Lake Junaluska

Tropical Precipitation Statistics - Methodology

by Dan McKemy

Introduction

Few studies have been conducted on rainfall climatologies associated with tropical cyclones in the Southeastern United States. Meteorologists and climatologists have estimated how much tropical cyclones contribute to annual rainfall by conducting broad studies over a considerable portion of the United States, but nothing has been done on a statewide level that focuses on various regions in that state. This study addresses in detail the rainfall associated with tropical cyclones across North Carolina from 1980-2009 using more than 350 stations.

Methods

Precipitation data was provided by the State Climate Office of North Carolina's CRONOS database. The majority of the station types used for this study were COOP, ECONET, ASOS, and RAWS. Daily precipitation totals were used for all of the stations, even if some of them reported hourly precipitation to keep the data collection methods consistent.

The methods used to determine the tropical cyclone radius and collect precipitation data in this paper are more flexible and accurate compared to research methods by others. For example, both Nogueira and Kiem (2010) and Gleason (2006) used a fixed radius around the center of the storm (500km and 600km, respectively) regardless of a storm's actual size. For this project, each tropical cyclone was analyzed on satellite and/or radar to determine the radius of its precipitation bands, except for 1 tropical cyclone in 1980, where satellite data was not available. Using PHP/MySQL, a program was written that took the radius of the storm at a specific NHC advisory point and pulled precipitation data from stations within that radius. While more labor intensive than using a fixed radius of influence, this method ensures accurate attribution of rainfall to each tropical cyclone. Storms that cover a much larger geographic extent have equal weighting as those storms that have much smaller horizontal scales. This method also ensures that convective thunderstorm activity outside of the influence of the tropical cyclone is not counted as tropically induced. Precipitation totals for tropical rainfall events can then be easily compared to seasonal and annual precipitation totals to quantify the contribution of tropical cyclones to the overall climate.

The NHC issued advisory points every 6 hours (0z, 6z, 12z, and 18z), and the latest point issued for a given day was 18z (as 0z marks the beginning of a new day). To account for the motion of the tropical cyclone, the radius used for the storm was often extended to account for precipitation that fell after 18z for the day in an area.

Precipitation from tropical cyclones that transitioned into extra-tropical storms were included in the database, as long as the NHC had an advisory issued for it. While Nogueira and Kiem (2010) only included rainfall associated with hurricanes and tropical storms, Knight and Davis (2007) included precipitation of tropical cyclones during and after its transition into a extra-tropical cyclone. For some of the storms, the NHC would issue its last advisory early or mid-way through a day, with precipitation occurring after the advisory had been issued. Unfortunately, due to the setup of how the stations recorded precipitation, anything that fell during the day an advisory was issued was included in the database. Additionally, if the tropical cyclone interacted with a front or some sort of boundary, there was no way to easily separate the precipitation that just was associated with the tropical cyclone, and as a result, the enhanced precipitation rates due to the front/boundary were included in the database. We feel this is appropriate as the presence of rainfall in these situations is likely dominated by the proximity of the tropical cyclone as compared to any approaching synoptic forcing.

Results and Discussion

One purpose of this project was to not only analyze how much tropical precipitation contributes to annual rainfall, but also the percentage of tropical precipitation during "hurricane season", or the warm season as labeled in the graphs below (June-November). While tropical cyclones do form before and after the warm season period, they rarely affect North Carolina and have trivial impacts overall. In fact, the only storms to affect North Carolina outside of the warm season period are Tropical Depression #1 (2009) and Andrea (2007) in the past 30 years.


Fig. 1: Contribution of tropical precipitation to the warm season total from 1980-2009. Yellow and light green dots indicate a lower percentage, while dark green and blue dots indicate a higher percentage.
Fig. 2: Contribution of tropical precipitation to the annual total from 1980-2009. Yellow and light green dots indicate a lower percentage, while dark green and blue dots indicate a higher percentage.
Region Average annual tropical precip. Percentage of tropical precip. during the warm season Percentage of tropical precip. during the year Standard deviation of tropical precip.
Mountains 2.7 in. 8.9% 5.0% 4.4 in.
Piedmont 2.9 in. 11.6% 6.7% 3.4 in.
Coastal Plain 4.1 in. 13.6% 8.2% 4.9 in.
North Carolina 3.3 in. 11.5% 6.7% 4.3 in.
Table 1: Precipitation statistics for 1980-2009.

Figure 1 reveals that the coastal plain of North Carolina receives the most precipitation per year on average than anywhere else across the state, with the mountains ranking last in all of the categories above. Surprisingly, the coastal plain only averages about 3% more tropical precipitation per year than the mountains do (although that 3% amounts to over an inch of extra precipitation). The standard deviation for each region and the state were quite high, and this is due to the irregular year-to-year frequency of tropical cyclones that affect the state.


Fig. 3: Average annual tropical precipitation for North Carolina from 1980 to 2009.

Figure 3 demonstrates the irregular frequency of tropical cyclones that affect North Carolina, but there is general pattern that can be made out. Except for 1985, the 1980-1994 period was relatively calm for North Carolina. From 1995-2009, there appears to be an increase in tropical storm precipitation, and the figures and tables below support that notion.


Fig. 4: Contribution of tropical precipitation to the warm season total from 1980-1994. Yellow and light green dots indicate a lower percentage, while dark green and blue dots indicate a higher percentage.
Fig. 5: Contribution of tropical precipitation to the annual total from 1980-1994. Yellow and light green dots indicate a lower percentage, while dark green and blue dots indicate a higher percentage.
Region Average annual tropical precip. Percentage of tropical precip. during the warm season Percentage of tropical precip. during the year Standard deviation of tropical precip.
Mountains 1.6 in. 5.2% 2.9% 2.8 in.
Piedmont 1.3 in. 5.2% 2.9% 2.2 in.
Coastal Plain 2.0 in. 7.2% 4.2% 3.4 in.
North Carolina 1.6 in. 5.9% 3.4% 2.9 in.
Table 2: Precipitation statistics for 1980-1994.
Fig. 6: Contribution of tropical precipitation to the warm season total from 1995-2009. Yellow and light green dots indicate a lower percentage, while dark green and blue dots indicate a higher percentage.
Fig. 7: Contribution of tropical precipitation to the annual total from 1995-2009. Yellow and light green dots indicate a lower percentage, while dark green and blue dots indicate a higher percentage.
Region Average annual tropical precip. Percentage of tropical precip. during the warm season Percentage of tropical precip. during the year Standard deviation of tropical precip.
Mountains 3.6 in. 11.8% 6.7% 5.1 in.
Piedmont 4.0 in. 15.8% 9.1% 3.6 in.
Coastal Plain 5.4 in. 17.8% 10.8% 5.2 in.
North Carolina 4.4 in. 15.3% 9.0% 4.8 in.
Table 3: Precipitation statistics for 1995-2009.

The two distinct periods have large differences, with warm season precipitation varying as much as 200-300% between the two periods. The average tropical precipitation is two to three times higher in the 1995-2009 period, with the coast having the highest annual average at 5.4 inches. An interesting observation is that the standard deviation for the piedmont remains relatively low compared to the mountains, coast, and the state of North Carolina for all time periods. One likely reason for this may be due to the track of tropical cyclones across Southeast. For example, a tropical cyclone tracking up through the Gulf of Mexico could have its precipitation affect parts of the mountains and western piedmont, but the coast is more than likely going to miss out on most of the precipitation. Another possibility is that a tropical cyclone could track up along the Southeastern U.S. coastline, affecting the coastal plains and piedmont of North Carolina, but leaving the mountains fairly dry. In both situations, the piedmont receives precipitation from the tropical cyclones, leading to lower standard deviation values.

Conclusions

Tropical cyclones have respectable impacts on rainfall climatologies in North Carolina as they can make up anywhere between 8-13% of the warm season precipitation. While the amount of precipitation received per year from tropical cyclones is temperamental, there is a general pattern in which activity has increased over the past 15 years (1995-2009) compared to the previous 15 years before that (1980-1994). This is associated with the general increase in tropical cyclone activities that increased beginning in 1995 and is often associated with the shift in phase of the Atlantic Multidecadal Oscillation.

Future work with this project should expand on the number of states analyzed, with the main goal being to include all of the Gulf Coast, Southeastern, and Mid-Atlantic states. Additionally, data beyond 1980 should be included for future studies so it can be comparable to other articles such as Nogueira and Kiem (2010), which had data from 1960-2007. Having a larger timeframe will also make it more useful to compare active and non-active periods to phases of ENSO, the AMO, and other large scale phenomena that affect global atmospheric circulation. However, there are challenges in collecting accurate precipitation associated only with tropical cyclones in the periods prior to active satellite monitoring.

Acknowledgements

I would like to thank Dr. Ryan Boyles and the State Climate Office of North Carolina for the funding provided to work on the project. Additionally, I would like to thank Corey Davis and Bradley McLamb for their assistance with the computer programming and testing that went into the project.

References

D.B. Knight, Davis R.E. (2007). Climatology of tropical cyclone rainfall in the southeastern United States. Phys Geogr, 28: 126-147.

Gleason, B.E. (2006). Characteristics of tropical cyclone rainfall in the united states. 27th Conference on Hurricanes and Tropical Meteorology, Session 16C, Tropical Cyclones and Climate V -- Atlantic Basin.

R.C. Nogueira, Kiem B.D. (2010). Contributions of Atlantic tropical cyclones to monthly and seasonal rainfall in the eastern United States 1960-2007. Theor Appl Climatol. Retrieved from http://www.springerlink.com/content/g782841705227418/fulltext.html

Appendix

  • All stations are located in North Carolina
  • 1980-2009 Annual: Tropical contribution to annual precipitation from 1980-2009
  • 1980-2009 Warm Season: Tropical contribution to warm season precipitation from 1980-2009
  • 1980-1994 Annual: Tropical contribution to annual precipitation from 1980-1994
  • 1980-1994 Warm Season: Tropical contribution to warm season precipitation from 1980-1994
  • 1995-2009 Annual: Tropical contribution to annual precipitation from 1995-2009
  • 1995-2009 Warm Season: Tropical contribution to warm season precipitation from 1995-2009


show/hide appendix table