Proceedings of the South Dakota Academy of Science, Vol. 87 (2008) 269 A STUDY OF SOIL TEMPERATURE CLIMATOLOGY: COTTONWOOD, SOUTH DAKOTA SOIL TEMPERATURES FROM 1982 TO 2004 Dennis P. Todey, Joanne Puetz Anderson, and Chirag Y. Shukla Agricultural and Biosystems Engineering Department South Dakota State University Brookings, SD 57007 ABSTRACT Soil temperatures have been collected at some sites in South Dakota for over 20 years. The information is important for agriculture, hydrology, climatology, and other interests. The Cottonwood, SD, site has over 20 years of soil temperature data at the 5, 10, and 20 cm depths and several years at the 50 and 100 cm depths. This study used morning observed soil temperatures at the 5, 10 and 20 cm depths to develop a procedure for predicting the beginning and end of the annual freeze/thaw cycle at other similar sites across the state. Maximum, mean and minimum morning soil temperatures, sorted by Day of Year (DOY) and depth, were plotted to show the annual soil temperature oscillations and variability at each depth. The 5 cm and 10 cm temperatures were each divided into spring and fall sets. The first DOY observed temperature above 32ºF (0ºC) and the last DOY observed temperature at or below 32ºF (0ºC) were determined for each spring. The first DOY temperature at or below 32ºF (0ºC) and the last DOY temperature above 32ºF (0ºC) were determined for each fall. From these, percentiles were calculated to indicate the expected beginning and end of the spring and fall freeze/thaw cycle. The first DOY soil temperatures at or below 40ºF (5ºC) in the fall were also determined to compare time between soil temperatures reaching 40ºF (5ºC) and 32ºF (0ºC). A forecast model could be useful to the agricultural community and others for the potential period before soil freeze-up. The procedure developed for the Cottonwood data will be used at other sites across the state to develop maps of when frozen soils are expected in South Dakota. Keywords Soil temperatures, frozen soil, South Dakota climate variability, soil temperature trends, freeze, thawed soil INTRODUCTION Soil temperature information (current and historical) is important for agriculture, hydrology, climatology, weather forecasting, construction, transporta-
270 Proceedings of the South Dakota Academy of Science, Vol. 87 (2008) tion, recreation and other interests across the state. In agriculture there have been changes within the Natural Resources Conservations Service standards on spreading manure on frozen ground. The Wisconsin Department of Natural Resources discussed this topic in NR 243 CAFO (Concentrated Animal Feeding Operations) Winter Spreading Restrictions. The following was stated: Solid Manure [see s. NR 243.14(6)] Generally, CAFOs may surface apply solid manure on frozen or snowcovered ground except during February and March. Beginning Jan. 1, 2008, CAFOs may not surface apply solid manure during February and March on areas of fields frozen anywhere between the first ½ and 8 of soil or on areas that have 1 or more of snow. Liquid Manure [see s. NR 243.14(7)] Except for liquid manure that is frozen and cannot be transferred to storage, CAFOs may not surface apply liquid manure at any time during February and March regardless of soil conditions. In addition, CAFOs may not surface apply liquid manure during other winter months when the ground is frozen or snow covered, with the following exceptions: (Wisconsin Dept. NR243). The state of Iowa has been discussing restrictions for CAFO for application of manure on frozen ground. In Iowa s Manure on Frozen Ground Update #1, under The Practice, the following protocols and reasons are given for guidelines: As stated in the motion, this rule is intended to restrict surface application of manure on ground made impermeable by freezing soil moisture, snow pack or surface ice. If manure is injected or incorporated this rule would not apply. Proper injection is thought to be impossible with more than 4 inches of snow. Application during overnight freezes in the spring should be allowed (when the first ½ inch or less of soils is frozen or when there is less than 1 inch of snow). (Olson 2008). According to the Department Environmental and Natural Resources (DENR) in South Dakota, CAFO permits have manure application restrictions. Under manure application restrictions, d. Surface broadcasting liquid manure on frozen and snow-covered ground should be avoided. If surface broadcasting liquid manure, the land must have slopes of less than 4%, a 100 foot buffer zone must be maintained to wetlands and waterways, and DENR shall be notified prior to application. (DENR July 08). Knowing when to expect frozen soil would be helpful to CAFO permit holders in scheduling the spreading of manure. This information is also useful
Proceedings of the South Dakota Academy of Science, Vol. 87 (2008) 271 for any industry involved with earth work, such as pouring concrete, excavation, archeology, preparation of athletic fields, parking lots, and underground storage tanks, to name a few. Soil temperature is also a critical component in fall nitrogen application to cropland. The objective of this preliminary study was to use morning soil temperatures from Cottonwood, SD, at the 5, 10 and 20 cm depths to develop a procedure for predicting the beginning and end of the annual freeze/thaw cycle at other similar sites across the state. METHODS Seventeen COOP (National Weather Service Cooperative Observer Program) sites in South Dakota record soil temperatures. Some, such as the Cottonwood and Brookings sites, have over 20 years of data, while others, like Yankton, have only one year of data. These data were collected either manually or automatically. Manually collected data were recorded once or twice a day, usually morning and /or evening, to capture some of the daily temperature variation (roughly maximum and minimum temperature). Automated sites usually report hourly observations of soil temperatures. There are no written standards on what depths to monitor, but most are taken at 2, 4, 8, 20, 40 and 72 inches (5, 10, 20, 50,100 and 180 cm). Soil temperatures for Cottonwood, SD, were obtained from the National Climatic Data Center (NCDC, 2008). These have been collected since February 16, 1982 at this location (latitude 43.7 and longitude -101.87, COOP ID number: 391972). Because the soil temperatures were recorded in Fahrenheit, our study will report and evaluate data in the same units. For our study soil temperature observations from February 16, 1982 to November 29, 2004 were analyzed. The scope of this preliminary study was limited to the 5, 10 and 20 cm data due to limited data for 50 cm and 100 cm depths. Morning observations were used because they were the most numerous and continuous of the observation times at all sites across the state. Microsoft Excel was used to organize the data before plotting. Line plots depicting the temperatures at the 5, 10, and 20 cm depths over time were created for preliminary analysis. The dates were converted to DOY for each set of data and plotted as a scatter plot of temperature versus DOY and as a line graph of the maximum, average and minimum temperature for the study period. The data, organized by DOY, was then divided into the spring half of the year (DOY 183 or less) and the fall half of the year (DOY 184 or more). Individual days with missing temperature data were located. If temperature data for 5 or more consecutive days were missing, those dates were evaluated to see if they would impact the first or last day of the season when the soil temperature was above, below, or at 32 0F (00C). For example, 30 days of missing data in August was considered to have no impact, but 5 days of missing data in October would require further evaluation. If it was determined that the missing observation(s) fell within a critical time period, that season would be excluded from the analysis. Spring and fall seasons that were excluded (Table 1) due to missing data fell
272 Proceedings of the South Dakota Academy of Science, Vol. 87 (2008) on days that would have been needed to evaluate first and last days of above or below 32ºF (0ºC) soil temperatures. Table 1. Data years excluded from analysis due to missing data at a critical period. SPRING FALL first day > 32ºF last day 32ºF first day 32ºF last day > 32ºF 1982 1982 1984 5 cm 1985 1985 1992 1994 1982 10 cm 1985 1985 1992 1992 1995 1982 1982 20 cm 1984 1984 1985 1985 1994 1994 For each year with sufficient data, two series were plotted on a line graph for the spring: the first DOY when the temperature was above 32ºF (0ºC) and the last DOY when the temperature was 32ºF (0ºC) or below. Three series were plotted for the fall graphs: the first DOY when the temperature was 32ºF (0ºC) or below, the last DOY when the temperature was above 32ºF (0ºC), and the first DOY when the temperature was 40ºF (5ºC) or below. This third line was used to evaluate if the first DOY of 40ºF (5ºC) or below could be used as a predictor for the first day when the soil temperature would be 32ºF (0ºC) or less. Finally percentiles were calculated for each set of data and plotted in a box plot form. RESULTS Figures 1, 2 and 3 show plots of daily soil temperatures for each depth by date. An annual cycle is clearly notable. The daily morning temperatures at the 5 cm depth (Figure 1) and at the 20 cm depth (Figure 2) have similar characteristics. Because these were morning soil temperatures, the observations at 20 cm were warmer than those at 5 cm. In addition, the average temperature at 20 cm was about 7ºF (4ºC) warmer than those at 5 cm for that time of day. The decrease in the range of temperatures as the depth of the sensors increased was also expected. Note also that the period of the study was only 8 years at the 100 cm depth.
Proceedings of the South Dakota Academy of Science, Vol. 87 (2008) 273 O F O F 90 90 80 80 70 70 60 60 50 40 50 30 40 20 30 10 20 0 10 0 2/16/1982 2/16/1984 2/16/1986 2/16/1988 2/16/1990 2/16/1992 2/16/1994 2/16/1996 2/16/1998 2/16/2000 2/16/2002 2/16/2004 5 cm daily morning soil temperatures 5 cm daily morning soil temperatures Figure 1. Daily morning soil temperatures at Cottonwood, SD, from 1982 to 2004 at the 5 cm depth. The horizontal line represents the mean soil temperature at 5 cm from 1982 to 2004. 2/16/1982 2/16/1984 2/16/1986 2/16/1988 2/16/1990 2/16/1992 2/16/1994 2/16/1996 2/16/1998 2/16/2000 2/16/2002 2/16/2004 Figure 1. Daily morning soil temperatures at Cottonwood, SD, from 1982 to 2004 at the 5 cm depth. The horizontal line represents the mean soil temperature at 5 cm from 1982 to 2004. 90 80 70 60 O F 50 40 30 20 cm daily morning soil temperatures 20 10 0 2/16/1982 2/16/1984 2/16/1986 2/16/1988 2/16/1990 2/16/1992 2/16/1994 2/16/1996 2/16/1998 2/16/2000 2/16/2002 2/16/2004 Figure 2. Daily morning soil temperatures from Cottonwood, SD, from 1982 to 2004 at the 20 cm depth. Figure 2. Daily morning soil temperatures from Cottonwood, SD, from 1982 to 2004 at the 20 cm depth. The horizontal line across graph represents the mean soil temperature at 20 cm from 1982 to 2004.
274 Proceedings of the South Dakota Academy of Science, Vol. 87 (2008) 70 65 O F 60 55 50 100 cm daily moring soil temperatures 45 40 35 30 6/1/1997 6/1/1998 6/1/1999 6/1/2000 6/1/2001 6/1/2002 6/1/2003 6/1/2004 Figure 3. Daily morning soil temperatures at Cottonwood, SD, from 1997 to 2004 at the 100 cm depth. Descriptive statistics calculated on each set of data show that the range of soil temperatures decreases with depth (Table 2). Table 2. Morning soil temperature statistics. Monitor level 5 cm 10 cm 20 cm 100 cm Mean 44.02 48.39 51.02 52.14 Median 42 47 50 52 Range 77 73 70 33 Minimum 3 10 15 34 Maximum 80 83 85 67 Confidence Level (95.0%) 0.378 0.385 0.385 0.333 Number of observation 7832 7861 7783 2742 Because morning observations were used the 5 cm depth was cooler than the 10 cm depth. The 20 cm depth was the warmest of these three depths. Figure 4 shows a pronounced thermal delay for the 100 cm depth compared to the upper three layers. The thermal delay between 5 cm and 100 cm was about 30 days. The range of soil temperatures also decreased with an increase in soil depth. The first and last DOY when soil temperature was 32ºF (0ºC) or below (fall) and the first and last day when soil temperature was greater than 32ºF (0ºC) (spring) were also calculated for each year. Figures 5 and 6 display the 5 cm spring and fall data, respectively, while Figure 7 displays the 10 cm fall data. Notice that in some years, when the soil temperature first reached 32ºF (0ºC), it remained at or above freezing for the rest of the spring. In other years the
Proceedings of the South Dakota Academy of Science, Vol. 87 (2008) 275 75 10 cm 65 55 100cm F O 45 35 100cm 20 cm 20cm 10cm 5cm 25 5 cm 15 1 31 61 91 121 151 181 211 241 271 301 331 361 DOY Figure 4. Average morning soil temperatures by day of the year (DOY) for four depths at Cottonwood, SD. 5/19 4/29 First day above 32 last day 32 or below 4/9 3/20 2/29 2/9 1/20 1982 1986 1990 1994 1998 2002 2006 Figure 5. First day of the year (spring) with 5 cm temperature greater than 32ºF and the last day below 32ºF (0ºC). year
276 Proceedings of the South Dakota Academy of Science, Vol. 87 (2008) temperature fluctuated above and around freezing for over 30 days. The mean difference between the first day above 32ºF (0ºC) and the last day at or below 32ºF (0ºC) was 11 days with a range of 0 to 34 days. While this extended period of fluctuation is uncommon, it must be explained to producers because refreezing impacts manure application. The variation likely is a function of snow cover, vegetation difference, or moisture content between years, and would be an excellent topic for future study. Figures 6 and 7 show the first DOY when soil temperature was greater than 32ºF (0ºC), at or below 32ºF (0ºC) and at 40ºF (5ºC) for the fall half of the year for 5 cm and 10 cm depths, respectively. The mean difference between the first day the temperature was 32ºF (0ºC) or below and the last day the soil temperature was greater than 32ºF (0ºC) was 11 days at the 5 and 10 cm level. The range of this difference was 34 days for 5 cm with a minimum of 0 and a maximum of 34 days. The range of the difference was 33 days at 10 cm with a minimum and a maximum of 0 and 33 days, respectively. The difference between the first 40ºF (5ºC) or less day and the first 32ºF (0ºC) or less day may be important in order to create a forecast model to predict when frozen ground would occur. The mean differences were 37 days at 5 cm and 21 days at 10 cm. The minimum and maximum differences were 19 and 53 days at the 5 cm level and 3 and 35 days at the 10 cm level. Percentiles were calculated for 5 and 10 cm data for the fall half of the year. Figure 8 shows the percentile information in a box plot format and illustrates how the first and last day values progress through the fall at the10, 25, 50, 75, and 90th percentiles.. 12/17 11/27 first day 32 or below last day above 32 first day 40 or below 11/7 10/18 9/28 9/8 1982 1986 1990 1994 1998 2002 2006 years Figure 6. First day fall soil temperatures (5 cm) were 40ºF or below, the first day 32ºF or below, or last day above 32ºF (0ºC) for the season.
Proceedings of the South Dakota Academy of Science, Vol. 87 (2008) 277 12/31 12/17 12/3 11/19 Date 11/5 10/22 10/8 9/24 first day soil temperature 40 degree or less first day 32 degrees or less 9/10 last day greater than 32 degrees 1982 1986 1990 1994 1998 2002 Figure 7. First day fall soil temperatures (10 cm) were 40ºF or below, the first day 32ºF or below, and the last day above 32ºF (0ºC) for the season. Year 12/31 12/17 12/3 11/19 Percentile 25% Percentile 10% average Percentile 50% Percentile 90 % Percentile 75% Fall Box Plots of Cottonwood 1982-2004 11/5 10/22 10/8 9/24 5 cm first day 40 degrees or less 10 cmfirst day 40 degrees or less 5 cm first day 32 degrees or less 10 cm first day 32 degrees or less 5 cm last day greater than 32 degrees 10 cm last day greater than 32 degrees Figure 8. Percentile box plots of first and last day dates in the fall half of the year. Figure 8. Percentile box plots of first and last day dates in the fall half of the year.
278 Proceedings of the South Dakota Academy of Science, Vol. 87 (2008) DISCUSSION The main purpose of this study was to set up and refine a procedure for analysis of soil temperatures at Cottonwood, SD, that could be used to analyze data from other sites across South Dakota. When data from the entire state are published and available, users of this information will be able to better plan and schedule such activities as when to spread manure on their fields. Replicating this study at other COOP sites that report soil temperatures across South Dakota would allow for a spatial study of soil temperatures statewide. One result of the broader study would be a map of South Dakota with contours of critical soil temperatures. Such a map would display values such as the average date when the first day in the fall the 5 cm soil temperature would be 32ºF (0ºC) or below, etc. This would allow users concerned about frozen soils to better plan their fall schedule. More work also needs to be done to find a method to predict frozen soil in the fall. Examining other locations or looking at the relationship between the different depths may reveal a possible forecast model. Also with more sites being analyzed, trends over the last 20 years may also be revealed and evaluated. ACKNOWLEDGEMENTS This work was funded through a United States Geological Survey 104b grant through the South Dakota Water Resources Institute. A special recognition goes to Janelle A. Anderson for helping improve the graphs. LITERATURE CITED Olson, J. 2008. Manure on Frozen Ground Update #1 June 10, 2008, Watershed Monitoring & Assessment Section Iowa Department of Natural Resources Wallace State. http://www.iowadnr. gov/epc/08jul/12.pdf (obtained July, 2008). NCDC. 2008. National Climatic Data Center, data references. http://cdo. ncdc.noaa.gov/ pls/plclimprod/poemain.accessrouter?datasetabbv=sod (obtained Jan, 2008). SDDENR. 2008.South Dakota Department of Environment and Natural Resources, The 10 Most Asked Questions about the State General Permit Process for Concentrated Animal Feeding Operations. http://www.state.sd.us/ DENR/DES/Surfacewater/feedlotquestions.htm (obtained July, 2008). Wisconsin Department of Natural Resources. NR 243 CAFO Winter Spreading Restrictions Wisconsin Department of Natural Resources CAFO Applications When Ground is Frozen, Snow-Covered or Saturated. http://dnr. wi.gov/runoff/pdf/rules/nr243/ WinterSpreading.pdf (obtained July, 2008).