DATA LIMITATIONS, FURTHER RESEARCH

There is a high degree of error and uncertainty inherent to making data projections on the scale of the entirety of British Columbia.  The initial data resolution was 769 meters, and it was resampled to 5 kilometers for the sake of data-processing ease.  ClimateBC inputted the resampled latitude, longitude and elevation points into a database of climate coverages that had already been interpolated from weather station data that a) has inherent sampling error to begin with; b) is not evenly distributed throughout the province nor throughout time, with a chronic under representation of weather data at high elevations, in the far north, and in the earlier part of the 20^th century; and c) is known to misrepresent precipitation patterns, which fluctuate on a very narrow spatial scale.  ClimateBC uses scale-free interpolation methods that project climate variables across elevation ranges better than most models, but nonetheless, on a five kilometer spatial scale there will be inaccuracies in representing topographically heterogeneous areas.

These data limitations are inevitable within a map series that is designed to show general climate patterns over time over a large spatial area.  For example, using data at a resolution finer than one kilometer would not add accuracy to the animated maps if they are viewed at a provincial scale.  If the intention of the project had been to create accurate climate coverages for particular locations within the province, those places would have had to be enlarged and finer DEM resolution used.  However, the nature of recreating climate coverages using monthly temperature and precipitation averages is that data smoothing occurs; it is simply impossible to generate precise historical climate records.  Fortunately, climate trends are more important than exact variables for many scientific studies pertaining to ecosystem functions over time.

Many further analyses could be done using the process described in this project.  It would be interesting to create climate coverages for the other dozen climate variables generated by ClimateBC.  In particular, mean coldest month temperature would be interesting to examine, as nighttime low temperatures are known to have risen the most dramatically during the twentieth century.  ClimateBC version 3.0 has the capacity to calculate coverages for particular seasons, so a coverage specific to springtime lows would be very interesting in terms of looking at the impact of climate on vegetative growth.  It would also be interesting to extend climate coverages to include the 2020, 2050 and 2080 projections and to add them to the existing animations to see if the projected climate changes appear dramatic in sequence with the decadal averages.