Bruce Smith
National Weather Service
Gaylord, Michigan


Christopher Blough (1)
Central Michigan University
Mt. Pleasant, Michigan



Meteorologists in the Great Lakes region are presented with a wide range of forecast challenges. Much of this variability can be attributed to the orientation of the Great Lakes within the North American continent, coupled with the physical and thermal characteristics of the lakes themselves. While the Great Lakes region is perhaps best known for its harsh winters, consisting of heavy snowfall and cold temperatures, the region is not immune from such adverse weather conditions as summertime heat waves, severe thunderstorms, heavy rainfall, and tornadoes. Other weather patterns in this area, such as lake breezes, land breezes, and lake effect snow are more regional in scope, and are the result of complex interactions between air, water, and land in and around the Great Lakes.

Lake effect snow is perhaps the most unique and formidable of these regional weather concerns. Lake effect snow is most common in the fall and winter, when air moving across the Great Lakes is frequently much colder than the lakes themselves. When this occurs, the thermal instability which develops (as approximated by the magnitude of the temperature difference between the lake and the air) causes warmth and moisture from the lakes to be transferred into the lower troposphere. This process frequently leads to enhanced cloudiness and lake effect snow showers downwind of the lakes.

Part of what makes lake effect snow complex is the number of variables which must be assessed and accurately predicted in order to determine1) whether heavy snow will occur, and 2) the location of the heavy snow. Many of these variables are meteorological, including thermal instability, wind speed and direction, inversion height, moisture, large-scale lift, and cloud physics. Additional variables which influence snowfall intensity and location are associated with variations of the physical geography around the Great Lakes. These factors include fetch (distance traveled by air over water), shoreline orientation, and elevation. These geographic features are responsible for producing tremendous spatial variations in snowfall during lake effect snow events. It is not uncommon for snow accumulations to range from a few inches to more than a foot over a distance of only a few miles. The ultimate challenge facing forecasters is to understand how each of these elements interact with one another to produce localized, heavy snowfall.

Figure 1 shows the forecast area served by the National Weather Service (NWS) in Gaylord, Michigan. This area consists of 25 counties in northern Lower and eastern Upper Michigan, and also borders three of the five Great Lakes (Lake Superior, Lake Michigan, and Lake Huron). Not surprisingly, the NWS Gaylord forecast area receives abundant lake effect snowfall. Since lake effect snow episodes are generally associated with cold air outbreaks from the west or north, the associated snowfall is typically deposited on the east and/or south shores of the Great Lakes (i.e., downwind). Figure 2 shows a plot of mean annual snowfall for the NWS Gaylord forecast area. The locations of the maximum snowfall in this figure (to the lee of the lakes with a prevailing west or north wind) help to demonstrate the influence of the lakes on mean annual snowfall.

Figure 1


Figure 1. The area shaded blue represents the 25 counties served by the National Weather Service in Gaylord, Michigan.


As noted, elevation plays a key role in the development of lake effect snow. This is due to the fact that upward vertical velocities and snowfall intensities increase as air is forced to rise over higher terrain. The connection between topography and precipitation has been previously well documented (Muller 1966, Hill 1971, Reinking and Boatman 1986, Niziol 1989, Houze 1993). Hill (1971) noted a strong correlation between annual snowfall and elevation. He found that for every 100 foot increase in elevation, an increase of approximately 8 to 12 inches of mean annual snowfall could be expected. Figure 3 shows the topography of northern Michigan. Though the highest elevation in northern Lower Michigan is only about 850 feet above the level of Lake Michigan, Figure 2 and Figure 3 suggest that this relatively small change in elevation is still sufficient to increase mean annual snowfall by about 60 inches per year across parts of northwest Lower Michigan.

Figure 2 Figure 2. Mean annual snowfall (inches) for the Gaylord, Michigan forecast area. Light blue shading represents snowfall between 100 to 120 inches. Dark blue shading represents snowfall in excess of 120 inches.

Figure 3 Figure 3. Elevation above mean sea-level (feet) for the Gaylord, Michigan forecast area. The average lake levels for each of the Great Lakes are also depicted.


This local variation in mean annual snowfall is one of the primary motivations for conducting this study. Our goal is to compute and graphically show vertical velocity over northern Michigan as a function of topography and wind direction using an objective Geographic Information System (GIS). Such plots are beneficial, because they help quantify the impact the higher elevations of northern Michigan have on vertical air motions, snowfall intensities, and mean annual snowfall. This is useful to operational forecasters because they are able to graphically see where vertical lift resulting from elevation changes would be greatest, and subsequently where the highest snowfall intensities would be most likely during lake effect snow events.


A. Integration of Elevation Data with GIS

GIS techniques were employed in this study to determine where topography-induced vertical velocities would most likely lead to the enhancement of lake effect snow. Though meteorologists frequently make use of maps, the concept of a GIS is likely unfamiliar to many. The benefit of a GIS is that it combines the functionality of a database with the spatial attributes of a map, such that it provides an association between nonspatial information and the location of a feature. GIS links basic map features (including points, lines, and areas) to database records which contain attribute information relating to these spatial features. Additional information on GIS can be found in Berry (1995), Jensen (1996), and Burrough and McDonnell (1998).

The GIS attribute information incorporated into this study was a grid of elevation points called a Digital Elevation Model (DEM). A DEM is a series of regularly spaced points of elevation across a particular area derived from detailed topographic surveys compiled by the United States Geological Survey. The fact that every elevation grid point in a DEM is referenced to a precise latitude and longitude makes it very appropriate for use with a GIS.

The grid point spacing within each raw DEM was initially about 215 feet in the east-west direction, and 303 feet in the north-south direction. It was determined that this relatively narrow grid spacing would provide more detail than required for an operational forecaster, and that an equally spaced and somewhat coarser grid spacing in each direction would be desirable. As a result, the database was modified (or "resampled") such that the horizontal grid spacing in each direction became a consistent 2500 foot interval. This modification yielded results that were appropriately detailed and operationally more interpretable. For ease of computation and display, the twelve DEMs required for full coverage of northern Michigan were merged to create a single, seamless grid covering the entire area. The development of such a grid of appropriately spaced elevation points was necessary to calculate vertical velocity as a function of wind direction.

B. Vertical Velocity Calculation

In the atmosphere, vertical velocity can be the result of numerous processes. Some vertical velocities are driven by thermal instabilities, such as surface heating or horizontal temperature advection. Other vertical velocities are driven by atmospheric dynamics, such as large-scale trough-ridge interactions or jet streaks. This study quantifies the vertical velocity which develops when the horizontal wind at the surface is forced to ascend as it encounters changing topography.

The vertical velocity produced when the horizontal surface wind impinges upon enhanced terrain is a function of two variables: 1) the wind speed at the surface, and 2) the slope of the terrain in the direction the wind is blowing. The vertical velocity will tend to be greatest when the wind speed is high, and/or when the terrain is steeply sloped. The vertical wind component near the surface wo is given by the equation:


Although lake effect snow is most often associated with flow regimes from the west and/or north, it can occur with any wind direction. Consequently, vertical velocities were calculated for 16 different wind directions, one for each of the "primary" compass direction points. For conciseness, however, only 8 of the 16 vertical velocity maps will be presented in this paper.


The eight vertical velocity maps are shown in Figures 4 through 11, for north, northeast, east, southeast, south, southwest, west, and northwest wind directions, respectively. On each of these maps, elevations greater than 1000 ft above mean sea-level are shaded green. The red contours show vertical velocity every 2 cm/s (starting at 3 cm/s). The shaded red regions depict upward vertical velocities of at least 7 cm/s. Vertical velocities of this magnitude fall considerably short of those associated with summertime thunderstorms (which can be tens of meters per second). However, vertical velocities on the order of several centimeters per second are not unlike those found within strong mid latitude synoptic-scale storm systems.


Figure 4
Figure 4. Upward vertical velocity produced when a surface wind of 7 m/s from the north moves across the terrain of northern Michigan. Red contours show upward vertical velocity every 2 cm/s (starting at 3 cm/s). Red shaded regions depict upward vertical velocities of at least 7 cm/s. Green shaded regions show elevations exceeding 1000 ft above mean sea-level.

Figure 5Figure 5. Same as Figure 4, except for a surface wind of 7 m/s from the northeast.

Figure 6Figure 6. Same as Figure 4, except for a surface wind of 7 m/s from the east.

Figure 7 Figure 7. Same as Figure 4, except for a surface wind of 7 m/s from the southeast.

Figure 8Figure 8. Same as Figure 4, except for a surface wind of 7 m/s from the south.

Figure 9Figure 9. Same as Figure 4, except for a surface wind of 7 m/s from the southwest.

Figure 10Figure 10. Same as Figure 4, except for a surface wind of 7 m/s from the west.

Figure 11Figure 11. Same as Figure 4, except for a surface wind of 7 m/s from the northwest.

These figures show that the resultant maximum vertical velocity is not only a function of the overall elevation change, but also the angle at which the surface wind strikes the slope. The greater the slope and/or the more perpendicular the wind direction is to the slope face, the greater the resulting vertical velocity. For example, a comparison of a north wind regime (Figure 4) and a northwest wind regime (Figure 11) reveals that the area of at least 7 cm/s vertical velocity is more expansive across northwest Lower Michigan when surface winds impinge on the higher terrain from the northwest, as opposed to north. These figures indicate that while the overall change in elevation is similar with both wind directions, the area of abrupt elevation change is smaller for the north wind regime. As a result, the size of the maximum vertical velocity area is reduced. This example highlights how an operational forecaster would be able to utilize these maps during a lake effect snow event to anticipate the areal coverage of highest upward vertical velocities and snowfall intensities.

While these maps were developed primarily for use during lake effect snow events, they also have application to any weather situation where subtle changes in low level vertical velocity resulting from the interaction between wind and elevation plays a role. Two such examples include the development/dissipation of low clouds (including fog), and the initiation of thunderstorms.


Lake effect snow represents a significant weather forecast challenge in northern Michigan. Forced ascent resulting from elevation is one of the parameters which helps to determine the location of the most intense lake effect snow. The goal of this study was to use a Geographic Information System (GIS) to objectively compute vertical velocity as a function of elevation and wind direction. From these results operational forecasters are able to graphically see where upward vertical velocities resulting from elevation changes are greatest. This information helps an operational forecaster to more accurately predict the location of high snowfall intensities

Maps depicting vertical velocity resulting from elevation changes for eight different wind directions are presented in this paper. The maps show that the resultant vertical velocity is not only a function of the overall elevation change, but also the angle at which the surface wind impings upon the slope. Our hope is that these maps allow forecasters to more accurately isolate areas likely to receive the intense lake effect snowfall, and that this detailed information can be incorporated into subsequent forecasts.


Berry, J.K., 1995: Where is GIS? Earth Observation Magazine, 29-32. Burrough, P.A., and R.A. McDonnell, 1998: Principles of Geographical Information Systems. Oxford University Press, 36pp.

Hill, J.D., 1971: Snow squalls in the lee of Lake Erie and Ontario. NOAA Technical Memorandum NWS ER-43. DOC, NOAA, NWS, Easter Region Headquarters, Scientific Services Division, Bohemia, NY, 20pp.

Houze, R.A., 1993: Cloud Dynamics. Academic Press, 573pp.  Jensen, J.R., 1996: Introductory Digital Image Processing.Prentice-Hall, Inc., 316pp.

Muller, R.A., 1966: Snowbelts of the Great Lakes. Weatherwise, 19, 249-255.

Niziol, T.A., 1989: Some Synoptic and Mesoscale Interactions in a Lake Effect Snowstorm. Postprints, Second Natl. Winter Weather Workshop. Raleigh, NC, 260-269.

Reinking, R.F., and J.F. Boatman, 1986: Upslope Precipitation Events Mesoscale Meteorology and Forecasting, P.S. Ray, Ed., AMS, 437-471.







1. Current affiliation Wade-Trim Inc., Gaylord, Michigan is the U.S. government's official web portal to all federal, state and local government web resources and services.