SR SSD 2000-02
1-2000

Technical Attachment

Twelve-hour Climatological Frequencies
of Precipitation in South Florida

Daniel P. Brown
NCEP Tropical Prediction Center / NWSFO Miami, Florida

Introduction

The use of climatological frequencies of precipitation can be a great tool for forecasters to become familiar with precipitation patterns in their forecast area. Because forecasting precipitation probabilities is much harder to visualize than temperature forecasting, an updated climatological precipitation study was needed for South Florida. Since large variations in precipitation frequency occur, studies computing the climatological frequencies should be updated on a regular basis. The more recent the study has been updated, the more useful the results will be to a forecaster's knowledge of current precipitation patterns.

This study continues a previous study (Pifer and Haydu 1990) which was completed in the same manner at the NWSFO in Miami from 1985 through 1989. With the continuation of the previous study climatological probabilities of precipitation can now be completed for a longer period of record. This updated study can also be compared with previous studies completed in the 1960s and mid 1980s.

Procedure

The National Weather Service began a local verification program called AEV (AFOS Era Verification) at the local level in the early 1980s (Barker 1987). This program is run twice daily on the AFOS (Automation of Field Operations and Services) computer system. The AEV verification program produces output which has the potential to improve local forecasts generated at NWS offices. Using output from the verification program, climatological frequencies of precipitation (> .01") for the twelve hour daytime and nighttime periods were computed for Palm Beach, Miami, and Key West.

The AEV local verification program was run for each month to obtain climatological frequencies of precipitation. Due to local computer problems, such as hardware or software crashes, some data were missing and were not available for this study, therefore, for some months the number of days used in this study may not equal the total number of days in those months. For example, hardware problems in January 1989 caused the loss of 12 days of data, so only 19 days of data were available. The period covered by this study is January 1985 through November 1999 for Miami, from October 1986 through November 1999 for Palm Beach, and October 1986 through June 1999 for Key West.

Each day was broken into daytime (1200 to 0000 UTC) and nighttime (0000 to 1200 UTC) periods. These are the periods for which forecasters verify both temperature and precipitation forecasts. The two periods match well with the public perception of the daytime ("today" or "tomorrow") and nighttime ("tonight") periods of the local forecasts. To obtain climatological frequencies of precipitation, the number of precipitation cases (>.01") was divided by the total number of cases (days) available for each month at each observing site.

Results and Comparisons

The results of the entire study are listed in Table 1, broken down by month as well as daytime and nighttime periods for each station. In Tables 2 through 4 results from local verification were compared with similar studies completed by Jorgensen (1967) and Jensenius and Erickson (1987). Results from the study by Jorgensen were available only for Miami for the years 1949 through 1964, however, the study by Jensenius and Erickson provided results for all three south Florida verification stations for the years 1972 through 1985. Because each study was very similar in length, between 13 and 16 years, the average differences should be quite similar.

The average differences in climatological precipitation frequencies for Miami among the previous studies and this study ranged between 3.2 and 3.8 percent, with the average overall difference of 3.5 percent. The overall average differences at Palm Beach and Key West between the Jensenius and Erickson study (1972-1985) were 4.3 percent and 2.8 percent, respectively. Since each report has a similar period of record, the small differences between the climatological precipitation frequencies are most likely the result of normal variations in climate. A few large differences between the studies were observed, however.

One large difference was detected between the Jorgensen (1949-1964) study and the two more recent studies at Miami, where climatological frequencies have increased for the daytime periods in the summer months of June through August. The differences ranged from 7 percent in July to as much as 13 percent in August. It appears the earlier study by Jorgensen (1949-1964) may have used an observation point much closer to the coast than Miami International Airport, which was used in the later two studies. Miami International Airport is about seven miles inland and receives much more rain than downtown Miami due to the setup of the afternoon sea breeze front. It is noted that record keeping began at Miami International Airport in 1942, which was prior to the Jorgensen study. National Climatic Data Center archives show weather observations were maintained at a downtown Weather Bureau Office until 1964. It is not certain, but the possibility of different observing locations could certainly explain the differences in climatological precipitation frequencies noted in the different studies.

A second interesting difference in computed precipitation frequencies was noted in the months of January and March, where frequencies at each south Florida location are about 5 percent larger in the present study, compared to the previous two studies. This could result from the effects of more frequent and stronger El Niño events since 1985. El Niño weather patterns are typically characterized by warmer and wetter winters in south Florida (Lushine 1997). It is uncertain if El Niño is the cause for the increase, but future studies may be able to establish a more reliable connection. One last notable difference in climatological frequencies occurred at Key West in the months of September and October, where the present study shows precipitation frequencies larger on average by about 6 percent. One possible explanation for the higher frequencies is the increase in tropical cyclone activity in the 1990s, compared to years examined in the previous studies. The months of September and October are generally the peak of the hurricane season for south Florida. The increase in precipitation frequency may not be a result from tropical cyclones directly affecting the Keys, but could result from a more general increase in tropical activity, such as more or stronger tropical waves or other disturbances affecting the area. Additional studies need to be completed to determine if the increased tropical activity is a cause of increased climatological precipitation frequencies at Key West.

Conclusion

Hughes (1980) presents a discussion of problems associated with precipitation probability forecasting, and provides suggestions for forecasting improvements. From a forecaster point of view, forecasting precipitation probability is the hardest element to visualize and correctly forecast, at least with good resolution and reliability. Temperature forecasts can be more easily monitored from surface observations each hour and can be verified by plotting current temperature fields. Verifying a point probability forecast generally requires collecting a large number of forecasts and verifying observations.

When verifying precipitation probabilities (PoPs), radar estimates of precipitation coverage can provide similar information as plotted temperature fields. Smith (1977), Smith and Smith (1978) and Naber and Smith (1982) argue that 12-hr composites of hourly radar echo coverage can provide a sufficient estimate of precipitation coverage to allow verification of a single PoP forecast, at least in the southern sections of the county in the summertime. They also demonstrate that such coverage estimates match closely the observed precipitation frequencies at any point in the area. It follows that since most forecasters need a significant amount of time to become familiar with the precipitation patterns (and frequencies) in their area, studies such as this one should help provide background for improving the PoP forecast process.

Additional products from the WSR-88D could be produced to aid the forecaster in verifying probability forecasts. Currently, a 24-hr User Selectable Precipitation (USP) map can be produced to show total 24-hr radar rainfall estimates, however, it would be to the forecasters' benefit to produce a composite 12-hr areal coverage product from the WSR-88D for each zone in the forecasters area of responsibility. This has been done at NWSFO Tulsa (Spaeth 1999). The graphic produced uses Stage III Precipitation data from the radar. As Spaeth notes, during typical summertime convective forecast regimes, forecasters could use areal coverage from the previous day as a "first guess," then adjust forecast probabilities up or down depending on their analysis of numerical model guidance and other indicators of whether the environment is more or less favorable for convection to develop and spread. The same idea has occasionally been used in forecasting PoP for South Florida, except using climatological precipitation frequency as the first guess.

After reviewing the results and variability of precipitation frequency, one realizes the importance of long term studies to smooth the variability. A long term study to determine climatological precipitation frequencies is one way forecasters can get an idea of a first guess PoP. One advantage of the local AEV verification program is that the results of this study can be updated locally on a regular basis. The data and results of this study can be a great asset in forecasting probabilities of precipitation. It is hoped this study can be continued so variations in precipitation frequencies can be smoothed out, therefore, representative climatological frequencies of precipitation can be a computed for any recent changes in precipitation patterns and aid local forecasters in forecasting probabilities of precipitation for South Florida.

Table 1. Climatological Frequencies of Precipitation for South Florida.*

Nighttime (0000 to 1200 UTC) Frequencies of Precipitation

Station

January

February

March

April

May

June

Palm Bch.

22%

16%

20%

15%

19%

32%

Miami

17%

12%

15%

13%

17%

28%

Key West

16%

11%

13%

9%

17%

21%

Station

July

August

September

October

November

December

Palm Bch.

21%

31%

39%

26%

19%

15%

Miami

24%

25%

33%

27%

18%

12%

Key West

21%

26%

35%

25%

13%

13%

Daytime (1200 to 0000 UTC) Frequencies of Precipitation

Station

January

February

March

April

May

June

Palm Bch.

24%

20%

24%

20%

22%

42%

Miami

17%

14%

17%

16%

23%

44%

Key West

14%

11%

12%

11%

12%

23%

Station

July

August

September

October

November

December

Palm Bch.

36%

41%

44%

28%

22%

17%

Miami

42%

48%

43%

30%

19%

14%

Key West

25%

32%

38%

24%

15%

13%

* From (1985-November 1999) for Miami, October 1986-November 1999 for Palm Beach,

and October 1986-June 1999 for Key West.

Table 2. Comparison of Miami Climatological Frequencies of Precipitation

Nighttime (0000 to 1200 UTC)

Months

1949-1964

ESSA WB 5

1972-1985

NOAA NWS 39

1985-1999*

Local Verification

January

11%

10%

17%

February

15%

15%

12%

March

10%

10%

15%

April

12%

12%

13%

May

15%

17%

17%

June

24%

24%

28%

July

25%

24%

24%

August

26%

34%

25%

September

36%

33%

33%

October

31%

29%

27%

November

16%

21%

18%

December

16%

13%

12%

Daytime (1200 to 0000 UTC)

Month

1949-1964

ESSA WB 5

1972-1985

NOAA NWS 39

1985-1999*

Local Verification

January

12%

12%

17%

February

15%

17%

14%

March

12%

11%

17%

April

15%

13%

16%

May

24%

28%

23%

June

36%

41%

44%

July

35%

37%

42%

August

35%

45%

48%

September

40%

46%

43%

October

33%

28%

30%

November

13%

21%

19%

December

12%

12%

14%

* January 1985-November 1999

Table 3. Comparison of West Palm Beach Climatological Frequencies of Precipitation

Nighttime (0000 to 1200 UTC)

Month

1972-1985

NOAA NWS 39

1987-1999*

Local Verification

January

17%

22%

February

17%

16%

March

12%

20%

April

10%

15%

May

21%

19%

June

22%

32%

July

25%

21%

August

26%

31%

September

33%

39%

October

22%

26%

November

24%

19%

December

18%

15%

Daytime (1200 to 0000 UTC)

Month

1972-1985

NOAA NWS 39

1987-1999*

Local Verification

January

17%

24%

February

21%

20%

March

13%

24%

April

16%

20%

May

31%

22%

June

37%

42%

July

34%

36%

August

40%

41%

September

44%

44%

October

27%

28%

November

24%

22%

December

19%

17%

* October 1986-November 1999

Table 4. Comparison of Key West Climatological Frequencies of Precipitation

Nighttime (0000 to 1200 UTC)

Month

1972-1985

NOAA NWS 39

1987-1999*

Local Verification

January

12%

16%

February

14%

11%

March

8%

13%

April

10%

9%

May

17%

17%

June

23%

21%

July

21%

21%

August

31%

26%

September

29%

35%

October

18%

25%

November

15%

13%

December

14%

13%

Daytime (1200 to 0000 UTC)

Month

1972-1985

NOAA NWS 39

1987-1999*

Local Verification

January

11%

14%

February

13%

11%

March

7%

12%

April

10%

11%

May

13%

12%

June

21%

23%

July

26%

25%

August

32%

32%

September

33%

38%

October

17%

24%

November

15%

15%

December

10%

13%

* October 1986-June 1999

References

Barker, Timothy W., 1987: AEV Local Verification for Aviation, Precipitation, and Temperature Programs: AV, REL, TEM. Western Region Computer Programs and Problems, NWS WRCP-No. 42, National Oceanic and Atmospheric Administration, U.S. Department of Commerce, 33pp.

Hughes, Lawerence A., 1980: Probability Forecasting - Reasons, Procedures, Problems, NOAA Technical Memorandum NWS FCST 24, National Oceanic and Atmospheric Administration, U.S. Department of Commerce, 84 pp.

Jensenius, John S. Jr. and Erickson, Mary C., 1987: Monthly Relative Frequencies of Precipitation for the United States for 6-, 12- and 24-H Periods, NOAA Technical Report, NWS 39, National Oceanic and Atmospheric Administration, U.S. Department of Commerce, 262 pp.

Jorgensen, Donald L., 1967: Climatological Probabilities of Precipitation for the Conterminous United States. ESSA Technical Report WB-5, Environmental Science Services Administration, U.S. Department of Commerce, 60 pp.

Lushine, James B., 1997: El Nino Public Information Statement. National Weather Service, Miami, FL.

Naber, Pamela S. and Smith, Daniel L., 1983: Evaluation of Point Precipitation Probability Forecasts Using Radar Estimates of Rainfall Areal Coverage, NOAA Technical Memorandum NWS SR-108, National Oceanic and Atmospheric Administration, U.S. Department of Commerce, 13 pp.

Pifer, Bob E. and Haydu, Kenneth J., 1990: Utilizing the AEV Program Output in the WSFO Public Forecast Program.

Smith, Daniel L., 1977: An Examination of Probability of Precipitation Forecasts in Light Rainfall Areal Coverage, NOAA Technical Memorandum NWS SR-89, National Oceanic and Atmospheric Administration, U.S. Department of Commerce, 20 pp.

Smith, Daniel L. and Smith, Matthew, 1978: A Comparison of Probability of Precipitation Forecasts and Radar Estimates of Rainfall Areal Coverage, NOAA Technical Memorandum NWS SR-96, National Oceanic and Atmospheric Administration, U.S. Department of Commerce, 17 pp.

Spaeth, D.R., 1999: Alternative PoP Verification at WFO Tulsa. Proc. 15th Conference On IIPS for Meteorology, Oceanography and Hydrology, 79th AMS Conference, Dallas, Texas, Amer. Meteor. Soc., 95-98.