for precipitable water (mm) and precipitable water anomalies from a) 0600 UTC 12 to f) 1200 UTC 12 August 2016. Return to text.
Figure 7. As in Figure 6 except for the period of a) 1800 UTC 12 August through f) 0000 UTC 14 August 2016. Return to text.
Figure 8. As in Figure 7 except for mean sea level pressure every 6 hours from a) 1200 UTC 12 to f) 1800 UTC 13 August 2016. Return to text.
Figure 9. As in Figure 4 except for 850 hpa v-wind anomalies. Return to text.
Figure 10. As in Figure 1 except for 6-hour rainfall from 0000 UTC 13 August through 0600 UTC 14 August 2016. Return to text.
igure 11. Accumulated maximum 12-hour (upper) and 24-hour (lower) QPF at each grid point for the periods ending at 1200 and 1800 UTC 13 August 2016. eturn to text.
Figure 12. NCEP GEFS forecasts of QPF for the 48 hour period ending at 1200 UTC 13 August 2016. Spaghetti plots show each member 200 mm contour of present. GEFS cycles are initialized at a) 1200 UTC 10 August, b) 0000 UTC 10 August, c) 1200 UTC 9 August, d) 0000 UTC 9 August, e) 1200 UTC 08 August, and f) 0000 UTC 8 August 2016. Return to text.
Figure 13. As in Figure 12 except for the probability of 100 mm or more QPF and the 100 mm mean contour if present. Return to text.
Figure 14. As in Figure 12 except for the 72 hour period ending at 1200 UTC 14 August 2016. Return to text.
Figure 15. Plume diagrams of NCEP GEFS QPF for a point near Baton Rouge, LA showing accumulated QPF (green) and instantaneous QPF (gray with dots) from each of the 21 GEFS members. Values in mm. Return to text.
Figure 16. Predictability horizon diagram showing 24-hour QPF for a point near Baton Rouge valid at 1800 UTC 13 August 2016. The horizontal axis shows the forecast length of each GEFS forecast with each point representing a QPF from a GEFS member. Y-axis shows the scale of the QPF in mm. Return to text.