|
 |
The ability to forecast precipitation values using an interpolation method is dependent on having access to current values from all of the weather stations in the GVRD and then being able to run a model on the fly based on these current values. The ability to gather this data at a daily time scale is impossible. In the interpolation modeling of average monthly values, 43 stations were used. This number of measured points was found to have limitations. It is therefore unreasonable to believe that an accurate forecasting model could be run using even fewer points.
Another way to go about creating an interpolated forecasted precipitation surface would be to calculate daily historical mean values. The data for this is available, but as was discussed in previous sections, the correlation between precipitation on the same day of the year over 15 years is practically non-existent. That is, there could be an average value of 8 mm on January 1st, but a range between 0 and 50 mm. This is highly inaccurate and would involve the creation of 365 maps and a very large database.
Cannon et al. (2002) have done work on map-pattern and classification for possible use in precipitation forecasting, but this again requires many stations with current values available. They have identified 25 distinct climate patterns that may occur in British Columbia and the northeastern Pacific Ocean, based on the clustering of decades of climatological data. A forecasting method would then select the pattern representative of current conditions and a map would be created. This would also require an extensive database as well as complex programming skills, which we do not have. Another difficulty with this method would be the fact that it would demand unreasonable computational capacity for the current work.
 |
back to
top |
 |