Dr. J. Scott Greene
Oklahoma Climate Survey & Environmental
Verification & Analysis Center
University of Oklahoma
Norman, OK 73019
Dr. Laurence S. Kalkstein
Synoptic Climatology Laboratory
Center for Climatic Research
Department of Geography
University of Delaware
Figures and Tables not currently available for the Web.
Results indicate that the air mass patterns identified are both spatially and temporally consistent. Maps of air mass frequencies and afternoon temperatures and dew points illustrate the modification rates of each air mass.
As an example of the environmental applicability of the SSC, the impact of climate upon human mortality is analyzed at three climatically different locations (New Orleans, Memphis, and Chicago). Results show how the day-to-day mortality fluctuations are sensitive to air mass type. A particular air mass which is associated with distinctly high mortality is identified for Memphis and Chicago; no such air mass is apparent for New Orleans, where the weather/mortality signal is very weak.
The types of synoptic events identified by existing classifications can be loosely described as either "weather types" or "air masses". Weather types are synoptic events categorized by pressure patterns and wind fields; the resulting categories represent distinct flow regimes which can influence environmental factors such as pollution concentrations (e.g., Davis and Walker, 1992). An air mass is a large volume of air, "...acquiring characteristics of temperature and humidity related to the condition of the sea, land or ice beneath it" (Crowe, 1971).
Weather type classifications are effective in studies which analyze atmospheric transport mechanisms. For example, pollution dispersal and atmospherically-forced insect migration patterns have been conveniently analyzed using weather typing approaches. However, they are less useful in those studies for which thermodynamically homogeneous groups are required. If the surface or upper level flow are comparable on different days, it is probable that the thermal and moisture characteristics of these days would also be similar; however, there are notable exceptions. For example, consider an anticyclone located over the Great Plains in summer. This system would be supported by negative vorticity advection aloft, and could result in a variety of distinct meteorological conditions at the surface. One possible result would be cool, dry air which has been transported into the region. Under these conditions, skies would be clear with afternoon temperatures below 25 C. Another result might be modified Pacific air, which is warmed adiabatically on the lee side of the Rockies, and possesses afternoon temperatures around 30 C in this area. An additional possibility is the intrusion of dry, warm air from the desert Southwest, which would result in afternoon temperatures above 35 C. Thus, for applications where the thermal and moisture characteristics of the air are especially important, weather typing may be less desirable because of this meteorological heterogeneity.
Air mass approaches typically classify days based on a wider variety of weather elements, such as cloud cover, visibility, and precipitation, as well as various thermal and atmospheric moisture variables. Therefore, these classifications are distinctly less dependent upon pressure patterns and wind fields. The classic description of an air mass as defined by Bergeron (1930) identified polar, tropical, continental, and maritime air masses according to their source regions. However, the goal of this study is to characterize the different air masses based upon differences in thermodynamic character, as environmental responses to air masses are dependent on the meteorological character of the air at a place at a given time. Thus, source region determination is not as important as the identification of "umbrellas of air" which traverse a region and possess distinctive meteorological characteristics. The criteria of the Spatial Synoptic Classification (SSC) described here is for categorization of thermal and moisture variables, which ultimately influence the activity of a variety of environmental parameters.
Some synoptic classifications have been used as point indexes, such as the weather-typing approach adopted by Muller (1977), or the Temporal Synoptic Index (TSI; Kalkstein et al.,1987). These indices have proven useful in examining a range of environmental variables. However, expanding either of these point indices would be very time-consuming and unwieldy from an analytical standpoint (Kalkstein, et al., 1995). On a larger scale, several regional approaches have described weather types or air masses across an area of several thousand kilometers squared. For example, Schwartz (1991) evaluated July air mass frequencies at 15 sites in the north central U.S. He successfully identified five air masses by incorporating the "partial collective" method developed by Bryson (1966). While numerous point and regional classifications have been accurately used to identify weather types and air masses, continental-scale air mass classifications have been more difficult to devise. Davis and Kalkstein (1990) applied procedures similar to those of the TSI to develop a continental-scale classification (over the contiguous U.S.) for 1984. This classification was used to analyze air quality variations; however, the procedure suffers from important shortcomings. First, the technique must be developed on a day-to-day basis, which can be exceedingly time-consuming when treating a dataset of many years. Second, there were considerable problems tracking a given air mass across the country, especially if modification occurred during transport. Third, the procedure produced an unwieldy number of synoptic types which varied both seasonally and regionally.
Despite recent advances in the classification of synoptic-scale events, particularly at the point and regional scales, there remains the need for development of a simpler, automated, large-scale air mass-based procedure. A new method of analysis to identify resident air masses on a national scale has recently been developed (Kalkstein et al., 1995; Greene 1994). This procedure is less complex than previous classifications, yet permits inter-site comparison of daily air masses across a large region. The resulting spatial air mass inventory could have important environmental applications, such as the determination of climatic limiting factors for certain biotic communities, the impact of temporary surface features, such as snow, on air mass modification rates, and an evaluation of precipitation characteristics for individual air masses. This paper will describe results derived from an air mass classification performed at 126 first-order weather stations east of the Rocky Mountains during a 30 year period (1961 through 1990) for summer (June, July, and August), and give an illustrative example of the classification's usefulness in analyzing environmental problems.
Air Mass Identification
Unlike existing classification techniques, the new procedure requires initial identification of the meteorological characteristics of typical synoptic conditions at each station. Based on a detailed analysis of historical weather maps, six synoptic situations (or air masses) which are present in the contiguous United States have been identified. The term air mass is used here to indicate homogeneous, holistic weather categories based on meteorological characteristics. The six air mass types over the eastern United States are:
The physical character of the six air masses in summer can be described as follows. Dry polar (loosely synonomous with continental polar) is relatively cool and dry, with generally clear skies and little chance of convective precipitation. High pressure, typically found within an anticyclone of Canadian origin, is most frequently associated with dry polar air.
Dry temperate air is generally associated with zonal flow aloft, which permits moist Pacific Ocean air to dry adiabatically as it descends the Rockies. This air mass brings mild, dry conditions to the Midwestern and eastern United States due to adiabatic warming, and is analogous to Pacific air identified by Schwartz (1991) and others. In the southeastern United States, this air mass can also be produced when polar air is transported around a surface anticyclone and modified by a long trajectory over the Atlantic Ocean. This subset of dry temperate air has been identified as "East Coast" air by Kalkstein (1972).
Dry tropical air is often associated with the hottest and driest conditions, as well as clear skies. It typically originates in the Desert Southwest or northern Mexico, and frequently intrudes into the southern Plains around the western edge of an anticyclone. In addition to dry desert air advected from the Southwest, this air mass includes chinook events in the Northwestern portions of the study region. The violent, dry downsloping of chinooks produce conditions that, at the surface, as similar to desert air. Hence, areas of northern Colorado and Wyoming can occasionally experience as much dry tropical air as western Texas, even during summer.
Moist polar air is associated with generally overcast conditions, cool temperatures, a small dew point depression and, frequently, easterly flow at the surface. This can occur when a front is located well to the south, producing an overrunning situation. However, in the eastern portion of the study region, MP is often advected from the Atlantic around the northern flank of a mid-latitude cyclone. Both of these processes result in virtually identical surface expressions; thus, these conditions are grouped together.
Moist temperate can also be associated with frontal activity, and is thus characterized by many of the same elements as moist polar. However, the temperature and dew point are considerably higher for moist temperate, and the responsible front is typically located much closer to the region. Both moist polar and moist temperate conditions can persist for some time if a quasi-stationary front is located nearby.
Moist tropical air, commonly recognized as maritime tropical, is warm and moist, as it generally originates from the Gulf of Mexico or tropical oceans. This air is found in the warm sector of a mid-latitude cyclone, or on the western flank of a subtropical anticyclone. Of the six types, convective activity is most common to this air mass, particularly during summer.
The Spatial Synoptic Classification
The Spatial Synoptic Classification (SSC) was developed to classify all days within one of these predetermined, readily identifiable, air mass categories. Each day is classified using "seed" days and linear discriminant function analysis. What follows is a brief description of the procedure; for greater detail, refer to Kalkstein, et al. (1995) or Greene (1994).
The procedure relies on the input of groups of "seed days" for each air mass at each locale (Figure 1; from Kalkstein, et al., 1995). Seed days represent days possessing the typical meteorological character of each air mass at a location and are used to classify all other days. Each group of seeds is chosen by the specification of ranges in afternoon surface temperature, dew point, dew point depression, wind speed, wind direction, cloud cover, diurnal temperature change, and four six-hourly dew point changes. Afternoon meteorological observations are most useful for selecting seed days because it is at this time that air masses are most physically distinct from one another. Narrow ranges in five trend variables are used to prevent those days which have undergone significant changes (such as frontal passages which represent air mass transition days) from being selected as seed days. Initial estimates of these criteria are specified for each air mass at each location after careful evaluation of surface meteorological data and maps, and days which satisfy them are selected to represent that air mass. The criteria are then iteratively adjusted until all seeds selected comprise a homogeneous representation of the air mass.
Following seed day selection, discriminant function analysis is used to generate a linear function for each air mass from its group of seed days. Since the goal is to classify each day into one of the pre-existing air masses described above, and to use the seed day means as input into the classification, discriminant function analysis was selected as the appropriate classification tool (James, 1985; Klecka, 1980). Discriminant analysis is similar to the use of multiple regression. For each air mass type, a discriminant function is developed based upon the means of the seed day groups. Each day in the period of record is subsequently evaluated using all air mass discriminant functions to determine which group it most closely resembles. The day is classified into the category possessing the highest score. The result of this evaluation is a calendar listing the air mass to which each day has been assigned.
While a majority of these air mass designations are correct, a significant number of days may be incorrectly classified because of the occurrence of a transition between air masses. In order to account for this possibility, the SSC performs a second discriminant function analysis to determine whether or not a transition between air masses occurred during each day. Two groups of seed days are used for this analysis: the group of seed days representing transitions between air masses (these contain substantial trends in meteorological character through the day), and a group containing the combined seed days of all air masses at that location. This discriminant analysis produces a second calendar which designates each day as either transitional or non-transitional. All days classified as non-transitional are placed into the category assigned them by the initial discriminant analysis. All other days are classified as transitions.
Using the SSC, days were classified into one the six air masses described above, and air mass frequencies over summer were examined (Figures 2). The least frequently occurring air mass in the summer is dry polar (Figure 2a). The frequencies range from over 15 percent in northern Minnesota and North Dakota, to less than 1 percent along the Gulf Coast. Dry polar air never occurred in southern Florida during the period of study. These values offer an interesting contrast to DP winter frequencies (December, January, and February) using SSC (Kalkstein et al., 1995). Winter frequencies in the north were at least twice as high, approaching 30 percent in the extreme north central and northeastern U.S. The proportional disparity between summer and winter frequencies is even greater toward the Gulf Coast, where DP occurred about 10 percent of the time in winter.
The other two dry air masses have a more significant presence during summer. In the East, dry temperate air exhibits a north-south decrease in frequency, ranging from about 30 percent in New England to about 5 percent in southern Florida (Figure 2b). Trends were somewhat the opposite during winter, where greatest frequencies of DT air along the East Coast were found in Georgia and South Carolina. Toward the western boundary of the study region, the summer frequency pattern shifts, and there is a more pronounced west-east frequency distribution over the Great Plains. Much of the DM air in this area originates from the Pacific, and moves east when the upper-level circulation exhibits a zonal pattern. This type of pattern occurs regularly in the summer; consequently, the highest frequency of DM air is just to the lee of the Rocky Mountains, with values near 50 percent in eastern Colorado and Wyoming. This is similar to the winter DM pattern in this region, although summer frequencies are generally higher, especially over the Northern Plains.
Not surprisingly, DT experiences a marked decrease rapidly away the Desert Southwest (Figure 2c). Unlike the other air masses, there is a consistent west-east decrease in frequency across the study area. The values range from 25 percent in western Texas to near zero in the Southeast. Although infrequent throughout the eastern U.S., there were a few incursions of the air mass in the Northeast, usually around the northern flank of a deep ridge when it was centered over the Southeast. Many of the stations in the Northeast (e.g., New York, Philadelphia) exhibited a small frequency of DT (less 1 percent), but only during extreme summers such as 1988 and 1995. This never occurred in the Southeast, where the deep ridge prevents the incursion of hot, dry desert air. In addition to dry desert air advected from the Southwest, this air mass encompasses chinook events in the northwestern portions of the study region. The violent, dry downsloping produces conditions that, at the surface, are almost indistinguishable from desert air. Hence, areas of northern Colorado and Wyoming exhibit DT frequencies of about 20 percent during summer. Winter DT frequencies are much less, and only exceed 10 percent in West Texas.
Moist Polar is relatively rare in summer, and increases slightly toward the north (Figure 2d). While most of the days associated with frontal conditions in the summer are classified as MM, a few of the coolest days (often associated with the strongest anticyclones) are classified as MP. This air mass probably exhibits the greatest frequency differential between seasons. MP frequencies are over 40 percent during winter in a large area from western New York to eastern Wisconsin, and over 30 percent in much of the northern third of the country. This contrasts strongly to the 4-7 percent frequency of this air mass over the same region in summer.
Moist Temperate is also not predominant in summer, but does occur at frequencies of 15 to 20 percent across much of the U.S. east of the Mississippi River (Figure 2e). There is a steep decline in the frequency of this air mass across Texas due to the dominance of dry DT and DM air in this area, and no MM air was found in western Texas or in eastern Colorado. In the winter, the frequency maximum for MM is along the Gulf Coast (Kalkstein, et al., 1995), where stationary fronts often align. In the summer, however, the maximum shifts north into Tennessee, southwestern Virginia, and western North Carolina. Some of this might be orographic, as cloud cover increases along the western flank of the Appalachians during southwesterly flow situations might lower temperatures and decrease diurnal range. Convective activity maxima (and concurrent cloud cover) around the central Gulf Coast might also be responsible for the higher MM frequencies from Baton Rouge to Mobile. In any case, the summer MM frequency pattern is interesting, and requires further evaluation.
The most dominant air mass in summer, particularly in the southeastern United States, is MT (Figure 2f). Perhaps the most dramatic feature of this air mass is the sharp decrease in frequency across Texas, which is associated with a concomitant increase of dry air from the Desert Southwest. Occurrences of MT range from over two out of three days from Corpus Christi to Brownsville to zero in El Paso. The entire Southeast is dominated by this air mass, with maximum frequencies found over the Florida peninsula. The frequencies decrease toward the north, and less than 20 percent of the days are classified as MT north of a line from Boston to Detroit to Minneapolis. Although frequency orientations are similar in winter, actual values are obviously much less. The winter 20 percent frequency line roughly parallels the Gulf Coast.
Afternoon temperature and Dew Point The temperature gradients of DP and DM show a general north-south orientation (Figures 3a and b), with a slight dip along the western edge of the Appalachians in western Pennsylvania and West Virginia. The highest temperatures for both air masses are found in Texas, and drier conditions in the Great Plains seem to contribute to higher maximums than further east. The summer temperature gradient for both air masses is much weaker than that for winter. For example, DP mean afternoon temperatures increase by over 24oC from northern Minnesota to southern Texas in winter (Kalkstein et al., 1995). This value is about 8oC in summer. DM maximum temperatures vary even less spatially; there is only about a 6oC differential between highest and lowest values throughout the study region.
The thermal character of DT shows that it modifies only slightly as it traverses the region in summer (Figure 3c). The values range from near 40 C in southwestern Texas and Oklahoma to 35 C in the Northeast. The pattern shows a southwest-northeast trend, suggesting the direction of modification as the air flows out of the southwestern deserts. The mode in central Oklahoma is an interesting phenomenon; it confirms previous research indicating that this area possesses the most sultry summer climate in the United States because of unusually high apparent temperatures when DT and MT air masses are present (Kalkstein and Valimont, 1986).
The moist air masses display even smaller temperature gradients in summer, owing to their higher specific heat (Figures 3d, e, and f). During summer, the coolest air mass is actually MP, with average afternoon temperatures remaining below 20 C in the northern half of the study region. The MM and MT air masses have particularly small thermal gradients; the difference in mean afternoon temperature for both between Georgia and New York is only 2 C. Thus, these air masses barely modify as they traverse great distances in summer.
Dew point gradients are somewhat similar to temperature gradients for all six air masses, with a few interesting exceptions (Figures 4a through f). For example, afternoon dewpoint gradients for the three dry air masses have a more meridional component than temperature gradients. This is especially true for the DT air mass, where very low dew points are noted in the western portions of the study region. Thus, some significant modification takes place as this air mass approaches the eastern U.S. The moist air masses do not exhibit this pattern, and although some modification of MP is apparent, the gradient is north-south. MT shows a particularly small gradient; mean dew points vary by only about 5oC throughout the area.
The distinct hydrodynamic patterns displayed by the air masses are related primarily to the surfaces they must traverse during transport. For example, the relatively large west-east dew point gradient of DT is the result of a moist, vegetated cover that exists as the air mass moves eastward. This also explains the very small modification of the other air masses, as the generally homogeneous surface creates little opportunity for change. Of course, this is very different in winter, when surface conditions vary considerably, especially from north to south. Snow cover has been shown to be a major contributor to winter air mass modification rates (Kalkstein et al., 1995). In addition, significant latitudinal variations in solar radiation contribute to increased modification rates in winter.
The summer of 1995 is an example of the potential impact of weather on human health, particularly acute mortality. The most notable case is Chicago, where over 500 perished from heat-related causes (CDC, 1995). Deaths due to excessive heat were also reported in numerous U.S. cities, and across Europe (Kalkstein, 1995). While the definition of heat-related death varies from place to place, variations in heat-related mortality are also due to the dissimilar responses to similar weather patterns at different locations.
People respond to the total effect of all weather variables interacting simultaneously on the body. Therefore, an appropriate means to evaluate weather/health relationships is through the identification of high risk air masses that, when present, could negatively affect human health. Studies funded by the U.S. EPA and the Southern and Southeast Regional Climate Centers show very strong relationships between certain very hot, oppressive air masses and increased human mortality. For example, Kalkstein (1991) has shown that for many cities, it is apparent that one or two hot air masses possess a much higher mean mortality than the others, and these high risk air masses contain an inordinately high percentage of days with the greatest mortality totals. However, the use of the site-specific temporal synoptic index prohibited a comparison of the impacts of the same air masses at different cities. The SSC provides a unique ability to identify similar air masses across continent-sized regions. Thus, it is now possible to determine if human response to particular air masses is similar in differing climates. To illustrate this point, we examined daily acute mortality at three climatologically distinct cities (New Orleans, and Memphis, and Chicago) to determine if regional differences exist during similar weather conditions.
The mean daily mortality for each synoptic category, along with the standard deviation, was determined to ascertain whether particular categories exhibited distinctively high or low mortality values (Table 1). Potential lag times were accounted for by evaluating the daily synoptic category on the day of the deaths, as well as 1, 2, and three days before the day of the deaths. Daily mortality was also sorted from highest to lowest during the period of record to determine whether certain synoptic categories were prevalent during the highest and lowest mortality days for each of the cities.
Results show that strong regional differences do, in fact, exist. In New Orleans, no air mass seems to be associated with particularly high mean daily mortality totals. However, dry tropical air appears to have a significant impact on mortality at Memphis and Chicago. In addition, the moist tropical air mass in Chicago also contributes to elevated mortality. These results are consistent with other studies that suggest increased mortality sensitivity to weather in climates with great summer weather variations (IPCC, 1996; WHO/WMO/UNEP, 1996). However, this study is the first that permits direct comparison of similar air masses in different climates.
While some air masses possess a mean mortality well above the overall mean, not all days within this air mass type possess elevated mortality totals. Thus, it is important to not only identify the "high risk" air mass, but to determine the important variables within each air mass that will result in elevated mortality on certain days (Table 2). A regression procedure was employed to determine those parameters within the air mass which contribute to the highest daily mortality. Independent variables included meteorological (e.g. maximum temperature, cloud cover) and non-meteorological variables (e.g. number of consecutive days of the air mass, time of season) which may impact mortality (refer to WHO, WMO, UNEP, 1996 for a more detailed explanation of the regression procedure). In Chicago, the factors that contribute to high daily mortality when the DT air mass is present include the number of consecutive DT days and maximum temperature. Thus, for each consecutive DT day, mortality increases in Chicago by an average of 2.41. At Memphis, maximum temperature is the only statistically significant independent variable which explains variation in mortality during DT occurrences. For each degree C increase in maximum temperature, daily mortality increases in Memphis by an average of 1.94. The resulting algorithms satisfy the Box and Wetz (1973) criteria as statistically robust predictors, and can be used to estimate mortality for any given day within the high risk air mass provided that the meteorological data are available.
It is clear apparent that regional differences exist in the mortality/weather relationship. The same air mass can induce different human responses at different locales. Although MT air masses are associated with elevated mortality in Chicago, where it occurs on only about 25 percent of the days in an average summer, this air mass has little impact on the population in Memphis or New Orleans, where it occurs much more often. This differential is noted in spite of the fact that the air mass has similar characteristics at all three cities. Thus, the response of mortality to air mass type described here represents an example of the relative impact of weather on human well-being. MT represents a stressful condition in Chicago, where it is relatively rare, but produces little response in the southern cities, where some form of acclimatization is evident.
The SSC was used to identify occurrences of six air masses (dry polar, dry temperate, dry tropical, moist polar, moist temperate, and moist tropical) at 126 locations east of the Rocky Mountains for summer, 1961-1990. An analysis of results indicates that the air mass patterns identified by the SSC are both spatially and temporally consistent across the study region. Maps of air mass frequencies and afternoon temperatures and dew points illustrate the common trajectories and modification rates of each air mass.
The spatial and temporal consistency of the SSC make it applicable to environmental analyses, such as weather-health evaluations. The brief example shown here illustrates the inter-regional differences in human response to weather, and for the first time, permits impacts analysis of particular air masses across large regions. A more intensive study examining the summer mortality for 44 cities in the US using SSC is currently under way (Kalkstein, et al., in press).
Other health-related problems which are affected by atmospheric phenomena, such as asthma and allergy distress, are also well-suited to SSC analysis. Certain air mass types are associated with high concentrations of air pollutants, such as total suspended particulates and sulfates. With the availability of EPA-monitored air pollution data from numerous sites around the U.S., the SSC will permit continental-scale analyses of air masses which are responsible for increased asthma symptoms, and allow for the evaluation of weather/air pollution synergism on this and other health problems.
Figure 1: Procedure to select seed days. Figure 2: Air mass frequencies (percent for 1961-90). a. Dry Polar; b. Dry Temperate; c. Dry Tropical; d. Moist Polar; e. Moist Temperate; f. Moist Tropical. Points on the map represent all stations where a particular air mass was detected during the study period. Figure 3: Mean 1500 LST Temperature ( C) a. Dry Polar; b. Dry Temperate; c. Dry Tropical; d. Moist Polar; e. Moist Temperate; f. Moist Tropical. Points on the map represent all stations where a particular air mass was detected during the study period. Figure 4: Mean 1500 LST Dew Point ( C) a. Dry Polar; b. Dry Temperate; c. Dry Tropical; d. Moist Polar; e. Moist Temperate; f. Moist Tropical. Points on the map represent all stations where a particular air mass was detected during the study period.
Bergeron, T., 1930: Richlinien einer dynamischen klimatologie, Met. Zeit., 47, 246-262.
Box, G.E.P., and J. Wetz, 1973: Criteria for judging adequately of estimation by an approximating response function, University of Wisconsin Statistics Department Technical Report, #9, Madison, WI.
Bryson, R.A., 1966: Air masses, streamlines, and the boreal forest, Geographical Bulletin, 8, 228-269.
Centers for Disease Control, 1995: Heat related mortality - Chicago, July 1995, Morbidity
and Mortality Weekly Report, 44(31):577-579.
Crowe, P.R., 1971: Concepts in Climatology, St. Martin's press, New York, 589 pp.
Davis, R.E., and L.S. Kalkstein, 1990: Using a spatial synoptic climatological classification to assess changes in atmospheric pollution concentrations, Physical Geography, 11, 320-342.
Davis, R.E., and Walker, 1992: An upper air synoptic climatology of the western United
States, J. Clim., 5, 1449-1467.
Greene, J.S., 1994: Examination of precipitation variability and validation of satellite-based rainfall algorithm using a new spatial synoptic classification, Ph.D. Dissertation, University of Delaware, Newark, DE.
Intergovernmental Panel on Climate Change (IPCC) Impacts Assessment, 1996: Assessing the Health Impacts of Climate Change, Geneva, in press.
James, M., 1985: Classification Algorithms, John Wiley and Sons, New York, 210 pp.
Kalkstein, L.S., 1995: Lessons from a very hot summer, The Lancet, 346:857-859.
Kalkstein, L.S., 1991: A New Approach to Evaluate the Impact of Climate Upon Human Mortality, Environmental Health Perspectives, 96, 145-150.
Kalkstein, L.S., 1972: Spatial and Temporal Climatic Variations Along the Northern Gulf Coast by Means of Air Mass Analysis, Master's Thesis, Louisiana State University, Baton Rouge, 117 pp.
Kalkstein, L.S., C.D. Barthel, J.S. Greene, and M.C. Nichols. 1995: A New Spatial Synoptic Classification: Application to Air Mass Analysis, Int. J. Climotol., in press.
Kalkstein, L.S., G. Tan, and J.A. Skindlov, 1987: An evaluation of three clustering procedures for use in synoptic climatological classification, Journal of Climate and Applied Meteorology, 26, 717-730.
Kalkstein, L.S., and K.M. Valimont, 1986: An Evaluation of Summer Discomfort in the United States Using a Relative Climatological Index. Bulletin of the American Meteorological Society, 67(7), 842-848.
Klecka, W.R., 1980: Discriminant Analysis, Sage University Pres, Newbury Park, CA, 71 pp.
Muller, R.A., 1977: A Synoptic Climatology for Environmental Baseline Analysis: New
Orleans, J. Clim. Appl. Meteor., 24, 293-301.
Schwartz, M.D., 1991: An Integrated Approach to Air Mass Classification in the North
Central United States, The Professional Geographer, 43(1), 77-91.
WHO/WMO/UNEP, 1996: Climate Change and Human Health. Eds., A.J. McMichael, A.Haines, R. Slooff, and S. Kovats, Geneva, in press