5.4 Geographic Variation in the Maximum 7-Day Precipitation Accumulation

Tuesday, 30 January 2024: 9:15 AM
Key 10 (Hilton Baltimore Inner Harbor)
Owen A. Kelley, PhD, NASA, Greenbelt, MD; George Mason Univ., Fairfax, VA
Manuscript (5.6 MB)

Observational datasets disagree about how much rain can fall in seven days. While the observational record is imperfect, it suggests that the maximum 7-day accumulation varies with a location's average-annual accumulation and whether the location is over land or ocean. The present study formulates an upper bound to the maximum 7-day accumulation as a function of these variables. Such an upper bound can give context to the accumulation from an individual storm. Such an upper bound can help communicate the threat posed by a forecasted storm. In general, an unusually large accumulation for a particular location is associated with societal impacts there.

There is no universally acknowledged reference data set that specifies what storm-total accumulation should be considered unusually large for each location on the globe. There is a standard reference that covers only the United States, and that is NOAA's Atlas 14. Atlas 14 is based on rain gauges, some with over a century of daily observations. Atlas 14 includes maps of the maximum precipitation accumulation for various durations, including 1 to 7 days, and with various average-return intervals, including 2 to 1,000 years.

The present study examines a much simpler statistic that is near global in coverage. Specifically, the present study examines the maximum 7-day precipitation accumulation during the past 23 years of satellite observations.

Two satellite gridded precipitation datasets are used in the present study, one tuned for stability over decades (GPCP Daily v3.2) and the other designed to extract information from every available satellite observation (Version 7 Final IMERG). The present study averages each of the two datasets individually to create daily 1x1-degree latitude-longitude grids. In this way, some of the noise in the higher-resolution extreme statistics are damped out and the resulting two grids are easier to work with. A 1x1-degree grid is sufficient to resolve synoptic-scale weather systems.

The two satellites datasets show a similar pattern when the 23-year maximum 7-day accumulation is plotted against each 1x1-degree grid box's average-annual accumulation. The middle of the distribution is similar, as is the 99th-percentile extreme value of the distribution.

While the overall pattern is similar, the two satellite datasets report that a different 7-day period experienced the 2000-to-2023-maximum 7-day accumulation in most of 1x1-degree grid boxes. There is one class of grid boxes in which the two satellite datasets are more likely to agree on the 7-day period that brings the maximum accumulation. That class is the grid boxes with the 99th-percentile highest 7-day accumulation. The events that cause these truly extreme accumulations are so different from typical events that, even though the 7-day accumulation in the two datasets differs, both datasets agree on which event brought the highest 7-day accumulation to the associated grid boxes. For example, IMERG and GPCP agree that Hurricane Harvey (2017) brought to the Houston, Texas, area the largest 7-day accumulation that this area experienced during 2000-2023.

IMERG stands for Integrated Multi-satellitE Retrievals for the Global Precipitation Measurement (GPM). The HDF5 files output by the IMERG algorithm have half-hour and 0.1-degree resolution and cover the globe from June 2000 to the present. IMERG puts a priority on including all available satellite observations, including passive microwave and infrared observations to improve short-term accuracy of precipitation estimates. To achieve this goal, IMERG de-emphasizes avoiding biases that shift from year to year.

GPCP stands for Global Precipitation Climatology Project. GPCP Daily version 3.2 has 0.5-degree resolution and covers June 2000 to the present. Between 55 South and 55 North latitude, GPCP Daily is produced by scaling IMERG daily estimates by GPCP Monthly accumulations. GPCP Monthly, in turn, is created from a subset of satellite observations that are believe to minimize shifts in bias from year to year. GPCP puts priority on avoiding false trends rather than in including all possible observations during an individual time period.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner