DATA AND METHODS

There were four main steps involved in creating animated climate variable coverages:

1.     Creation of latitude, longitude and elevation data

2.     Generation of climate variables using ClimateBC

3.     Conversion of climate variables into pixilated maps

4.     Animation of climate variable maps

Creation of latitude, longitude and elevation data

ClimateBC can generate climate variables for any year between 1901 and 2000 given appropriately-scaled latitude, longitude and elevation data points.  It is possible to create these data points by adding xy coordinates to a vectorized digital elevation model (DEM).

1.     A DEM raster with a cell size of 769 meters was downloaded from the Geography data drive.  This resolution proved to be cumbersomely high for efficient data processing and unnecessary for the scope of the project; therefore the first step was to resample the DEM to 5000 meters.  Five kilometers is approximately equivalent to 3 arcmin resolution in southern British Columbia.

2.     The resampled DEM was converted to a feature file and saved as a vector shapefile in order to capture elevation values as points.

3.     The new shapefile was in an Albers projection, so it needed to be re-projected as latitude/longitude.  The shapefile was reprojected to North American Datum 83 (NAD83).

4.     Coordinates were added to the NAD83 shape file by creating new ‘x’ and ‘y’ fields in the attribute table and plugging in the appropriate processing code from the ‘help’ menu.

Creation of climate variables using ClimateBC

Using latitude, longitude and elevation data as an input, ClimateBC can generate the following output variables:

Table 1: Directly calculated and derived variables that are generated by ClimateBC

          All of these climate variables are relevant to tree growth and distribution.  There was not enough time to create animations for all 16 variables, so three variables were selected for analysis in this project: mean annual temperature (MAT), number of frost-free days (NFFD) and summer heat:moisture index (SH:M).  The summer heat:moisture index is a scaled, decadally-averaged ratio between mean warmest month temperature and mean summer precipitation.  Particularly hot, dry years have high SH:M indices while cool, wet years have very low indices. 

To create climate variables using ClimateBC:

1.     The attribute table from the NAD85 5k resolution shapefile was opened using Excel.  The data was reformatted into five columns with the headers ‘id1’, ‘id2’, ‘lat’, ‘lon’, ‘el’ as per ClimateBC input standards.

Figure 1: Csv-style data variables needed for ClimateBC

2.     Within ClimateBC, it is possible to calculate climate variables for a one or ten-year time period anytime between 1901 and 2000; it is also possible to generate climate coverages for 2020, 2050 and 2080 using a variety of climate model projections.  For the purposes of this project, ten-year averaged time periods from 1901 to 2000 were selected one at a time.  The output for each was saved as a csv.

Figure 2: ClimateBC user interface.  This is an image of ClimateBC version 2, which is not capable of calculating yearly nor decadal variables.  Version 3,  which is capable of generating these variables, was used for this project.

Conversion of climate variables into pixilated maps

The csv outputs from ClimateBC are not particularly exciting unto themselves, but when transformed into colorful maps they provide the viewer with enormous quantities of climatic information in an easily interpretable format.  For this project, mean annual temperature (MAT), summer heat:moisture index (SH:M) and number of frost free days (NFFD) were chosen for mapping and analysis because of their importance as correlates of vegetative growth.

1.     Before inputting the ClimateBC csv decadal outputs into ArcGIS, it was necessary to delete all symbols from the climate variable headings (for example, delete the ‘<’ from DD<18) and to convert all null values (symbolized as ‘.’) into numbers that would not be confused for real values (something like ‘-9999’).  ArcGIS is not able to interpret csvs that have symbols and non-numerical values.

2.     The edited ClimateBC csv was imported as xy data.

3.     The shapefile was interpolated to a raster format for each of the three chosen variables by kriging.  Kriging is a statistically-enhanced interpolation process that takes into account the relationship between the measured points.

4.     The output maps depicted the selected climate variables at a 5 kilometer resolution.  Initially, the map edges were not specific to British Columbia, but rather included large parts of Alberta and the Pacific Ocean.  It was necessary to multiply the output maps by a mask of BC using the raster calculator in order to exclude the unwanted areas.

5.     In order to prevent errors and save time, the data import, kriging and masking tasks were automated using ‘modelbuilder’.  The csv for each decade needed to be specified at the beginning of the model, and an output name for each map needed to be specified at the end, but otherwise the model ran itself.

Figure 3: The ArcGIS model created to automate the xy data input, kriging and masking processes for the output of 30 climate variable maps.

6.     An appropriate symbology was chosen for each of the three different climate variables.  Regular scaling was achieved by manually creating intervals that encompassed the full data range for all ten maps per climate variable.  This symbology was then exported to the other maps depicting the same variable.  The selected data ranges were:

a.      MAT: Data range was -9.5 to 10.5.  However, the histogram showed very few values below -6 on any of the maps, so -8 chosen as the lower boundary and 10.5 as the upper in the final symbology.

b.     NFFD: Data range was from 36 to 335.  This range was used in the symbology.

c.     SHM: Data range was from 2.6 to 181.  However, most of the values were lower than 140 in all of the histograms, so 140 was used as the upper boundary and 3 as the lower in the final symbology.

Figure 4: ArcGIS maps for mean annual temperature.

Animation of climate variable maps

While the decadally-averaged climate variable maps can be scrolled through using ArcGIS, it is easier to see climate trends when the images are animated into an automated slide show.

1.  Maps were exported individually as .gif files.

2.  Gif files were imported into Corel R.A.V.E. 2.0, resized, and then made into an animation.

3.  The animations for each climate variable were exported as Flash movies.