Doug Green
WFO Phoenix, Arizona
1. Introduction
Significant flash flooding occurred over a localized portion of western Gila
County, Arizona, an area of relatively rugged terrain east of Phoenix (Figure
1), during the early morning hours of 9 September 2003. KIWA WSR-88D precipitation
estimates indicate that 8 to 10 inches of rain fell over a 5 sq mi area centered
4 miles south of Roosevelt, Arizona, in a 4-hour period (11 pm MST 8 September
to 3 am MST 9 September), while in excess of 6 inches of rain fell over a
100 sq mi area just south and east of Roosevelt Lake (Figure
2). According to a storm spotter in the area, in excess of 10 inches of
rain was measured at a couple of rain gauges just south of Roosevelt Lake.
Flash flooding occurred on several creeks and numerous washes in the primary
drainage basin, with the worst flood damage reported over a small portion
of western Gila County between 2 and 4 am MST; 5 homes were declared total
losses due to flooding, while over 200 others sustained significant flood
damage (East Valley Tribune newspaper article, 13 September 2003). Peak rainfall
totals from this single event exceeded yearly average rainfall observed over
Greater Phoenix, and were on the order of 50-60% of the average yearly total
over the affected area.
This TA-Lite provides a brief overview of the synoptic and mesoscale conditions which preceded and accompanied the flash flooding, and highlights the importance of situational awareness with regard to the issuance of timely, accurately-worded flash flood products.
2. Synoptic Overview
At 1200 UTC 8 September, a mid-latitude trough existed near 120 W, while ridges
existed near 150 W and 90 W. A vigorous short wave trough was moving through
the Pacific Northwest, with a secondary upstream disturbance observed over
the eastern Pacific Ocean. A surface cold front extended from northwest Utah
through central Nevada. By 0000 UTC 9 September, the vigorous Pacific Northwest
trough had tracked east, staying mainly north of Arizona; however, the base
of the trough extended south to southern California (Figure
3). The low-mid level pressure gradient had tightened, leading to a strengthening
and backing of the low-mid tropospheric winds over Arizona. This trend continued
overnight as the secondary disturbance strengthened and moved southeast to
the California coast.
The low level (sfc-700 hPa layer) air mass over Arizona possessed an above-average equivalent potential temperature: Phoenix's average temperature for 8 September, 94 F, was 6 degrees above climatology, while the average surface dew point of 63 F was 7 degrees above climatology. Precipitable water was 1.25"-1.5" over south central Arizona during much of the day, then increased to in excess of 1.5" during the night of 8-9 September as moist southwest flow increased (precipitable water was 130-140% of normal over south central Arizona during the night of 8-9 September). GOES precipitable water imagery indicated that a moist axis extended from northwest Mexico to north-central Arizona. The GFS indicated that deep layered (700-300 hPa) synoptic scale vertical motion would increase over south central Arizona during the night of September 8, peak during the day of September 9, then decrease the following evening. The 0300 UTC Phoenix sounding (Figure 4) indicated relatively large CAPE (> 1000 j/kg) and very small CIN (convective inhibition), similar to 0000 UTC soundings taken at Tucson and Flagstaff; however, precipitable water had increased over Phoenix between 0000 and 0300 UTC. The combination of increasingly favorable dynamics, presence of a ridge of equivalent potential temperature, and increasingly moist, unstable air supported the potential for isolated strong thunderstorm development, with locally heavy rain, over south-central Arizona during the night of 8-9 September, with more widespread development during the day of 9 September.
3. Evolution of the Flash Flood
Isolated thunderstorms formed to the north and west of Phoenix shortly after
5 pm September 8, then affected northwest Maricopa County between 6 and 730
pm. A few storms developed south and east of Phoenix around 745 pm, but most
activity had died out and skies were generally clear west of Phoenix by 930
pm. Between 950 pm and 1115 pm, cloud-to-ground lightning was sparse over
south-central Arizona; only 2 negative strikes were detected. Radar reflectivity
images confirmed that convective activity had diminished (Figures
5, 6 , 7, 8
depict conditions as of 0600 UTC 9 September). However, infrared satellite
images and surface observations indicated that low and mid-level cloudiness
increased over and east of the Phoenix area between 930 and 11 pm, suggesting
that synoptic scale upward vertical motion was occurring. Given the unstable
conditions, with little inhibition noted on the soundings, coupled with the
increase in cloudiness over south-central Arizona, forecasters at WFO Phoenix
expected thunderstorm activity to increase over and east of Phoenix; shortly
after 11 pm, areal coverage and intensity of convective showers began to increase,
mainly east of Phoenix over extreme eastern Maricopa County and western Gila
County. By 1145 pm, an intense thunderstorm had developed over the Salt River
Lakes Recreational Area in extreme eastern Maricopa County, which prompted
issuance of the first severe thunderstorm warning for this event.
Between midnight and 3 am MST, a quasi-stationary MCS developed and affected extreme eastern Maricopa County and western Gila County (Figures 9, 10 , 11, 12, 13, 14, 15, 16, 17, 18, 19, 20). Although individual cells moved toward the east-northeast at 20 mph, discontinuous cell redevelopment occurred on the southwest flank of the thunderstorm complex over extreme eastern Maricopa County as thunderstorm outflow interacted with ambient southwest flow, resulting in the nearly stationary storm complex. Training of echoes was primarily focused in a 5-mile wide band extending from 10 miles south-southwest of Roosevelt to 35 miles east-southeast of Roosevelt, where rainfall rates in excess of 3 inches per hour were observed, especially between midnight and 3 am MST. Three hour precipitation (THP) accumulation valid at 3 am September 9 indicated several areas in the heavy rain band had received 7 to 8 inches of rain (Figure 21), with much of the aforementioned area receiving in excess of 3 inches of rain. This rain fell over a basin drained by several creeks and washes, including Campaign, Spring and Pinto Creeks, as well as Blevens Wash. Flash flooding occurred in Roosevelt Estates, a residential area on the southeast side of Lake Roosevelt, between 2 and 3 am; however, the most significant flash flood damage occurred over and near State Route 88, where flooded Pinto and Spring Creeks initially merged.
4. Discussion and Conclusions
Timely, accurate flash flood products were issued by WFO Phoenix before and
during this catastrophic event. Situational awareness, as usual, was key to
successful short-term forecast operations. Forecasters were well-aware that
flash flood conditions would be in place during the day on 9 September (a
flash flood watch was already in effect), and noted the following factors
which supported the potential for strong thunderstorms with locally heavy
rainfall over south central Arizona the night of 8-9 September: 1) the 1800
UTC 8 September GFS model run forecasted deep-layered (700-300 hPa) upward
vertical motion in excess of 4 microbar/sec to develop and be centered over
the affected area by 0600 UTC 9 September, with strongest low-level (700 hPa)
ascent centered over the Greater Phoenix area, which suggested that convective
development was likely over and just east of Phoenix, while more persistent
deep convection would be possible east of Phoenix (the 0000 UTC 9 September
GFS run generated a similar vertical motion forecast); 2) the GFS forecast
a 35 knot jet streak at the 500 hPa level over central Arizona by 0600 UTC,
with attendant differential positive vorticity advection; 3) water vapor satellite
imagery indicated increasingly diffluent upper level flow over south-central
Arizona during the afternoon and evening of 8 September; 4) infrared satellite
imagery and surface observations indicated low (7500-8500 foot MSL cloud bases)
and middle-level (15000 foot MSL cloud bases) developed rapidly over south
central Arizona between 0430 and 0600 UTC 9 September; 5) GOES precipitable
water images indicated that a moisture ridge existed over east-central Arizona,
with precipitable water increasing to 130-140% of normal over south central
Arizona during the night of 8-9 September; 6) west-southwest surface winds
10-15 mph were noted over south central Arizona through at least 11 pm on
8 September (stronger wind, with more upslope component than usually observed);
7) Phoenix Sky Harbor surface pressure bottomed out at 1003.6 mb at 0200 UTC
9 September, about 1 millibar lower than the climatological norm; 8) Phoenix
maximum temperature and average dew point were above climatological norms;
and 9) the 0000 UTC Phoenix sounding exhibited above normal CAPE and precipitable
water, with below normal convective inhibition.
The primary lesson
reinforced from this event was: when atmospheric instability is greater than
normal, convective inhibition is negligible, precipitable water is greater
than climatology (especially in the lower levels of the atmosphere), and low-level
southwest (upslope) flow is present, it takes only weak to modest kinematic
forcing to set the stage for a significant convective event. In this case,
a quasi-stationary MCS produced extremely heavy rainfall, with rainfall rates
and 3-hour rainfall totals seldom observed anywhere over the Phoenix CWFA.
TA-Lite Figure Captions for 9 September 2003 Flash Flood
Figure 1. High-resolution topographical map. "Home" identifies the
area that experienced the most significant flash flood damage.
Figure 2. Storm Total Precipitation (STP) from the KIWA WSR-88D, valid 1200 UTC 9 September.
Figure 3. Water vapor image, with boundary layer CAPE, metar observations, 1 hour lightning and 700-300 hPa layer div Q, valid at 0000 UTC 9 September.
Figure 4. Phoenix Arizona upper air sounding, valid at 0300 UTC 9 September.
Figure 5. Same as figure 3, except at 0600 UTC 9 September.
Figure 6. Infrared image, with 1 hour lightning and metar observations, valid at 0600 UTC 9 September.
Figure 7. 3.4 degree base reflectivity, 5 minute lightning, and metar observations, valid at 0600 UTC 9 September.
Figure 8. One Hour Precipitation (OHP) from the KIWA WSR-88D, valid at 0600 UTC 9 September.
Figure 9. Same as figure 6, except valid at 0700 UTC.
Figure 10. Same as figure 7, except valid at 0700 UTC.
Figure 11. Same as figure 8, except valid at 0700 UTC.
Figure 12. Same as figure 6, except valid at 0800 UTC
Figure 13. Same as figure 7, except valid at 0800 UTC.
Figure 14. Same as figure 8, except valid at 0800 UTC.
Figure 15. Same as figure 6, except valid at 0900 UTC.
Figure 16. Same as figure 7, except valid at 0900 UTC.
Figure 17. Same as figure 8, except valid at 0900 UTC.
Figure 18. Same as figure 6, except valid at 1000 UTC.
Figure 19. Same as figure 7, except valid at 1000 UTC.
Figure 20. Same as figure 8, except valid at 1000 UTC.
Figure 21. Three Hour Precipitation (THP), midnight to 3 am MST 9 September, from the KIWA WSR-88D.
