1. INTRODUCTION

Quantitative precipitation forecasting is an important aspect of the modernized National Weather Service as it predicts how much rain is expected over a specific river basin per rain event. Quantitative precipitation forecasts (QPFs) are issued on a daily basis and normally include four six-hour forecasts of areally averaged rainfall over local river basins. River Forecast Centers (RFCs) routinely include QPFs in their calculations of expected river stages making QPF critical in the prediction of floods and for the protection of life and property.

Some important aspects of QPF forecasting involve knowledge of the topography, river basins, normal rainfall patterns, rainfall frequency, and synoptic conditions of heavy rain events for the local area of responsibility. In the modernized National Weather Service, the NEXRAD Weather Service Forecast Office (NWSFO) Memphis, TN will issue QPFs for its present Hydrological Service Area (HSA) which includes the Missouri bootheel, northeast Arkansas, west Tennessee and northern Mississippi (Fig. 1). This HSA is basically the same area as the Memphis County Warning Area (CWA) with only a few minor differences described in Figure 1. The purpose of this paper will be to provide a complete rainfall climatology for the entire Memphis CWA addressing the important aspects of QPF forecasting which will serve as guidance for the quantitative precipitation forecaster.

2. DATA

Much of the data used in this study was obtained from the National Climatic Data Center (NCDC) in Asheville, North Carolina. Monthly and annual rainfall 'normals', thirty year averaged values from 1961-1990, of cooperative stations across the Memphis CWA were obtained from the Climatography of the United States Series #81 paper published by NCDC. The locations of the cooperative stations, included in the calculation of 'normal' rainfall values, are shown in Figure 2.

Maximum rainfall frequency information for the Memphis CWA was obtained from a Weather Bureau publication entitled 'Rainfall Frequency Atlas of the United States'. This technical paper was published in 1961 using 20 years of data and is the most current information at the time of this publication. It is assumed that this rainfall frequency information is still relatively accurate, even over thirty years later. Rainfall frequency graphs were compiled for Memphis using hourly precipitation data extracted from the Solar and Meteorological Surface Observation Network (SAMSON) CD-ROM distributed by NCDC. The manipulation of the raw hourly precipitation data for Memphis into six hour frequency categories was accomplished with a computer program developed by Brain Walawender at NWSFO Topeka.

Hourly precipitation data for six stations across the Memphis CWA were obtained from NCDC on floppy discs. The stations were selected on the basis of having a relatively continuous thirty year hourly precipitation record from 1961 through 1990 and being spatially representative of the Memphis CWA. After analyzing the hourly precipitation data to determine 'heavy rain' events, 'Daily Weather Map' charts were obtained from the Climate Prediction Center in Washington D.C. in order to evaluate the surface and 500 mb synoptic patterns on the selected heavy rain days.

3. TOPOGRAPHY AND RIVER BASINS

Most of the NWSFO Memphis CWA is located in the lower Mississippi river valley with little variation in topography. The map produced in Figure 3 is a general representation of the local topography using the specific elevations of the cooperative stations presented in Figure 2. From this map, one can get a general, almost complete, idea of the variations in local topography across the Memphis CWA.

The terrain of eastern Arkansas and the Missouri bootheel is relatively flat except for Crowley's Ridge which is an incline of around 150 to 200 feet above the surrounding topography. This ridge stretches north to south from central Clay County south to central Lee County and is generally 10 miles in width. Most of the vegetative cover across eastern Arkansas and the Missouri bootheel is found on Crowley's Ridge, a mixture of deciduous and evergreen forests, with the rest of the area comprised mostly of open farmland.

The topography of west Tennessee consists of a gentle upward slope away from the Mississippi river, ranging from 265 feet at the Memphis station to 580 feet at the Paris station. The vegetative cover across west Tennessee is thicker than that of eastern Arkansas and is comprised mainly of deciduous forests.

Northern Mississippi has the greatest variation in topography of the cooperative stations in the Memphis CWA ranging from Ashland at 630 feet above sea level to Swan Lake at 145 feet above sea level. The actual highest point in the Memphis CWA is Woodall Mountain in Tishomingo County Mississippi which is 806 feet above sea level. This elevated topography separates the lower Mississippi River basin, into which most rivers across the Memphis CWA drain, from the Tombigbee River basin in northeast Mississippi, which flows south directly into the Gulf of Mexico. The vegetative cover across most of northern Mississippi is comprised of a mixture of deciduous and evergreen forests.

The locations of the river basins in the Memphis HSA are seen in Figure 4 with Table 1 providing a listing of the definitions of the river basins. These river basins are the water sheds of the river forecast points in the Memphis HSA. This means that any rain that falls, for example, in the MEMT1 river basin area will cause a direct rise in the MEMT1 gage reading at Memphis. QPF forecasters attempt to predict the average areal precipitation for each of these river basins in order to assist the river forecast centers in their prediction of river conditions for each individual gage.

4. NORMAL RAINFALL

An important tool for QPF forecasters is knowledge of the normal monthly rainfall of the local area and its spatial distribution. Normal rainfall for a specific station is determined from averaging the past thirty years of data every ten years, i.e. 1941-1970, 1951-1980, and 1961-1990. The 'normal' rainfall for the Memphis CWA as a whole was determined by averaging the known normal (1961-1990) monthly rainfall amounts from seven stations which adequately represent the CWA spatially (Fig. 5). The stations used to represent the Memphis CWA in Figure 5 were Jonesboro, AR, Paris, TN, Jackson, TN, Memphis, TN, Clarksdale, MS, Tupelo, MS and Columbus, MS.

Looking at Figure 5, the wettest month of the year across the Memphis CWA is December with around 5.5 inches while October is the driest with around 3.2 inches. The five month period consisting of June through October (summer through most of fall) can be considered the dry season as only a third of the total annual rain occurs during this time period. The seven month period of November through May (late fall through spring) can be considered the wet season as two-thirds of the annual rainfall across the CWA occurs during this time period.

The spatial distribution of annual rainfall across the Memphis CWA is generally from the northwest to the southeast (Fig. 6), ranging from 48 inches across sections of northeast Arkansas and the Missouri bootheel to over 58 inches in northeast Mississippi. This general pattern of rainfall could be the result of developing low pressure systems over the Gulf of Mexico creating an increase in rainfall across the CWA as they move northeast, on average, and also the idea that most cold fronts are oriented southwest to northeast as they cross the Memphis CWA. As these fronts move from west to east, on average, the available moisture, from the Gulf of Mexico to the south, would have more time to be pulled northward thus revealing a more northern extent as one traveled east. Although variations in topography are minimal across the Memphis CWA, there does seem to be a slight increase from southwest to northeast which might have a slight affect on this rainfall distribution.

Monthly rainfall distribution maps were grouped together according to the seasons of the year which are defined in this study according to the standard meteorological classification: December - February (winter), March - May (spring), June - August (summer), and September - November (fall). In the winter (Fig. 7), rainfall patterns reveal a definite northwest to southeast pattern with large differences in rainfall amounts. This can be attributed to the fact that synoptic scale fronts (predominately oriented southwest to northeast) and developing lows in the Gulf of Mexico are the dominate features during these months. These synoptic fronts create widespread, definite rainfall patterns without much localized effects.

During the first two spring months (Fig. 8), the northwest to southeast rainfall pattern is still seen, but it is not as definite and does not have as large a difference in rainfall amounts as the winter months. In May, rainfall amounts are similar across the entire CWA which could possibly be attributed to an increased likelihood of stalling or slow moving cold fronts which generally become weaker during the late spring in the Memphis CWA.

The summer rainfall patterns (Fig. 9) reveal that topography might have somewhat of a role in the formation of afternoon thunderstorm activity. While rainfall amounts are generally evenly divided across the CWA, there appears to be a minimum over eastern Arkansas with the higher terrain over west Tennessee and northern Mississippi apparently enhancing amounts somewhat in these areas. Also, the prevailing south to southwesterly wind direction during the summer, bringing moisture from the Gulf of Mexico, would likely create higher rainfall amounts across the southeastern sections of the CWA.

Rainfall amounts during the fall (Fig. 10) are fairly evenly distributed especially during September and October. This is likely the result of a lack of afternoon convection and strong fronts during these months which would usually cause significant variations in rainfall distribution in other months. The northwest to southeast rainfall gradient does begin to develop in November as stronger cold fronts reach the area, but the differences in amounts are still not as great as during the winter.

5. RAINFALL FREQUENCY

A frequent question about rainfall in any area is 'what is considered a heavy rain and how often does it occur?'. Using charts from the 'Rainfall Frequency Atlas of the U.S.', the maximum amount of rain in a 3, 6, 12 and 24 hour time period across the Memphis CWA once every 100 years was determined (Fig. 11). A compilation of maximum rainfall amounts, using the other return periods of one, two, five, ten, twenty-five and fifty years from the 'Rainfall Atlas of the U.S.', was produced (Table 2) in order to show the maximum amounts of rainfall expected in a three, six, twelve and twenty-four hour period.

Graphs showing the frequency of rainfall for different categories of rainfall amounts were developed using the Memphis hourly precipitation data. On an annual basis (Fig. 12), a maxima in light rainfall amounts (0.01-0.49 inch) tends to occur between noon and 6 pm local time while a maxima in heavy rainfall amounts (>0.49 inch) tends to occur between 6 pm and midnight. While the temporal differences in the distribution of rainfall amounts are subtle, the slight maxima in both light and heavy rainfall categories during the afternoon and evening hours is most likely the result of the diurnal increase in convective activity from solar heating. It is also interesting to note that in all seasons, a noticeable drop in the rainfall frequency in the 0.05-0.09 inch category is observed. This result eludes any explanation from these two authors.

During the winter months (Fig. 13), rainfall is fairly evenly distributed between the different time periods with a nocturnal bias noted between midnight and 6 am for most rainfall amounts. In the spring (Fig. 14), light rainfall (<0.25 inch) occurs mainly during the 6 am to noon time period while the majority of the heavier rainfall (>0.24 inch) occurs during the 6 pm to 6 am time period, again showing a nocturnal bias. The summer months (Fig. 15) have most rainfall occurring between noon and 6 pm with the 6 pm to midnight hours nearly as active. During fall (Fig. 16), most rainfall amounts are fairly evenly distributed with an afternoon maxima observed for most categories between noon and 6 pm.

There seems to be a slight increase in rainfall occurrence for amounts greater than an inch between the hours of 6 pm and 6 am during the winter and spring. While during the summer and fall, there is an afternoon and evening maxima between the hours of noon and midnight. The observed maxima during the summer and fall months is most likely the result of afternoon and early evening thunderstorm activity from solar heating. The nocturnal bias during the winter and spring months is likely the result of 1) the average wind speeds in the boundary layer being greater at night leading to greater advection of low level moisture 2) greater instability in cloudy locations due to cooling aloft and little temperature change below the clouds and 3) afternoon convection leading to mesoscale boundaries and echo training several hours later, often after sunset. (Hoxit et.al., 1978)

6. METEOROLOGICAL CONDITIONS OF PAST HEAVY RAIN EVENTS

An important tool for local forecasters in the prediction of heavy rain is the recognition of the surface and upper air patterns which lead to 'heavy rain' events across the area of concern. Thirty years of hourly precipitation data were analyzed at six selected stations which were chosen to spatially represent the Memphis CWA (Fig. 17). A 'heavy rain' event is defined for the Memphis CWA as one that produces, at two or more stations, three or more inches of rain in twelve hours or less, which is roughly the maximum amount of rain expected in a twelve hour period every two years (Weather Bureau, 1961). In addition, a 'heavy rain' event is defined for this study as one that produces at least six inches in twelve hours or less at a single station, which is about the maximum amount of rain expected once every 100 years at a particular site in the Memphis CWA (Weather Bureau, 1961).

Selecting two or more stations with the criteria of three or more inches in twelve hours or less was done to ensure that the heavy rain was truly the result of a large scale pattern and not just an isolated event such as a heavy thunderstorm. Also, by using at least two stations, any flaws or incorrect data should be filtered out. After selecting the dates from which at least three inches of rain fell in 12 hours or less at a station, the dates were cross checked with the other five stations to see if heavy rain fell at another station within the same general timeframe. 'Heavy rain' at two separate stations was considered from the same event if the starting time of rainfall at one station was within twelve hours of the ending time of rainfall at the other station.

After the dates of heavy rain events were determined, copies of the Daily Weather Map charts for each of the selected dates were obtained and analyzed in order to classify the events as defined by Maddox et.al. (1979) (Table 3). Maddox et.al. defined four different conditions that produce flash flooding rains (synoptic, frontal, mesohigh and western) which will be considered in this study to categorize heavy rain events. The 'western' category deals with conditions that are exclusively found in the western U.S. and thus will not be considered for this study. Also, Maddox et.al. excluded those events that were tropical in origin, but tropical events will be used in this study.

Maddox et.al. defined 'synoptic' events as those which are the result of an intense synoptic scale cyclone or frontal system. Synoptic events normally develop in association with a quasi-stationary or slow-moving front, usually oriented from southwest to northeast, with an east to northeast moving trough at 500 mb and with heavy rains occurring in the warm sector ahead of the front. 'Frontal' events were defined as those developing in association with a stationary or very slow moving front, usually oriented from west to east, with heavy rains found near the large-scale 500 mb ridge, and with heavy rains usually occurring on the cool side of the surface front. The definition of 'mesohigh' events were those that occur in association with quasi-stationary thunderstorm outflow boundaries which had been generated by previous convective activity. The heaviest rains usually occurred near the 500 mb large-scale ridge position and on the cool side of the surface boundary, usually to the south or southwest of the mesohigh pressure center. With the criteria of using two or more stations to define most heavy rain events in this study, some mesohigh events may have been filtered out. Also, because the Daily Weather Map publication only shows large scale features with limited mesoscale resolution, detection of some mesohigh events may have been inadvertently overlooked and classified as synoptic events. However, most mesohigh events which produced widespread heavy rains, instead of the typical isolated heavy rain events common during the summer, were examined which, because of their rarity, will be of better use to the QPF forecaster.

Of the thirty-three heavy rain events examined in this study, over half were classified as synoptic with eighteen total events: seven occurring during spring, seven during fall, and four during winter. Only seven events were classified as frontal: three in the winter and four in the spring. Five events were classified as mesohigh: two in the summer, two in the fall and one in the spring, and three events were classified as tropical: one in the summer (July) and two in the fall (September). All frontal events and the vast majority of synoptic events occurred during the CWA's wet season of November through May, while all but one mesohigh and tropical events occurred during the CWA's dry season of June through October. The final seasonal breakdown revealed that heavy rain events are possible throughout the year with twelve heavy rain events occurring during the spring, eleven during the fall, seven during the winter and three during the summer. As for extremely heavy rainfall events of six inches or greater, four occurred during the fall, three during the spring, two during the winter and one during the summer. Four were synoptic events, three were mesohigh events, two were frontal events and one was a tropical event.

Surface dewpoint temperatures for each event were also evaluated using the highest 6 am CST dewpoint temperature at the Memphis station compiled from each series of Daily Weather Maps which comprised a single event. While this method doesn't necessarily always reveal the highest dewpoint temperature before the start of a heavy rain event, especially when there is significant moisture advection, it is the only feasible way to estimate dewpoint temperatures from the Daily Weather Maps and is assumed to be fairly representative in most cases. Using this method, it was discovered that over 90% of the heavy rain events occurred with a dewpoint temperature between 50 and 75 degrees Fahrenheit. The range of dewpoints during the winter and early spring (including March) was from 40 to 65 degrees with over three-fourths occurring between 50 and 59 degrees. During the late spring (April and May), all events occurred with a dewpoint above 60 degrees with half occurring with a dewpoint of 70 degrees or above. All events during the summer had dewpoints of 70 degrees or above. During the fall, dewpoints ranged from 55 to 75 degrees with more than half of the events occurring with a dewpoint above 65 degrees.

Maddox et.al. also found that most flash flooding events occurred at night, particularly with frontal and mesohigh events. In this study, a heavy rain event was considered nocturnal if the starting time of rainfall occurred between 7 pm and 7 am local time during the spring and fall, 5 pm and 7 am during the winter, and 8 pm and 6 am during the summer, which is roughly the average time of sunset and sunrise throughout these seasons. With this criteria, around a third of the heavy rain events in this study were found to be exclusively nocturnal in nature. Of the thirteen total nocturnal events, seven were synoptic in nature with four being frontal, and two being mesohigh in nature.

7. CONCLUSIONS

The spatial distribution of rainfall across the Memphis CWA is generally from the northwest to the southeast, ranging from 48 inches across sections of northeast Arkansas and the Missouri bootheel to over 58 inches in northeast Mississippi on an annual basis. This general pattern of rainfall is likely the result of developing low pressure systems over the Gulf of Mexico creating an increase in rainfall across the CWA as they move northeast, on average, and also the idea that most cold fronts are oriented southwest to northeast as they cross the Memphis CWA. As these fronts move from west to east, on average, the available moisture, from the Gulf of Mexico to the south, would have more time to be pulled northward thus revealing a more northern extent as one traveled east. Although variations in topography are minimal across the Memphis CWA, there does seem to be a slight rise from the southwest to the northeast which might have a slight affect on the annual rainfall distribution, especially with summer convective activity as a slight maxima of rainfall is observed across west Tennessee and northern Mississippi during the summer months.

It was determined that the 'wet season' for the Memphis CWA can be considered from November through May with frontal and low pressure systems being the dominate heavy rain producing systems during these months. The 'dry season' can be considered from June through October with mesohigh and tropical systems being the dominate heavy rain producing events during these months. During the dry season, there is an afternoon maximum in frequency of rainfall which is most likely the result of thunderstorms caused by solar heating. The maximum in rainfall frequency during the wet season was found to be slightly greater at night which could be the result of destabilization effects from the radiation budget near the tops of middle and high clouds, the diurnal cycle in boundary layer wind speeds and the evolution of mesoscale pressure systems generated by convective activity (Hoxit et.al., 1978).

A heavy rain event can be defined for the Memphis CWA as one that produces rainfall of three inches or greater in a twelve hour period or less at two or more selected stations, or six inches or more in a twelve hour period or less at a single station. Although there is a minimum during the summer, heavy rain events are possible in all seasons throughout the year across the Memphis CWA. Most heavy rain events in this study were synoptic in nature, as defined by Maddox et.al., with roughly a third of all events being exclusively nocturnal, as defined in this study. One common feature for the vast majority of heavy rain events in the Memphis CWA was a southwest flow at 500 mb. Also, most heavy rain events during the year occurred with dewpoint temperatures above 60 degrees with 50 degrees being the minimum in winter for most cases.

For forecasters that have been in the local area for some time, much of this information is likely already known. However, new forecasters, particularly those from other parts of the country, may not be as familiar with this information. It is important for these new forecasters to quickly become knowledgeable of the local rainfall climatology. By having this knowledge, they will be able to issue better QPFs which in turn will make for better river forecasts which can be critical in times of major flooding.

ACKNOWLEDGEMENTS

The authors would like to thank Jack Jackson and Paul Close at NWSFO Memphis for their assistance in the gathering and manipulating of the hourly precipitation data, Brian Walawender at NWSFO Topeka for his computer assistance in the composition of the Memphis rainfall frequency charts, Jamie Kousky of the Climatic Prediction Center for her retrieval of numerous Daily Weather Maps, Buzz Merchlewitz (NWSFO Memphis hydrologist) for his hydrological insight, and Julie Shinko at the LMRFC for her assistance in the retrieval of the river basin map used in this paper.

REFERENCES

Hoxit, L.R., R.A. Maddox and C.F. Chappell, 1978: On the Nocturnal Maximum of Flash Floods in the Central and Eastern U.S. Preprints, Conference on Weather Forecasting and Analysis and Aviation Meteorology (Silver Springs), AMS, Boston, pp. 52-57.

Maddox, R. A., C. F. Chappell, and L. R. Hoxit, 1979: Synoptic and meso-a scale aspects of flash flood events. Bull. Amer. Meteor. Soc., 60, No. 2.

NOAA, 1961-1990: Daily Weather Maps-Weekly Series. U.S. Government Printing Office, Washington D.C.

NOAA, 1961-1990: Hourly Precipitation Data for Tennessee, Mississippi and Arkansas. National Climatic Data Center, EDIS, Federal Building, Asheville, NC.

NOAA, 1992: Monthly Station Normals of Temperature, Precipitation, and Heating and Cooling Degree Days 1961-1990. Climatography of the United States Series No. 81, National Climatic Data Center, EDIS, Federal Building, Asheville, NC.

NOAA, 1993: Solar and Meteorological Surface Observation Network 1961-1990 (CD-ROM). Version 1.0, National Climatic Data Center, EDIS, Federal Building, Asheville, NC.

Weather Bureau, 1961: Rainfall Atlas of the United States. Technical Paper #40, Hydrological Services Branch, Washington D.C.