Ecological Archives A-010-002-A1

Christopher Daly, Dominique Bachelet, James M. Lenihan, Ronald P. Neilson, William Parton, and Dennis Ojima. 2000. Dynamic simulation of tree-grass interactions for global change studies. Ecological Applications 10: 449-469.

Appendix A

Preparation of MC1 input data for the Wind Cave study area

Following is a discussion of the methods used to prepare the soils and climate data sets necessary to apply MC1 on the Wind Cave study area. A digital elevation model (DEM) of the study area (50-m resolution) supplied by the EROS Data Center served as a base map for input data set development.


MC1 requires soil texture (i.e., percent sand, silt, and clay), rock fraction, and depth to bedrock. Lacking a detailed soils map specific to the study area, a map of terrain classes was used to develop a dataset that reflected the observed topographic variation in these soil characteristics (Stephen Ogle, pers. comm.). Five terrain classes ranged from ridgetops to swales. Mean soil texture and percent rock were estimated in each of the terrain classes from samples of the first 50 cm of the profile collected in the field. For cells within a given terrain class, a corresponding mean texture was assigned to all soil layers, but below the 50 cm level the mean percent gravel was increased by 12 percent, based on published soil series descriptions for the Black Hills (USDA Soil Conservation Service, 1990). Soil depths were assigned to each of the terrain classes by linear interpolation along an assumed gradient from the shallowest soils on ridgetops to the deepest soils in swales. The two extremes in soil depth were also estimated from the Black Hills soil series descriptions.

Historical climate

A gridded, 100-year monthly climate dataset (1895-1994) was prepared for the study area and included precipitation, minimum and maximum temperature (used by the biogeochemistry module), vapor pressure and wind speed (only used by the fire module), and solar radiation. All the climatic variables are necessary to run MAPSS in the initialization phase. Since local stations had much shorter periods of record, the generation of 100 years of climate data was made possible by a 99-year temperature and precipitation station dataset produced for the VEMAP climate data project (Kittel et al 1997). Incomplete station records were statistically in-filled using information from surrounding stations to form a complete record back to 1895. Complete VEMAP records for Hot Springs (1085 m, 50 km south of WCNP) and Custer (1670 m, 20 km west of WCNP) were used in this analysis.


Precipitation data were available for the period 1991-1994 at WCNP headquarters (1244 m, 5 km south of the study area). Monthly observations at this site were found to be highly correlated (r2 = 0.85; n=48) with those from Hot Springs. A 100-year precipitation record for the headquarters site was therefore synthesized using a regression relationship with Hot Springs. A sense of the patterns of precipitation in the study area relative to the headquarters site was obtained from temporary rain gauges operated in 1993 as part of the National Biological Service CEGR-1 (Central Grasslands) project (Detling and Ojima 1998). Using this network as a rough guide, precipitation in the study area was varied from 103% of headquarters in the southeast to 111% of headquarters in the northwest.


Temperature data were available at park headquarters for the period 1991-1994. Thanks to extremely high correlations between minimum and maximum temperatures at headquarters and both at Custer and Hot Springs (r2 = 0.99; n = 48), the headquarters site could be eliminated and the long-term data at Custer and Hot Springs used directly. Using Custer as the high-elevation site and Hot Springs as the low elevation site, monthly lapse rates were calculated and applied to the DEM elevations in the study area to produce flat-surface temperature estimates for each grid cell.

A relationship between slope exposure and vegetation pattern is easily seen in the study area. For example, coniferous forest cover is typically greatest on shady slopes. This is thought to be partially a result of the influence of direct-beam solar radiation on maximum temperature. The flat-surface maximum temperatures described previously were modified to reflect these various radiation regimes in the study area. Hypothetical "clear-sky" solar radiation was simulated by calculating direct and diffuse beam radiation for each DEM pixel, accounting for pixel orientation, slope, surrounding horizon and ratio of diffuse to direct beam radiation (Frew, 1990). The ratio of diffuse to direct beam is a function of atmospheric transmissivity, usually due to cloudiness (Bristow and Campbell 1985). Transmissivity was inferred from the difference between actual solar radiation in Rapid City and the hypothetical clear-sky radiation calculated for the same location. However, solar radiation observations were available from Rapid City only for the period 1992-1994. A 100-year record of transmissivity was synthesized by developing a relationship between the temperature range at Hot Springs and the ratio of observed radiation at Rapid City to the calculated clear-sky radiation at WCNP (r2 = 0.50; n=23). Monthly direct-beam radiation was then calculated for each pixel in the study area using the synthesized transmissivity estimates. Based on observations by Running and Nemani (1985) on the heating effect of solar radiation on various slopes in Montana, maximum temperature deviations from the flat-surface values in the study area were varied between ± 1-2C. Minimum temperatures were not observed to vary in a predictable way on different slope exposures (Running and Nemani 1985), and thus were not modified for slope exposure.

Wind Speed

No wind speed data were available in the park, requiring that observations at Rapid City for 1992-1994 be used as wind speed estimates for the study area. Attempts to extend the record to 100 years by developing relationships between wind speed and precipitation or temperature were unsuccessful. However, a strong, repeating seasonal cycle in wind speed was observed; mean wind speeds were typically greatest in the winter and spring, and least in the summer. This seasonal pattern was used to represent all 100 years of record.

Vapor Pressure

The nearest observations of dew point were again at Rapid City, for the period 1992-1994. A strong regression relationship was developed between monthly dew point at Rapid City and minimum temperature at Hot Springs (r2 = 0.93; n=36). This function was applied to the gridded flat-surface minimum temperatures in the study area to obtain gridded dew point estimates. Dew point was converted to vapor pressure by calculating the saturated vapor pressure at the dew point temperature.

Future Climate

A future climate scenario for the years 1995-2094 was derived from a climate scenario based on output from the Hadley Center for Climate Prediction and Research general circulation model (GCM) (Johns et al. 1997). The scenario was derived from a Hadley GCM application that included the effects of sulfate aerosols (HADCM2SUL, Johns et al. 1997). The scenario was available on the VEMAP United States grid, which covers the continent at 0.5-degree resolution. The grid cell encompassing the Wind Cave area (row 9, column 41) was selected to provide the future climate change values for our study region.

The future climate scenario was not expressed as actual values of the climate elements. Instead, it was provided as monthly changes in precipitation, minimum and maximum temperature, vapor pressure, and solar radiation between two time periods in the GCM run: 1961-90 and 2061-90. Changes in precipitation, vapor pressure and solar radiation were given as ratios, i.e., (2061-90 mean) / (1961-90 mean), while those for minimum and maximum temperature were differences, i.e., (2061-90 mean) - (1961-90 mean). Using the 1895-1994 historical time series as a base (to provide realistic interannual and decadal variability), climate values for the years 1995-2094 were created by applying the GCM changes to each element in a linearly ramped fashion. For example, maximum temperature for January 1995 was calculated as the maximum temperature for January 1895 + 1/100th of the GCM change in January maximum temperature, January 1996 was January 1896 maximum temperature + 2/100th of the GCM change, and so on to January 2094, which was the January 1994 maximum temperature + all of the GCM change in January maximum temperature.

To avoid the potential for step-changes in climate values between the end of the historical time series in 1994 and the beginning of the future scenario in 1995, the historical time series was checked for significant trends (p < 0.05) over the 100-year period and these trends removed. Slight, but significant trends were found and removed for February and March minimum temperature and vapor pressure, February maximum temperature, and September solar radiation. Trends were removed by developing a linear regression function between a climate element and year, then subtracting this function from the original data.


Literature Cited

Bristow, K. L., and G. S. Campbell. 1985. An equation for separating daily solar irradiation into direct and diffuse components. Agricultural and Forest Meteorology 35: 123-131.
Detling, J. K., and D. S. Ojima. 1998. Predicting the effect of climate change on vegetation in park landscapes in the Central Grassland region. National Biological Survey - Global Change research Program CADGER-2, Annual Report. Fort Collins, Colorado, USA.
Frew, J. E. 1990. The image-processing workbench. Dissertation. University of California, Santa Barbara, California, USA.
Jones, T. C., R. E. Carnell, J. F. Crossley, J. M. Gregory, J. F. B. Mitchell, C. A. Senior, S. F. B. Tett, and R. A. Wood. 1997. The second Hadley Centre coupled ocean-atmosphere GCM: model description, spinup and validation. Climate Dynamics 13:103-134.
Kittel, T. G. F., J. A. Royle, C. Daly, N. A. Rosenbloom, W. P. Gibson, H. H. Fisher, D. S. Schimel, L. M. Berliner, and VEMAP II participants. 1997. A gridded historical (1895- 1993) bioclimate dataset for the conterminous United States. Proceedings of the 10th conference of Applied Meteorology. 20-24 October 1997, Reno, Nevada, USA. American Meteorological Society, Boston, Massachusetts, USA.
Running, S. W., and R. Nemani. 1985. Topographic and microclimate control of simulated photosynthesis and transpiration in coniferous trees. Pages 53-60 in H. Turner and W. Tranquillini, editors. Establishment and tending of subalpine forest: research and management. Proceedings of the Third IUFRO Workshop, Bermensdorf, Switzerland.
USDA Soil Conservation Service. 1990. Soil Survey of Custer and Pennington Counties, Black Hills Parts, South Dakota. U.S. Department of Agriculture.

[Back to A010-002]