Mn/DOT Agreement No. 74708

"Climatological Characterization of Snowfall and Snowdrift in Minnesota"

Summary Report for Task 3 of Project

Task 3: Tabler model performance in Minnesota and agricultural impact assessment

Duration: 12 months

Deliverable: Task summary report


     Site Descriptions
Winter Conditions

  Materials and Methods
     Snow Deposition
     Snowpack Density
     Snow Transport
     Soil Temperature
     Soil Moisture
     Crop Yields


Comment: This task report addresses phase three of the project, which is to evaluate snow transport models, evaluate the Tabler methodology for snow fence design and assess agricultural impacts. It was agreed that the report should include a summary of the methods used for calculating snow transport and associated parameters following the Design Guidelines for the Control of Blowing and Drifting Snow publication (Tabler 1994). This includes the calculation of potential snow transport, relocation coefficient, mean seasonal snow transport and snow fence storage capacity. It was further agreed that deliverables would include a field assessment giving a detailed analysis of results from the 2000 - 2001 field investigation at each of the three living snow fence sites. Products include observed snow transport, snow density, soil temperature, freezing depth, soil moisture and crop yields. A second component of Task 3 is a case study analysis for each field site. This product is Web-based and takes the user through the steps necessary to design a living snow fence, while using a combination of products developed in Task 2 with data collected on-site. These steps follow those given in a training session conducted by Dan Gullickson of Mn/DOT and Paul Flynn of the USDA/NRCS. As with Task 2, 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 and the case study could be facilitated using an internet browser (but does not require Internet access).

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.


Site Descriptions

A field investigation was conducted in southwest Minnesota for the 2000 - 2001 winter season to test the applicability of the Tabler model for snow transport and to assess the agricultural implications of living snow fences. Three sites in southwest Minnesota were used, located near the towns of Gaylord (Sibley Co.), Lamberton (Redwood Co.) and Mountain Lake (Cottonwood Co.) (Figure 1), each site representing a different living snow fence design. The first site was located 1.5 miles north of Gaylord, MN, designed to protect an S-shaped, north-south oriented section of trunk highway 22 (Image 1). It consisted of two 8-row strips of standing corn 2,150-ft in length with 150-ft distance between the strips, and positioned 150-ft from the west side of the highway (Images 2, 3). The height of the corn for the winter season ranged between 5 and 7-ft. Prior to snow deposition the corn was approximately 7-ft (Image 4) and roughly 4-ft after snowmelt (Image 5). After snowmelt, the corn was hand-harvested by the landowner to eliminate volunteer corn during the growing season. For the 2001 season, corn was again planted. A second site located two miles to the west of Lamberton, MN at the Southwest Research and Outreach Center was established in collaboration with MN/DOT and the Redwood Soil and Water Conservation District. It consists of a 350-ft long east-west oriented twin-row planting established in 1998, and placed 230-ft on the north side of Highway 14 (Image 6). Three varieties of honeysuckle (Arnold's Red, Freedom and Hawkeye) were planted at a spacing of 4-ft and the rows spaced 4-ft on center (Image 7). At the time of planting, a 10-ft wide black geotextile fabric (termed "EarthMat") was installed at the top of the soil surface for permanent weed control. The height of the honeysuckle for the winter season was also 6 - 7-ft tall. For the 2001 growing season, the area north of the fence was left fallow and wheat was planted on the south side of the fence. Finally, the third site is located 8 miles north of Mountain Lake, MN on the north side of Highway 30 (Image 8). The fence is a 1998 planting of alternating honeysuckle and red cedar; however, the cedar is only 2 - 3-ft tall with the honeysuckle at 8 - 9-ft tall, as of the 2000 winter season. The position of the fence is 300-ft from the north side of the highway and is 2000-ft in length. Soybean was planted in spring 2001 at this site.

Winter Conditions

In determining the effectiveness of living fences as snow barriers, it is necessary to report the meteorological conditions so the results can be viewed in context. The seasonal snowfall and historical ranking, precipitation, temperature, winds and soil moisture conditions of the field sites are presented. Federal meteorological observing stations are located at Gaylord and Lamberton; however the closest station to the Mountain Lake site (by straight-line distance) is in Springfield, MN, approximately 15 miles from the field site. Data from these stations are used here because of the observation frequency and instrument standardization.


Annual (July 2000 - June 2001) snowfall totals for each location were 6-in to 38-in above the 30-year (1961 - 1990) climatological mean, as seen in Table 1. The greatest positive departure from normal for the year occurred in December, with monthly totals of 22-in, 24-in and 20.6-in for Gaylord, Lamberton and Springfield, respectively. When put in the context of the last 100 years, the 2000 - 01 winter season snowfall total (Figure 2) for southwest Minnesota ranks in the 85th to 98th percentile, meaning that 85% to 98% of all observations are less than this total (Figure 3). This was also the case for much of the southern half of Minnesota and along the north shore of Lake Superior.


Both the maximum and minimum temperature data show a two to three degree negative departure from the 30-year normal for the 2000 - 01 season, with the greatest departure occurring in December (Tables 2, 3). Each of the three locations reported approximately 12 F below normal for both the maximum and minimum temperature for December and historically, these rank second for each station. This also corresponds with the period of anomalous snowfall for these locations.


Wind speed and direction is essential when investigating snow transport and blowing snow events. Hourly wind data were collected at the Gaylord and Lamberton sites, but the Windom, MN airport observing station was used to represent the Mountain Lake location (no wind observing station was available for Springfield, as was used for the previous parameters). Frequency distributions of the wind direction observations for the dates within the snow accumulation season are divided into the 16 cardinal directions for each location (Figs. 4, 5, 6). The predominant wind directions for Gaylord, Lamberton and Windom were 280, 60, and 335, respectively for the 2000 - 01 winter season. However, this does not necessarily correspond to the direction of snow transport, as will be discussed in the results.

Precipitation and Soil Moisture

Finally, two crucial components to assessing the agricultural implications of living snow fences are precipitation for the pre- and post-winter season and soil moisture conditions. Precipitation observations for Gaylord, Lamberton and Springfield for the months of November and April illustrate a noticeable positive departure from the 30-year normal (Table 4), specifically in April, which set a record for Lamberton and Springfield. This anomalous rainfall occurred at crucial stages, just prior to the onset of the snow accumulation season and during and after snowmelt. The implications of the November rain are that the soil goes into the winter season with a high water content. Moreover, the excess rainfall during snowmelt, coupled with the anomalously high snowfall for the season, produced exceedingly wet soil conditions in the spring. Data from Lamberton show that available soil water was above average from the time of snowmelt until July (Fig. 7). Tile lines at the Southwest Research and Outreach Center began running March 21, 2001 and ended on August 2, 2001.

Materials and Methods:


Automated meteorological observing stations were deployed at the Gaylord and Lamberton living snow fence sites. Each station was equipped with instrumentation to measure wind speed and direction, air temperature and soil temperature. Wind speed and direction was measured with an RM Young Wind Sentry for each site (Model WS-03002, Campbell Scientific). Air temperature was measured using a temperature probe and standard radiation shield (Model HMP35C and 41002, Campbell Scientific). Thermocouples were prepared using standard thermocouple wire (Type T, Omega Engineering) and installed at a 4-in depth in the soil. Data from these sensors were recorded using electronic data logging equipment at intervals of one hour. At Gaylord, a 21X was used in conjunction with an AM25T Multiplexer (Campbell Scientific). Similarly, data at Lamberton were logged using a CR10 and AM416 Multiplexer (Campbell Scientific). Snow stakes were installed at these two sites for manual snow depth measurements, which consisted of a graduated wooden stick. Frost tubes were installed at Lamberton for measuring the freezing depth in the soil profile. The instruments were fabricated using a set of plastic tubes, the inner tube a flexible vinyl half-inch diameter tubing 3-ft in length, filled to 90% capacity with a dye solution which changes colors when below 32 F (Image 9). The outside of this vinyl tubing was graduated and inserted into a 2-in diameter, 8-ft long PVC tube, capped at both ends. The cap on the top end of the outer tube was attached by wire to the top of the inner tube such that the zero line corresponded to the soil surface. Using a hydraulic probe, a 2-in diameter core of soil was extracted such that the frost tube could be inserted to a depth of 3-ft into the soil, with 5-ft above the soil surface.

Experimental Design

A transect approach was utilized at each field site for the soil temperature, snow stake depth measurement and frost tube instrumentation arrangement. The soil temperature sensors were positioned in a line perpendicular to the Gaylord and Lamberton fence lines at 20-ft intervals. For Lamberton, the sensors extended 100-ft from the leeward side of the fence line and 60-ft on the windward side (Image 10), with a control site established another 40-ft beyond that distance (Image 11). At Gaylord, soil temperature and snow stakes extended between the two corn row strips and 25-ft past the outer rows of each 8-row strip. Instruments were also placed 120-ft from the windward corn rows for the purpose of control measurements (or not under the influence of the snow fence). Correspondingly, snow stakes were installed in the same manner. Frost tubes at the Lamberton site were installed at 40-ft intervals, or at alternate soil temperature and snow stake locations (Image 6). There was no automated instrumentation installed at the Mountain Lake site; however, the snow depth measurements were taken at 20-ft intervals in a transect perpendicular to the fence.

Summer Soil Temperature Measurements

Soil temperature measurements were also taken under the geotextile fabric at the Lamberton planting site during the summer of 2000 and 2001. Since the fabric is utilized by MN/DOT to establish snow fence plantings, the objective of the measurements was to compare the soil temperature at specific depths under the fabric with those under bare soil. The fabric on the west end of the fence line had no mature plantings; therefore the area was not under the influence of vegetative cover (Image 12). Thermocouples were installed at 2-in, 4-in, 8-in and 16-in depths under the fabric. Measurements under bare soil were obtained from an automated weather station located (distance?) at the Southwest Research and Outreach Center. Daily maximum and minimum temperatures are compared at the two locations.

Snow Density

Late season snow density measurements were taken to determine the density of the entire depth of the snowpack for snow volume and snow transport estimates. To sample an entire snow column, hollow 2-in diameter graduated PVC tubes were inserted into the snowpack to the soil surface (Image 13). The snow depth was recorded and the snow inside the tube was then bagged, sealed, and later melted and the volume was recorded (Image 14). In this regard, the volume of snow could be compared to the corresponding volume of water. The same transect approach was used for these measurements as previously mentioned; however the interval was 10-ft (rather than 20) in areas where there was a strong height gradient. For each location along the sampling transect, a pair of measurements was taken to increase the sample size. These measurements were taken at each of the three sites and two times during the season, March 2-3 (Image 15) and March 30, 2001 (Image 16), with the first date prior to the onset of snowmelt.

Soil Moisture

After the completion of snowmelt, gravimetric soil moisture samples were taken for each site using the same transect approach. The samples were taken at locations where the thermocouples were installed, in a line perpendicular to each fence and at each control site (upwind of the fence). A composite of 4 samples were taken to determine soil moisture in the top 0 - 18-in of the soil profile at each sampling location. Once extracted, the soil was placed in plastic bags and sealed while transported to the laboratory. Four samples (weighing approximately 25 grams) from each sampling bag were weighed to obtain a wet weight, dried in an oven at 105 F for 24 hours and weighed again for a dry weight. With this data, the moisture content was then calculated.

Crop Yields

Areas adjacent to the snow fence at Lamberton and Mountain Lake were harvested using yield equipment from the Southwest Research and Outreach Center. Yields were determined using a weigh wagon to weigh samples taken from transects parallel to the fence line at both sites. At Lamberton, five 20-ft wide and 900-ft long transects were run on the downwind side of the fence. While at Mountain Lake, three 30-ft wide and 500-ft long transects were obtained from each side of the fence.


Snow Deposition

Given the type of land use in the areas of Minnesota where snow fences are utilized, it is important to determine the amount of snow, and water, captured at these locations. Two important parameters to consider for snow deposition are fence porosity and height. Porosity of living snow fences is difficult to estimate, as opposed to structural fences, because it is not a stable characteristic of the vegetation. Living fences are 3-dimensional, affected by the wind, and change over time as the plants mature. The configuration at Gaylord, with 8-row strips of standing corn, function as a fence with 50% porosity (Gullickson 1999, p. 47). Studies using structural fences show that this porosity allows for the maximum snow storage capacity, and is therefore the most effective (Gullickson 1999, p. 34). The porosity of a double shrub row fence is estimated to be 27.5%, at maturity. Through visual inspection of the Lamberton fence (twin row honeysuckle), it was determined that the porosity is much higher, and a value of 70% is more representative (Image 7). For the Mountain Lake fence design, the porosity of the single row honeysuckle is estimated at 20% (Image 8).

It is important to note that with an increase or decrease in porosity from 50%, the effectiveness of any fence for a given height will be reduced. In other words, the fence storage capacity decreases with an increase or decrease in porosity from 50%. The following empirical equation has been developed from snow deposition observations for structural snow fences,

Qc = H2.2(3 + 4P + 44P2 - 60P3) [1]

where Qc is the storage capacity [t/m], H = fence height [m] and P = porosity [decimal]. Using the porosity and height values for each fence, the storage capacity has been computed at the three field sites. The storage capacity at Gaylord is found to be the highest, at 38.2 t/m for each 8-row strip. Again, this is because of its porosity (0.5) and it consists of two fence rows, with eight rows of corn each. For the Lamberton fence, the storage capacity is, 30.8 t/m and for the Mountain Lake fence design the capacity is 41.2 t/m.

In figures 8, 9, and 10, a profile view of snow depth is illustrated for the Gaylord, Lamberton and Mountain Lake sites, respectively. The snow depth measurements were taken March 2 - 3, 2001, which was late in the snow accumulation season and prior to the onset of snowmelt. At Gaylord, approximately 300 lineal feet (perpendicular to the cornrows) were under the influence of the fence with an increased snowpack. This extended approximately 80-ft upwind of the first set of rows and roughly 100-ft downwind of the second set. For Lamberton and Mountain Lake, the lineal distance affected was roughly 120-ft and 150-ft, respectively. For all three fences, the snow depth profile showed greater deposition on the leeward, or downwind side of the fence than the windward, or upwind side, meaning a larger portion of the transported snow was deposited on the downwind side.

Since the snowpack is stored on agricultural land, an important result of the snowpack is the additional amount of water that accumulates over the winter. The following table illustrates differences in the amount of snow captured at each site, in water equivalence.


Water from snowpack


10.8 in / acre


10.3 in / acre

Mountain Lake

17.9 in / acre

The greatest amount of snow (water) over a unit area was captured at the Mountain Lake site (Image 19). The Gaylord and Lamberton fences did not capture as much snow in comparison (Image 17, 18). It should be noted that these units are indicative of water from the snowpack over a unit area. Due to the snow fence design, the snow deposition at the Gaylord site covers a larger area (18.1 acres) than the Lamberton and Mountain Lake sites (1.5 and 6.9 acres, respectively), therefore overall, the Gaylord site captured the greatest volume of water (or snow). This will also be evident when looking at the observed snow transport for each site, as will be described in the snow transport section.

Snowpack Density

The snowpack density measurements taken at each site were compiled into two data sets (divided by date) and compared to model estimates of snow density for the range of snow depths observed. The empirical snow densification equation used by Tabler (1994, p. 57) is represented by the following,

ps = 522 - (304/1.485h)(1 - e-1.485h) [2]

where Ps = density (kg/m3) and h = snow depth (m).

As can be seen from Figure 11, the observations of snow density are generally greater than the model predictions. Both the observed and modeled density show increasing density with depth, which is attributed to a rearrangement and compression of snow particles from overburden pressure (Tabler 1994, p. 57). However, the underestimation of the model from our observations could be attributed to the difference in location of where the model was developed (Wyoming) versus where the observations were taken. The arid climate of Wyoming results in lower snow water equivalent (SWE) values than Minnesota for fresh snowfall, which can have an impact on snow densification. The measurements taken on March 30, 2001 show high densities (~ 0.6) at snow depths less than 1-m (3.28-ft). This value is often the assumed density of an actively melting snowpack, and by this time in the season, snowmelt was in progress at each site (Image 16).

Snow Transport


The snow depth and corresponding density measurements taken at each field site allowed for computing the total snow transport over the snow accumulation season. The methodology used for calculating the observed snow transport followed previous work conducted in southern Minnesota by Tabler (1997, p. 109). Table 5 illustrates the differences in total snow transport for the Gaylord, Lamberton and Mountain Lake sites. The snow fence design at Gaylord (two 8-row strips of standing corn) was the most effective of the three, capturing the most snow at 30, 30.5 t/m. As mentioned previously, with a porosity of approximately 50%, studies have concluded that this allows for maximum snow deposition (Gullickson 1999, p. 34). The design at Lamberton (twin-row honeysuckle) captured approximately half of the Gaylord total with 16.6 t/m; however, note that this design was a single fence line rather than two. Also, the porosity was higher at Lamberton, which has an effect on snow deposition. Finally, the honeysuckle/red cedar at Mountain Lake captured slightly more than the Lamberton total with 18.6 t/m. For this fence, the lower porosity also decreased the efficiency of the fence.

Comparing observed snow storage with storage capacity, the Gaylord fence captured 40% of capacity, Lamberton was at 55% capacity and Mountain Lake was at 45% capacity. It is important to note that the empirical equation for snow storage capacity was developed using 50% porous structural snow fences, rather than living. In addition, the living snow fence designs in this field study did not have a bottom gap, or a section of fence just above the surface with 100% porosity. Without the presence of an open area at the bottom of a snow fence, it will become buried with snow during the course of a winter season. If this occurs, as was the case for this study, the snow trapping efficiency will be reduced. This can also result in damage to a living fence due to heavy snow loads. Given the snow deposition profile for each fence design (Fig. 8, Fig. 9, Fig. 10), the fences appeared to be near equilibrium and therefore unable to further capture a significant amount of snow (see Gullickson et al., p. 36).

Nonetheless, given the 2000-01 winter snowfall (>90th percentile snowfall), the design of each fence was more than sufficient to capture the windblown snow. This is important from the standpoint of the design year and will be discussed further in the mean seasonal section under snow transport. In each case, a significant amount of snow was stored upwind of a roadway, helping to mitigate hazardous blowing snow events and increasing safety for travelers.


Previous studies have shown that snow transport in the first 16-ft above the surface is proportional to the wind speed raised to a power, assuming a constant snowcover (Tabler 1994, p. 37). The potential amount of transported snow for a specific area can be calculated using the following empirically-developed equation,

Qupot = (ui3.8/233847)(f)(86400)(n) [3]

where Qupot is the potential snow transport downwind from an infinite fetch with unlimited snow cover [kg/m], ui is the midpoint of the 10-m wind speed [m/s], f is the frequency of observations in the ui wind speed class over a snow accumulation season with n days. The potential snow transport (Qupot) was calculated from wind data for the 86 locations given in Task 2, using only the days within the snow accumulation season for each location. The calculations were restricted to hours when the air temperature was below 32˚ F (using hourly temperature observations) and to wind speeds greater than 11 kts. Above the freezing point and at low wind speeds, it has been shown that snow transport is negligible.

One use of the potential snow transport calculation is to determine the prevailing direction of snow transport, as seen in Table 6 (1000 kg/m = 1 t/m). Note that the direction of greatest snow transport is in the range of 270 - 300 for Gaylord, 300 to 330 for Lamberton, and 300 to 330 for Mountain Lake. Note that the snow transport for Lamberton is significantly different from the wind direction frequency distribution (Figure 5). From the standpoint of snow fence design criteria, this information is needed for calculation of the attack angle.

Relocation coefficient

A second use of the potential snow transport equation is for determining the correct relocation coefficient, which is used in the mean seasonal snow transport equation based on snowfall according to

Qt = 1500 (Swe)(r)(1 - 0.14F/3000) [4]

where Qt is the mean seasonal snow transport [t/m], Swe is the water equivalent of the total mean (1971 - 2000) snowfall over the accumulation season [m], r is the relocation factor and F is fetch distance [m]. Previous studies in Wyoming, U.S. and Siberia have shown the relocation factor to be no higher than 0.7, meaning that no more than 70% of the snowfall is relocated over the winter, therefore this number can be taken as the extreme case.

The relocation coefficient is defined as the proportion of winter snowfall water equivalent that is relocated by the wind. To calculate this, a ratio is taken of the potential transport by wind (Qupot) to the seasonal snow transport based on an unlimited fetch. The latter can be represented by the following,

Q = 1500 r (Swe) [5]

where Q is the mean seasonal snow transport [t/m], Swe is the water equivalent of the total mean (1971 - 2000) snowfall over the accumulation season [m] and r is the relocation coefficient that is solved for. Through personal communication with Tabler (2001), it is expected that this value will range from 0.3 to 0.5 for Minnesota, whereas in the northeast US, the range is 0.2 to 0.3 (Tabler, 1994, p. 44). From Figure 12, it is evident that there is a geographic variability to the relocation coefficient. The topography and land-use characteristics of western Minnesota are such that this area experiences a greater frequency of higher wind speeds, yielding a higher total potential snow transport (Qupot), and therefore a higher relocation coefficient. Conversely, south-central portions and sites in the forested north have comparatively lower coefficients. Averaging all locations, the relocation factor is 0.35.

Mean seasonal

With the calculated relocation factor for southern Minnesota, a better estimate of seasonal snow transport can be calculated using equation 3. For each field site, the relocation coefficients of 0.25, 0.50 and 0.63 were used for Gaylord, Lamberton and Mountain Lake, respectively. The average snow water equivalent for November through March was obtained using Figure 13 and is 0.09 for Gaylord, Lamberton and Mountain Lake. Mean snowfall (1971 - 2000) for each site was obtained using Figure 14. To obtain fetch distance, aerial photographs and topographic maps were used, in addition to visual inspection of each site. For Gaylord, Lamberton and Mountain Lake, the fetch distances were found to be 1280m, 640m and 700m, respectively. Using these variables, the mean seasonal snow transport was calculated to be 14.6 t/m, 16.4 t/m, and 22.3 t/m, respectively. As can be seen, snow transport calculated using equation 3 yields an underestimate for Gaylord and Lamberton and a slight overestimate for Mountain Lake in relation to the observed snow transport (~15.3 t/m per 8-row strip, 16.6 t/m and 18.6 t/m). Given the historical ranking of the 2000 01 winter snowfall, this field investigation is more representative of an extreme case. Using the mean seasonal snowfall in Eq. [4] for design purposes, rather than the 2000-01 annual total, or a 99th percentile ranking, would be satisfactory for the fences to function properly.

Soil Temperature


The substantial snowpack that was sustained over the snow accumulation season provided effective insulation for the underlying soil. The frost tube data at Lamberton and the 4-in soil temperatures at Lamberton and Gaylord illustrate the influence of snowcover on the depth of the freezing layer. For the control site at Lamberton (100-ft upwind of the fence) the freezing layer penetrated to a depth of 35-in, which was reached March 15 (Figure 15). The rate of change for this 32 F isotherm was relatively constant at approximately 4.5 inches per week from November 15 to March 15. At this time, the freezing depth remained constant for roughly 2 weeks until it began moving toward the surface in response to warming temperatures.

Data from all frost tubes showed the same approximate temperature initially, after installation at the beginning of November. However, as the snow depth increased at the 4 frost tube locations in close proximity to the fence, the insulation provided by the snowpack resulted in a warming of the soil. The frost tube located 20-ft downwind of the fence was under the influence of a relatively deep snowpack from the end of November through the end of the snow season. This resulted in soil temperatures beneath the snowpack that remained above freezing until snowmelt. Similarly, the instrument located 10-ft from the upwind side showed a slight decrease in the freezing layer from mid-December (3-in) through March (1-in). Also, the frost tube 40-ft from the downwind side reported an increase in the freezing depth from 2-in to 8-in from installation to mid-December. This was followed by a warming, and a decrease in the freezing depth to 5-in by the end of March. Again, the snowpack provided sufficient insulation to impede further frost penetration in the soil profile.

The 4-in soil temperature data also show a dependence on associated snow cover and respond to the diurnal changes in air temperature (Figure 16, see Figure 8 for soil temperature locations). At the locations under a deep snowpack (greater than 3-ft) the 4-in soil temperatures were in the range of 32 - 34 F, while temperatures in the absence of a snowpack were in the range of 20 - 28 F. The diurnal temperature range varied as well, with insulated locations having a 1 - 2 F range and unprotected locations in the 2 - 5 F range. This is most evident in the December - January observations and by March, the difference in soil temperature between locations is negligible (Figure 17). By April, the effect of snowmelt is evident in the soil temperatures in which locations where the temperature sensors are no longer covered by snow show a marked increase in temperature and a large diurnal variation (Figure 18). At this point, soil temperature is largely a function of air temperature, solar radiation and soil water content.

In conclusion, the insulating properties of the snowpack helped to mediate the underlying soil temperature such that when snowmelt began, the meltwater was able to infiltrate the unfrozen soil. In unprotected areas, the soil is frozen much deeper resulting in meltwater runoff, unable to penetrate the frozen soil.


A comparison of the daily soil temperature measurements in summer under the fabric and bare soil at the 2-in depth revealed a consistently higher maximum temperature under bare soil (Figure 19). This was also the case for the 4-in and 8-in as maximum temperatures as well (Figs 20, 21). However, the minimum temperatures under the fabric showed higher values under the fabric, rather than bare soil. The thermal conductivity of the fabric could be such that less heat is allowed to escape from the soil surface. During nighttime conditions in summer, sensible heat flux is generally away from the surface such that thermal energy is transferred from the warm soil to the cooler air. However, with the fabric covering the soil surface, the heat loss is shown to be less. Deeper in the soil profile, however, both the maximum and minimum daily soil temperature under the fabric were consistently higher than under bare soil (Figure 22). This could be attributed to the decreased heat loss under the fabric, such that more thermal energy is stored in the soil profile. In summary, despite the presence of the black fabric, surface temperatures are no higher than temperatures under bare soil with the same ambient conditions. Therefore, by utilizing the fabric to establish living snow fence plantings, the survival will not be hindered by unusually high daily maximum temperature conditions.

Soil Moisture

For April, the precipitation departure from normal was 4 to 6-in, which historically ranks 1st for Lamberton and Mountain Lake and 2nd for Gaylord. Data collection was delayed due to the abundant rainfall received in the spring of 2001 (Table 4). Samples were collected April 20 at Lamberton and April 27 at Gaylord and Mountain Lake. The soil moisture measurements show no significant difference based on proximity to fence (and over-winter snowpack) (Figs. 23, 24, 25). The values ranged from 23% to 32% at Gaylord, 25% to 31% at Lamberton and 33% to 41% at Mountain Lake. From speaking with the agricultural producers at each site, it was learned that planting was not delayed due to the deep snowpack and allowing time for snowmelt. Rather, the abundant rainfall resulted in late planting, both at these locations and for much of southern Minnesota.

Crop Yields

Samples taken from the south side of the fence at Lamberton show an increase in yield with increasing distance from the fence (Figure 26). Of the five samples, the lowest yield was in the first 20-ft south of the fence at 39.5 bu/A. This corresponds to the area of deepest over-winter snowpack (~ 4.5-ft). The yields from 40- to 100-ft away from the fence showed little variation at 53.6, 53.2 and 54.1 Bu/A. Recall from Figure 9 that the snowpack height decreased rapidly at approximately 40-ft on the south side of the fence, such that snow depths were 1 - 2-ft. In the area directly south of the fence, there appeared to be foxtail growing in the wheat, resulting in a competition problem. The wheat may have been planted when the soil conditions were too wet resulting in an emergence problem, which allowed the foxtail to develop.

A similar correlation between snow depth and crop yields was found for the Mountain Lake site. From Figure 27, it is evident that yields closer to the fence, and in areas sustaining a deep over-winter snowpack, were lower than areas further from the fence (see Fig. 10 for snow depth). Yield for the 30 - 60-ft transect downwind of the fence had slightly lower yields than the area directly south of the fence (the area influenced by the deepest snowpack). However, this is thought to be due the fact that 2 rows were knocked down for various lengths due to prior damage from fertilizer equipment. Also, in past years the land north, or upwind, of the fence, has yielded higher than the land south, or downwind, of the fence. These results confirm this observation and coincide with the wheat yield variation as described previously.

At Gaylord, yield samples were not taken due to several factors. The two 8-row strips planted in 2000 followed the contour of the roadway, having an S-shaped curvature. However, for the 2001 growing season, corn was planted in a north-south straight-line orientation; therefore only a relatively short distance of the corn rows could be harvested for comparable yield observations. In addition, much of the crop experienced lodging due to severe winds just prior to harvest in early November. Although no quantitative measurements were taken at this site, the landowner thought yields for the 2001 growing season were down from previous years. This was the case across the entire field rather than just in the area affected by increased snow deposition.

In summary, yields in the direct vicinity of the Lamberton and Mountain Lake living snow fence were hindered, which appears to be the result of conditions associated with the over-winter snowpack. However, these results are only indicative of one season and should be regarded in such a manner, and in the context of the 2000 - 01 environmental conditions, which is on the extreme end of the distribution. It should also be noted that previous studies have shown that living windbreaks alter the microclimate of the environment in the vicinity of the fence (Brandle and Hintz 1987, Kort 1988, NRCS). Such influences include reduced wind speed, higher relative humidity and temperature and reductions in soil moisture that often result in a yield decrease adjacent to the fence.


(1) Brandle, J. and Hintz, D. 1987. An ill wind meets a windbreak. Science of Food and Agricuture, Volume 5, Number 4.

(2) Kort, J. 1988. Benefits of windbreaks to forage and field crops. Agricuture Ecosystems and Environment. 22/23: 65-190.

(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.

(4) Tabler, Ronald D., 1994. "Design Guidelines for Control of Blowing and Drifting Snow", SHRP-H-381, Strategic Highway Research Program, National Research Council, Washington, D.C., 364 pp.

(5) Tabler, Ronald D., 1997. "Recommended Drift Control Measures for Selected Sites in Southern Minnesota", Prepared for MN/DOT by Tabler and Associates, Niwot, Colorado, 111 pp.

(6) Windbreaks for Conservation, Natural Resources Conservation Service Agriculture Information Bulletin 339.