Data & Data Management

Two major data sets are required for this project. In order to compare the past to the present, a modern baseline climate data set is needed. Also, if ecosystem classifications are to be made on the past data, the baseline data will need existing ecosystem classifications also. Second, a complete paleoclimate data set for the study area is required at each of the temporal intervals of concern (6,000 and 21,000 YBP). Ideally, both data sets will have similar resolutions and coverages.

Ecological Niche Modelling
Baseline Data
Paleoclimate Data
ClimateBC Input & Output Variables
Ecosystem Classification Data

1) Ecological Niche Modelling (Top)
The modelling process itself is based on the concept of ecological niches. Simply put, an ecological niche is a geographic arrangement of abiotic factors affecting the survivability of an individual of a plant species (Jackson & Overpeck 2000). These abiotic factors include climate, soils, topography, etc. However, because interspecies interactions also occur (typically in the form of competition), a species does not occupy all of the ecological space in which it potentially could survive (its fundamental niche). The actual ecological space that is inhabited by the individuals of a species (its realized niche), is determined by all these factors in combination: climate, soils, topography, competition, migration, seed dispersal, etc. (Jackson & Overpeck 2000). It is these realized niches that an effective model should produce.

Ecological niche modelling attempts to do just that. The process uses abiotic data such as climate and topography and attempts to use them as predictive variables to generate simulations of biotic communities based on changes in that data. Essentially, the model correlates certain plant communities to certain abiotic conditions, then uses those correlations to predict plant communities given a novel arrangement of abiotic data. Of the available abiotic data, climate is largely accepted as the primary driver of ecosystem and species distribution (Rehfeldt 1999). In fact, it has been suggested that vegetation may also respond very quickly to climatic fluctuations, often indicating a threshold response to certain variables (Williams et al. 2002).

The differentiation of the fundamental and realized niche of a species is critical for modeling. It differentiates whether a simulated species distribution is based on theoretical abiotic and biotic constraints or whether the projected distributions are based on the actual occurrence of species in the field, where they are exposed to interactions (both positive and negative) of other species and individuals (Guisan & Zimmerman 2000). This is an important distinction for paleo-modelling, as modeling interspecies competition is both critically important and very difficult. This project is indirectly based on realized species niches as the baseline ecosystem data is derived from field observations of species.

2) Baseline Data (Top)

High resolution climate data for the current era is actually relatively simple to acquire, thanks to an efficient software package developed through the University of British Columbia by Wang, Spittlehouse, Hamann, and Aitken (Wang 2006). ClimateBC is an MS Windows application that generates interpolated climate data at virtually any resolution based on monthly temperature values for specific locations throughout the Pacific Northwest.

(Click for larger image)

The ClimateBC software allows the user to input spatial locations (lat/long) paired with elevation data at any resolution. From this input, the software will generate a variety of climatic variables (listed below) for each data point (location). In addition to British Columbia, the software was used in this project for coverages through the Canadian prairies and through the western United States.

University of British Columbia: ClimateBC Information & Download
Dr. Andreas Hamann, University of Alberta: ClimateBC & ClimatePP Information and Download

From the ClimateBC software, it is possible to generate high-resolution, topography-incorporated climate maps of the study area. The data set itself is a spatially explicit point file with 1 elevation, and 19 climate variables. In the baseline area, because we have ecosystem classifications (see below), it is possible to attach each of the ecosystem hierarchy classes to each spatial data point also. This is the format of the data that can now be displayed and processed in both SAS statistics software and ArcMAP GIS software.

(Click for larger image)	Baseline Data: mean annual temperature (MAT) Temperature Range (deg. C) = -6.9 to 10.4 (blue to red)
(Click for larger image)	Baseline Data: mean annual precipitation (MAP) Precipitation Range (mm/year) = 230 to 9450 (red to blue)
(Click for larger image)	Baseline Data: temperature difference / continentality (TD) Temperature Range (deg. C) = 9.1 to 42.1 (blue to red)
(Click for larger image)	Baseline Data: mean warmest month temperature (MWMT) Temperature Range (deg. C) = 3.8 to 20.5 (blue to red)

3) Paleoclimate Data (Top)

Paleoclimate data is just as the term would suggest: climate data for past time periods. There are many sources of paleoclimate data: some based on field-collected fossil data, some generated through more theoretical computer-based climate models. These computer climate models are similar to those used currently to predict future climate conditions and climate change impacts.

For this project, I chose to use the output data from a paleoclimate model that generates climatic conditions on a global scale. General circulation models (GCMs) are computer-based climate models which generate temperature and precipitation data. In general, the models are built based on global patterns of oceanic and atmospheric circulation.

The Center for Climate Research at the University of Wisconsin-Madison is involved in a wide variety of climatic research projects, including the development of climate and paleoclimate models. Based largely on planetary orbital patterns and atmospheric chemistry, the Community Climate Model (CCM) was one of the center's original global circulation models (Kutzbach 1988, Kutzbach & Guetter 1986). Orbital forcing has been, and remains to be, considered as the primary driver for terrestrial glaciations and climate variation throughout the Holocene (Overpeck et al. 1989).

I am aware that the Community Climate Model (CCM) program has seen many updates since its first development. For future analysis, it may be desirable to pursue a more current version of the GCM. Also, the Paleoclimate Model Intercomparison Project (PMIP) in another valuable source of paleoclimate model data and research. Thus far, I have been unable to contact the appropriate project coordinators. For the purposes of this project, the CCM1 data is sufficient for developing procedures and preliminary results. Ideally, to account for variations in GCM output data, a variety of GCM data should be considered (Newman 2006). However, for the scope of this project, the CCM1 data will suffice.

This circulation model was chosen for two simple reasons:

First, the data is publicly available and easily accessed online. The data set is posted at the World Data Center of the National Oceanic and Atmospheric Association (NOAA)—a division of the United States Dept. of Commerce, National Climatic Data Centre. The NOAA paleoclimate data library is impressive in breadth and uses a very interactive online data display application.

Second, the data itself is provided on a monthly basis for the years in which I am interested—namely 6,000 and 21,000 years ago. As discussed above, the ClimateBC software uses monthly temperature and precipitation data inputs to generate a plethora of more general and some more specific climate variables. Because the data formats match, I was able to use the paleoclimate data as an input data set for the ClimateBC. Furthermore, ClimateBC requires anomaly data values (from present-day) and the online display and download centre on NOAAs website allows for the data to be accessed as anomalies. This allowed me to use the same input spatial data set (lat, long, elevation) and generate high-resolution paleoclimate maps.

This high-resolution conversion was very important, as the GCM data is provided only in lower resolution (CCM1 is provided on a global scale at 1 degree lat/long resolution). The coarse nature of the paleoclimate data does not discourage any analysis work. As this is a theoretical, computer-generated data set based on general circulation patterns, it would not provide any more information to generate the paleoclimate data on a finer scale. A simple interpolation of the data will adequately generate the global coverage necessary for this project.

The most efficient means of generating high-resolution paleoclimate data sets at the same scale and with the same climate variables as the baseline data is to simply use the ClimateBC interface. By generating an input paleoclimate data set for the software, the climate variable calculations are held consistent. Also, by using the same lat/long/elevation input files, the resolution and actual geographic locations of the data points can also be standardized. Finally, and probably most importantly, this also saves the time and programming required to manually compute the variables.

All the paleoclimate data images below are presented as anomaly data (based on present-day normals). To generate actual values, the data was run through ClimateBC (see below). Anomaly values are shown from colder (blue) to warmer (red).

Raw GCM Data (Click for larger image)	Creating paleoclimate input files for the ClimateBC software is a relatively straightforward procedure. The coarse scale monthly paleoclimate files can be easily and quickly interpolated in ArcGIS to generate global coverage raster files. This must be completed for both the temperature and precipitation data for each month of each of the two years in question (6,000 and 21,000 YBP). The images on the left represent the raw GCM data (1 deg. lat/long resolution) and the interpolated raster image for the mean monthly temperature, July, 21,000 YBP.
Interpolated GCM Data (Click for larger image)
(Click for larger image)	Once the 24 raster interpolations have been created for each study year, the climate data may then be extracted. In ArcGIS, it is possible to extract the raster data into a single database for specific locations by overlaying a spatial point file. This point file must match identically the necessary input locations for the ClimateBC software (shown as white squares).

The extracted paleoclimate grid can now be used in the ClimateBC application to generate high-resolution paleoclimate maps for the study area (shown in the images below). Note that the values shown in these images represent actual temperature and precipitation values. They are no longer anomaly values, as ClimateBC has generated actual values based on present-day normals.

21,000 YBP
(Click for
larger image)

6,000 YBP
(Click for
larger image)

Paleoclimate Data: mean annual temperature (MAT)

21,000 YBP: Temperature Range (deg. C) = -35.6 to 18.7 (blue to red)
6,000 YBP: Temperature Range (deg. C) = -14.4 to 25.2 (blue to red)

21,000 YBP
(Click for
larger image)

6,000 YBP
(Click for
larger image)

Paleoclimate Data: mean annual precipitation (MAP)

21,000 YBP: Precipitation Range (mm/year) = 50 to 9,200 (red to blue)
6,000 YBP: Precipitation Range (mm/year) = 50 to 11,300 (red to blue)

4) ClimateBC input and output variables (Top)

The climate data used for this project contains 2 geographic variables which were input into the software:

Elevation (input)
XY Position (Lat/Long)

The output data from the software also contains 19 climate variables (described in more detail in the software help files):

MAT - mean annual temperature
MWMT - mean warmest month temperature
MCMT - mean coldest month temperature
TD - temperature difference between MCMT and MWMT (continentality)
MAP - mean annual precipitation
MSP - mean summer precipitation
AH:M - annual heat moisture index
SH:M - summer heat moisture index
DD<0 - degree days below 0C
DD>5 - degree days above 5C
DD5-100 - Julian date on which DD>5 reaches 100
DD<18 - degree days below 18C (not used in this analysis)
DD>18 - degree days above 18C (not used in this analysis)
NDDF - number of frost-free days
FFP - frost-free period
bFFP - beginning of the frost-free period (Julian date)
eFFP - end of the frost-free period (Julian date)
PAS - precipitation as snow
EXT - extreme minimum temperature

Hamann (2006) suggests that averaged climatic variables such as mean annual temperature and growing degree days should be included in such climate-based ecological analyses as they may drive species ranges by influencing competitive advantages between species, individuals, or populations. Furthermore, extreme climate variables such as extreme cold temperature must also be included, as they have the capacity to directly affect individuals (through frost damage, etc.) (Parmesean et al. 2000, as cited in Hamann 2006). This creates a problem when attempting to reduce climatic variables to simplify the simulation input. This is discussed further in 'Statistical Analysis'.

The abiotic data is further complicated by the difference in the nature of some abiotic data. For instance, soil and topographic variables remain spatially constant (ie. they don't move...or at least not much) for the timescales involved in this study. However, other niche variables such as temperature, precipitation, etc. are spatially dynamic on both long and short timescales (Jackson & Overpeck 2000). Major climatic fluctuations can occur both seasonally and gradually over decades, centuries, and millennia.

For the statistical analysis, some of the variables output by the ClimateBC software could not be used to to null values. For example, due to the extremely cold temperature in some of the paleoclimate data, calculations of growing degree-days and frost periods were not possible. For this reason, they have been excluded from some of the analysis.

5) Ecosystem Classification Data (Top)

As part of the baseline data set, it is necessary to assign known ecosystem classifications to the present-day climate data. This will serve as a calibration data set for part of the subsequent statistical analysis. For this, a combination of Alberta and British Columbia ecosystem classifications was paired with the baseline data.

The British Columbia ecosystem classes are based on the 2006 Biogeoclimatic Ecosystem Classifications (BEC), developed by BC Forests. This is a hierarchal classification system, dividing the province subsequently into the 14 zones, 97 sub-zones, and 152 variants. These divisions and sub-divisions are based on vegetation patterns, though it also incorporates climatic and site characteristics in its subdivisions (Pojar & Klinka 1987). For more information, consult the BEC Website.

The Alberta ecosystem classes have been developed by the Alberta Government -Tourism, Parks, Recreation and Culture (TPRC), and uses similar hierarchal divisions to the B.C. system: 6 regions, 21 sub-regions, and approximately 90 seed zones. Region and sub-region classifications are based largely on climatic and other abiotic conditions. Seed zone classes are based largely on genetic criteria (Natural Regions Committee 2006). For more information, consult the Alberta TPRC Website.

For a complete description of the ecosystem classifications and structures for both B.C. and Alberta, including the abbreviations used on this website, click here (Excel format).

Because these ecosystem classifications were specifically developed for use in Alberta and B.C., their applicability to more extensive geographical areas is questionable. However, due to the general southward shift in climatic conditions in the early Holocene (21,000 YBP), the ecosystems do provide a reasonable range of classifications for this area in this time period. This is discussed further in ‘Future Directions’.

There is another issue with using these two sets of ecosystem classifications. Between the two systems, there are certainly redundant classifications—similar or identical ecosystem structures for which Alberta and B.C. have different classes. This issue is also discussed in ‘Future Directions’.

<<<< Back to Project Overview | Top of this page | Continue on to Statistical Analysis >>>>

References:

Guisan, A. and N. E. Zimmermann (2000). "Predictive habitat distribution models in ecology." Ecological Modelling 135(2-3): 147-186.

Hamann, A. and T. L. Wang (2006). "Potential effects of climate change on ecosystem and tree species distribution in British Columbia." Ecology 87(11): 2773-2786.

Jackson, S. T. and J. T. Overpeck (2000). "Responses of plant populations and communities to environmental changes of the late Quaternary." Paleobiology 26(4): 194-220.

Kutzbach, J. E. (1988). "Climatic Changes of the Last 18,000 Years - Observations and Model Simulations." Science 241(4869): 1043-1052.

Kutzbach, J. E. and P. J. Guetter (1986). "The Influence of Changing Orbital Parameters and Surface Boundary-Conditions on Climate Simulations for the Past 18000 Years." Journal of the Atmospheric Sciences 43(16): 1726-1759.

Natural Regions Committee 2006. Natural Regions and Subregions of Alberta. Compiled by D.J. Downing and W.W. Pettapiece. Government of Alberta. Pub. No. T/852.

Newman, J. A. (2006). "Using the output from global circulation models to predict changes in the distribution and abundance of cereal aphids in Canada: a mechanistic modeling approach." Global Change Biology 12(9): 1634-1642.

Overpeck, J. T., L. C. Peterson, et al. (1989). "Climate Change in the Circum-North Atlantic Region during the Last Deglaciation." Nature 338(6216): 553-557.

Parmesan, C., T. L. Root, et al. (2000). "Impacts of extreme weather and climate on terrestrial biota." Bulletin of the American Meteorological Society 81(3): 443-450. (as cited in Hamann, 2006).

Pojar, J., K. Klinka, et al. (1987). "Biogeoclimatic Ecosystem Classification in British-Columbia." Forest Ecology and Management 22(1-2): 119-154.

Rehfeldt, G. E., C. C. Ying, et al. (1999). "Genetic responses to climate in Pinus contorta: Niche breadth, climate change, and reforestation." Ecological Monographs 69(3): 375-407.

Wang, T., A. Hamann, et al. (2006). "Development of scale-free climate data for western Canada for use in resource management." International Journal of Climatology 26(3): 383-397.

Williams, J. W., D. M. Post, et al. (2002). "Rapid and widespread vegetation responses to past climate change in the North Atlantic region." Geology 30(11): 971-974.