SR/SSD 96-49
12-1-96

Technical Attachment

MODEL PERFORMANCE DURING A SOUTH TEXAS HEAVY RAIN EVENT

Andrew R. Patrick and Steven R. Pfaff
NWSO Corpus Christi, Texas

Introduction

During late October 1995, a non-tropical heavy rain event occurred across much of South Texas, especially the Texas Coastal Bend. While rain was reported during the period of October 28-30, the majority of the rain fell on October 29, when rainfall totals ranged from 2 in across the western counties of the Coastal Bend to over 12 in along the Gulf Coast counties.

Heavy rain events of this magnitude across South Texas are usually associated with tropical systems; however, during this event upper dynamics played a much larger role. Beside a synoptic overview of this episode, NCEP models will be reviewed to examine the performance of Quantitative Precipitation Forecasts (QPFs) for South Texas.

Brief Event Description

Rainfall at the Corpus Christi International Airport totaled 7.90 in on October 29. This amount not only broke the record for the date, but it was also the third highest daily rainfall ever recorded at the airport. Figure 1 shows rainfall totals from surrounding areas. This event was the highest rainfall total since 1887 not associated with a tropical storm or hurricane.

Widespread flooding was reported by local law enforcement officials and local emergency management personnel. Although the flooding led to numerous road closings and also stranded many vehicles, no severe damage or injuries were a direct result of flooding. Several buildings in the Port Aransas and Aransas Pass areas were flooded due to the excessive rainfall amounts during such a short duration. The bulk of the rain occurred within an 8-hr period on October 29, from approximately noon until 8:00 pm CST.

Synoptic and Sounding Analysis

The weather pattern for the October 29, 1995, episode had characteristics of both a frontal-type and synoptic-type heavy rain event as defined by Maddox et al. (1979). A quasi-stationary front extended across far South Texas between Corpus Christi and Brownsville. A diffuse 850-700 mb east-west oriented boundary was also evident. An easterly 850 mb jet was feeding moisture along the surface boundary as a 500 mb trough was approaching from the southwest United States. The trough allowed weak short wave impulses to transverse the area during the day.

Strong upward vertical motion was also noted during this analysis. Both the 1200 UTC Eta and NGM revealed good low-level isentropic lift. Also, an upper level jet streak was moving through far South Texas. This feature enhanced the upper vertical motion and upper divergence in its right rear quadrant as seen in Fig. 2. Grice and Maddox (1983) found this to be a favorable characteristic of South Texas heavy rain events in previous studies. The strong vertical motion was also indicated by direct ageostrophic circulation along the right rear quadrant of the upper jet.

The 1200 UTC October 29 modified sounding (Fig. 3) was characterized by a layer of deep moisture resulting in precipitable water amounts above 1.80 in. Surface based stability revealed a neutrally stable atmosphere with a Lifted Index (LI) of 1C, while instability was more evident in the mid-levels where the 850 mb based K Index value was 34. The wind profile indicated significant veering in the low to mid levels, and upper level winds were aligned perpendicular to the frontal boundary.

The 850 mb moisture axis extended along the Rio Grande Valley into South Central Texas. Maddox et al. found that the heaviest rain is skewed to the right of the moist axis. Figure 4 depicts the 1200 UTC (1979) October 29 850-700 mb layer-average moisture axis and subsequent rainfall across the Coastal Bend. In this case, the heaviest rainfall occurred east of this axis and was aligned similar to previous heavy rainfall cases.

Model Similarities

Several meteorological parameters were similar in the NGM and Eta models. The most obvious and consistent were the moisture fields. From 48 to 36 hr prior to the heavy rain, both models depicted deep moisture in the vicinity of the event. This is reflected in the 1200 UTC October 29 Corpus Christi sounding and in gridded relative humidity, precipitable water, and moisture-flux convergence charts. Both models were in agreement with respect to the location and the overall flow and, more importantly, the low/mid level and upper level jet streaks.

The models agreed on a north-to-south orientation of the 200 mb subtropical jet streak, with the right entrance region along the coast of South Texas. The east-southeast 850 mb wind maximum supplied ample moisture from the Gulf of Mexico. This feature was indicated by both models. There was also agreement on the primary 500 mb trough over the southwest United States which ejected a series of short waves and associated vorticity maxima over northeast Mexico and South Texas.

Model Discrepancies

Despite these similarities, large scale model discrepancies were noted mainly in the upper levels. Figures 5a and 5b show the 36-hr forecast (valid 0000 UTC October 30) for the 200 mb wind speed and divergence from the Eta and NGM, respectively. While both models indicated a jet streak approaching South Texas, the Eta indicated stronger winds and greater upper divergence. Successive model runs also showed similar trends. Figures 6a and 6b show the subsequent model run forecasts for 24-hr, also valid at 0000 UTC on October 29. Once again, the Eta had the strongest upper level divergence and corresponding upward motion.

Total quantitative precipitation fields from each model were also analyzed. Figures 7a and 7b show the 36-hr precipitation forecasts (12-hr totals) for both models valid at 0000 UTC October 30. While both forecasts significantly underestimated rainfall, the Eta performed better in forecasting significant rain over the Coastal Bend. The NGM placement of precipitation along the Rio Grande Valley and northern Mexico remained consistent on successive model runs. Figures 8a and 8b show the 24-hr forecasts of 12-hr QPFs for each model (at the same valid time as Fig. 7). QPFs were underestimated by both models. Although the NGM produced greater precipitation forecasts, this model continued to forecast the heaviest rain west of the Coastal Bend. Despite smaller amounts and a slightly southern shift of precipitation during this forecast cycle, the Eta once again showed the best placement of the precipitation area.

Conclusion

It is apparent that the Eta outperformed the NGM in this particular case. The ability of the Eta to identify the evolution of the heavy rain event well in advance greatly benefitted forecasters. The Eta model captured the magnitudes and representative placement of not only the moisture fields and jet streaks, but retained continuity of upper divergence, implied vertical velocities, and isentropic lift. Despite producing higher QPF values than the Eta in some of the model runs, the NGM was not consistent, especially regarding the location of QPF maxima.

The Eta's greater resolution, in particular the greater number of vertical layers in the model, has likely contributed to its superior performance. Future studies of non-tropical heavy rain events across South Texas will be needed to confirm the superiority of the Eta model in similar instances, as well as overall performance of NCEP models. A more thorough understanding of model tendencies and limitations with regard to heavy rainfall events will enable forecasters to produce higher quality QPFs.

References

Grice, G.K., and R.A. Maddox, 1983: Synoptic characteristics of heavy rainfall events in South Texas. Preprints, Fifth Conference on Hydrometeorology, Tulsa, Amer. Meteor. Soc., 149-155.

Maddox, R.A., C.F. Chappell, and L.R. Hoxit, 1979: Synoptic and meso- scale aspect of flash flood events. Bull. Amer. Soc., 60, 115-123.