SR/SSD 99-5

2-1-99

Technical Attachment

Excessive Rainfall (?) in South Florida: January 2, 1999

Rusty Pfost

MIC, NWSFO Miami

While visiting the South Florida Water Management District in West Palm Beach on January 19, I inspected the tipping bucket rain gauge which reported the astounding 31 inch total in Palm Beach Gardens (north of West Palm Beach). Over 20 inches of that total supposedly fell in 3 hours between 5 and 8 AM on January 2 (see table 1):

Table 1

Rainfall Data for SFWMD Gage S44 North Palm Beach, Fla.

time period /99 hourly rainfall accumulated rain
2-3 AM 4.30 4.30
3-4 AM 2.18 6.48
4-5 AM 2.00 8.48
5-6 AM 7.08 15.56
6-7 AM 9.37 24.93
7-8 AM 5.08 30.01
8-9 AM 0.42 30.43
9-10 AM 0.28 30.71

The SFWMD tipping bucket has excellent exposure, was surrounded by at least an eight-foot high chain link fence, and appeared to be in excellent working order. It is an old gauge with what looked like a snow shield attached to the bottom, where the rain runs onto the tipper, to prevent water from running down the outside of the gauge from getting inside and affecting the total. However, we found a number of dead ants inside the gauge apparatus, and a substantial amount of water standing in the gauge, making me wonder if the gauge itself may have filled with water and perhaps the remains of an ant mound were floating inside, wreaking havoc with the tipping mechanism.

In my previous experience with torrential rains, where we had a standard eight-inch gauge, a weighing gauge, and a tipping bucket gauge on the roof of the Key West airport terminal, the tipping bucket in hurricanes Elena and Kate (in 1985) recorded from 20-30 percent less than the other gauges. Also, the flooding associated with the Palm Beach County rain was substantial (primarily affecting homes - and many of those of considerable value), but it was not of the same magnitude which I observed in Slidell, Louisiana, or Hancock County, Mississippi, in 1995 when 27 inches fell in a 30-hour period.

The Miami WSR-88D was operating with the tropical Z-R relationship 250 R1.2 (unintentionally) for this event, while Melbourne's radar used the nominal Z-R 300 R1.4. The rain gauge site is about equidistant from both RDAs. The KMLB radar in real time estimated about eight inches, and we estimated just over 20 inches at Miami; neither anywhere near the 31 inch gauge report. The next largest measured rainfall amount nearby was only in the teens, so the gauge in question was definitely an outlier.

From a subsequent WATADS study conducted by Miami SOO Jack Gross using level II archive data, again using the tropical Z-R for a time period 0351 UTC to 1350 UTC, the maximum precipitation estimate was 14.93" just northwest of Lake Park. The radar estimated 11.7" over North Palm Beach where SFWMD gage S44 is located. At no time during the period from 5:00 UTC to 13:50 UTC did Jack's study show estimated rainfall rates more than 4"/hr. Jack's study is available on the Miami NWSFO home page at http://www-mfl.nhc.noaa.gov/Palm_Beach.html.

The SFWMD personnel asked me what they might do to verify their total. Two suggestions I had were to run a backward check by using the nearby river gauge and integrating the amount of water under the hydrograph curve as runoff and comparing that with a mean areal precipitation amount for the basin (including the 31 inch report), and multiplying by the area of the basin. One could then compare the volume of water under the hydrograph with the approximate volume of water estimated from rainfall and see how they match up.

I also suggested they do a calibration of the tipping bucket with a larger rainfall rate than is typically done. From my understanding of the ASOS tipping bucket, almost ALL the studies done have been with rainfall rates typical of stratiform rain, none (that I know of) typical of south Florida convective rain, let alone the sometimes torrential tropical rain rates.

In short, I remain skeptical of this report, but I cannot rule it out, either. It is unusual for a tipping bucket to OVERESTIMATE a torrential rain, unless the circuitry was shorted out or (perhaps) there were bugs (specifically, ants) in the equipment. Further investigation may shed more light on what actually occurred.