Comment: This task report essentially addresses phase two of the project, which is the development of snow climatology and associated statistical descriptions and depictions from the complete database. Members of the Technical Advisory Panel (TAP), along with University of Minnesota staff, participated in a half-day training session to understand the design criteria utilized in deploying snow fences for the control of blowing and drifting snow. This training session was conducted by Dan Gullickson of Mn/DOT and Paul Flynn of the USDA/NRCS. Following the training session, and several discussions with TAP members, there was consensus on a list of deliverables related to the analysis of snow climatology. These deliverables fell into four broad categories (and are listed in detail in Table 1): (1) characteristics of the snow accumulation season (as defined by the Tabler method); (2) characteristics of snowfall frequency, including means and percentile rankings for various time periods and design criteria thresholds (following the precedent set in the MN/DOT and University of Minnesota Extension Service Guidebook "Catching the Snow with Living Snow Fences," 1999); (3) characteristics of snow water equivalent (density of freshly fallen snow) for each month of the snow season; and (4) wind speed and direction analysis in the form of wind roses for all available National Weather Service automated stations in Minnesota. It was further agreed that summaries would be presented in the form of maps and tables, with written descriptions of the database and how the statistical descriptors were derived. In addition, a map of MN/DOT district boundaries is provided for users to reference. These products were to be delivered in the form of a CD so that mapped details of the analysis in color format would be more easily interpreted.
Disclaimer: The Minnesota Department of Transportation (MN/DOT), in cooperation with the University of Minnesota (UMN) and the Minnesota Department of Natural Resources – State Climatology Office (SCO) provide these data, and the associated maps, tables, and analyses for informational purposes only. As with most government services, these data exist in the public domain, but users are encouraged to pay special attention to the conditions and provisions under which the data were collected, as there were some variations in method of observation. The user assumes the entire risk related to use of this database. Data are provided on an "as is" basis and MN/DOT, UMN, and the SCO disclaims any and all warranties, whether express or implied, including (without limitation) any implied warranties of merchantability or fitness for a particular purpose. In no event will MN/DOT, UMN, or the SCO be liable to users or to any third party for any direct, indirect, incidental, consequential, special or exemplary damages or lost profit resulting from any use or misuse of this data.
The Tabler method to calculate snow accumulation season (SAS) is a temperature dependent function (Tabler, 1994) based on the time when the average air temperature falls to the freezing point (32° F) in the fall, remains below freezing, and rises above it in the spring. The beginning and ending dates are computed from monthly mean temperatures and yield an onset date and an end date. For this analysis, mean monthly temperatures were derived from the mean daily temperatures (which is obtained from the average of the daily maximum and minimum temperatures). The time of observation for the daily maximum and minimum temperature records are not uniform for the entire database and/or may not be consistent with the recommended time. Because daily temperature data can be biased by the time of observation, it is necessary to apply a correction factor. The greatest spatial resolution was achieved using a time-corrected temperature data set for the period 1971-1996. With this data set, 10-km resolution grid spacing, derived from kriging techniques in the mapping software, was utilized to map onset and end dates for the snow accumulation season. Climatologically speaking, the data and maps represent the best possible spatial resolution of this important parameter. Onset of the snow accumulation season was found to range from October 30th (in northeastern St. Louis County) to November 20th in some southern counties. The end date for the snow accumulation season ranged from March 15th in some southern counties to April 4th in northeastern ST Louis County. This gives a range of 155 days in the SAS for northern sectors of Minnesota and 115 days for southern portions of the state.
The mean seasonal and annual snowfall data were summarized based on three distinct time periods: the Tabler-based snow accumulation season (SAS); the seasonal snowfall based on the October through March total; and the total annual amounts (July-June) regardless of season. The latter technique allowed for the accumulation and averaging of all available historical snowfall data at each location, including the anomalous, but often significant, early fall or late spring snowfalls. The frequency statistics for the snow accumulation season (SAS) and for the October through March snow season are based entirely on a fixed calendar sequence and therefore do not include the more unusual snowfall events that occur in the early fall or late spring. A further stratified view of the data sets was achieved by examining different historical periods: a 110-year period (1890-91 to 1999-2000), a 90-year period (1911-12 to 1999-2000), a 60- year period (1941-42 to 1999-2000), and a 30-year period (1971-72 to 1999-2000). These multiple views of the data were depicted in a series of maps showing an analysis based on snowfall isopleths (lines of equal value). As might be expected, the mean snowfall for the snow accumulation season (SAS) shows the narrowest range of values across the state, while the mean annual snowfall shows the widest range of values, with the fixed calendar-based seasonal period (October through March) showing a range of values intermediate to these other periods. The statewide range in mean snowfall values generally ranged from 30 inches to 70 inches, with smallest values in western counties and the largest values in the northeast. One characteristic evident in the map series is an apparent gradual increase in snowfall over time. Comparison of the patterns shown for the 1890-2000 period and the 1971-2000 period illustrates this tendency. This positive trend in seasonal and annual snowfall is compatible with the upward trend in annual precipitation across Minnesota reported by the DNR State Climatology Office (DNR-WaterTalk, Spring 2000). Consideration of such a trend may be warranted in evaluating the probabilities for snowfall thresholds used as design criteria for the deployment of snow fences around the state.
Frequency distributions of seasonal and annual snowfall were derived for two historical periods of 110-years, and the most recent 30-years. Maps were produced for the 25th, 50th, 75th, and 99th percentile values in the historical distributions for each station record. For example, the map showing the 99th percentile of annual snowfall infers that 99 percent of all years produced an annual snowfall amount that was less than the value shown. By examining this series of maps, road designers and MN/DOT engineers can assess the likelihood of snowfall amounts exceeding various critical thresholds for any given MN/DOT district.
As with the mean seasonal and annual snowfall maps, the percentile maps refer to three distinct types of total snowfall: the Tabler-based snow accumulation season (SAS), the calendar-based snow season (October to March), and the annual snowfall (July through June). The percentile values (in inches) shown on the maps based on SAS and the calendar-based seasonal snowfall are generally 10 to 20 inches less than those depicting the annual snowfall. This is undoubtedly the result of the occasional contributions of early fall or late spring storms to an extreme total snowfall value. Similarly, maps showing the percentile values for the longer time period, 110-year distributions, versus the 30-year distributions, will often show larger values of snowfall since more extreme snow seasons appear in the longer record periods. Since the costs associated with the control of blowing and drifting snow tend to escalate dramatically in extreme winters, it will likely behoove MN/DOT personnel to consider planning to cope with total seasonal or annual snowfalls that represent the 50th percentile or higher in the distribution. The risk-proof approach in planning might be to consider the costs and consequences of coping with a seasonal or annual snowfall that represents the 99th percentile historically. However, this may prove to be unrealistic in terms of costs and the deployment of personnel. Perhaps the most cost-effective planning would be based on another percentile value such as the 75th, or other level relatively high in the distribution but not representing the most extreme case.
Mapped exceedance rankings were produced based on thresholds for seasonal snowfall total used previously in MN/DOT training materials, particularly "Catching the Snow with Living Snow Fences (1999), a guidebook prepared jointly by the Office of Environmental Services and the University of Minnesota Extension Service. These thresholds are 32 inches, 64 inches and 96 inches. The mapped isopleths depict the probability that these thresholds will not be exceeded based on two historical periods, 1890-2000 and 1971-2000 and the two types of snowfall statistics (SAS and annual). These two expressions of the winter season represent the maximum (annual) and minimum (SAS), or extreme, cases. Consequently, the October - March time interval is therefore unnecessary for this product. These maps again illustrate the point that a positive temporal trend in snowfall across the state should be considered in planning endeavors, which involve an assessment of probability. The most recent 30 year period (1971-2000) generally shows a smaller probability that these threshold values will not be exceeded, than the probability derived from the full record period (1890-2000). Stated conversely, there is a higher probability that the given threshold will be exceeded in the most recent 30 years as opposed to the full record. With the 110 year period, the sample size is much greater, therefore the extremes in the distribution are likely to exceed that of the most recent 30 years. This feature is particularly evident in comparing the maps that show the ranking for a 32-inch and 64-inch snowfall total. Planners may want to give more weight to the values shown for the most recent 30 years.
The snow water equivalent (density of freshly fallen snow) is an important parameter in considering the potential for blowing and drifting snow. The more dense (or wet) the snow, the less likely that it will be subject to blowing and drifting. Since the majority of snow movement across the landscape occurs in the interval from November to March, mean snow water equivalent values were calculated for each month of this period. Complete station histories were used with some quality controls applied. Mixed precipitation (liquid and frozen) was not counted in the statistics. The snow water equivalent values were based on the ratio of melted liquid water to snowfall, a measurement made by all National Weather Service stations and many cooperative observer stations. An average was derived based on a compilation of the individual daily values from each station record. Similar to other maps, these data were gridded by a kriging technique and then mapped at 10-km resolution.
As expected, the snow water equivalent maps show greater density in the transition months, particularly November when values as high as 0.16 occur. This translates to about a 6:1 ratio of snow to liquid water, meaning 6 inches of snow is equivalent to 1 inch of water. Lowest values are found in January, with many areas showing values ranging from 0.06 to 0.08. This range conforms closely to the National Weather Service standard of 0.077, which is used for converting snowfall into liquid water equivalent when observers cannot melt down the gage-caught snow. Geographically, areas in southern Minnesota, the Twin Cities metropolitan area, and the shoreline along Lake Superior tend to show consistently higher snow water equivalent values, perhaps as a result of greater low level atmospheric water vapor content (or moisture). Conversely, northern regions of the state show consistently lower SWE, indicative of drier, and colder, atmospheric conditions.
Snow Water Equivalent
Number of observing stations = 180
In order to maximize the geographic distribution of wind speed and direction data, automated stations were selected from the National Weather Service Automated Surface Observation System (ASOS) network and the Federal Aviation Administration's Automated Weather Observing System Network (AWOS) deployed at many Minnesota city and county airports. All hourly observations since 1996 were available from these networks, yielding a total of 86 stations that were utilized to derive wind speed class distributions and wind rose plots. To avoid inhomogeneity in the wind analysis due to observation frequency, period of record and sensor characteristics, data from the Road Weather Information System and University of Minnesota Agricultural Experiment Stations were not utilized.
Table 2 gives a listing of the stations along with pertinent meta-data, such as length of record, and Figures 1 - 86 show the data derived from this network of stations. Wind speed class distributions follow those of Tabler (degree translation is given in Table 3), while the standard wind rose plots illustrate the frequency of occurrence pattern for the given wind direction. The wind roses clearly show the dominance of northwesterly and southeasterly winds at most locations in the state. These plots, along with the modal direction tables, are most useful in determining the attack angle of the wind when planning the orientation and length of snow fences to be placed in the landscape for protection of roads and highways from blowing and drifting snow. However, it is advised that the potential snow transport table also be utilized because valuable information is given expressing the amount of snow transport for each cardinal direction. In some cases, for example, the predominant winter wind direction is from the SSE, yet the majority of the blowing snow is transported from the NW.
Status of the Work Plan: The Web-based data sets and products from Task 2 have now been delivered to MN/DOT staff for review. All products make use of standard software such as Microsoft Visual FoxPro (6.0), Microsoft Visual Basic (6.0), Microsoft Excel (97) spreadsheets and Surfer (6.01) mapping software, following data quality control standards set by the National Climatic Data Center and the DNR-Minnesota State Climatology Office. Mapped products are based on standard kriging techniques (spatial regressions) to derive 10-km based grid point data sets. This is an attempt to maximize the spatial resolution of available data. All maps show county boundaries and can be interpreted to spatial scales that conform to the MN/DOT district level. All of these procedures and product formats will be followed in deriving products and deliverables for Task 3 of this project.
Conclusions: With Task 2 completion, the project work plan (Tasks 3 and 4) will progress more rapidly now. Similar temporal and geographic depiction and resolution of the Tabler parameters will be the focus in Task 3, along with an assessment of the field measurements collected at three southern Minnesota sites. Specific case scenarios will be evaluated in the context of designing an appropriate living snow fence based on the snow climatology and the Tabler parameters.
The Web site for hosting the data base and the summary products from this project (primarily maps, figures, and tables) is being developed and will be made available for evaluation by TAP members later this fall. Reports will be generated as CD’s and distributed to MN/DOT as recommended by the TAP members.(1) Tabler, Ronald D., "Design Guidelines for Control of Blowing and Drifting Snow", SHRP-H-381, Strategic Highway Research Program, National Research Council, Washington, D.C., 1994.
(2) DNR-State Climatology Office, “Minnesota’s Precipitation Climate at the End of the 20th Century,” DNR-WaterTalk, Spring 2000, 17 pp.
(3) Gullickson, Dan et al. 1999. Catching the Snow with Living Snow Fences. MN/DOT Office of Environmental Services and University of Minnesota Extension Service (MI-7311-S) 140 pp.