Mesoscale representativeness of rainfall measurements for Winnipeg

Open Access
Publisher Website
Google Scholar
Abstract
For a variety of applications, rainfall measurements from the AES (Atmospheric Environment Service) weather observing site in Winnipeg are extrapolated to ungauged points ‐ locations where the information is desired but not measured. Sometimes, they are used as the area‐average for the city. This study examined the spatial representativeness of point rainfall measurements for Winnipeg for two accumulation periods ‐ one day and one month. Eight to nine years of data from a mesoscale network were used to specify isotropic exponential correlation functions and typical coefficients of variation. Taken together, they suggest: (a) probable errors for amounts extrapolated over distances of 10 to 32 km of ± 126 to 165% for estimates of daily rainfalls and ±36 to 48% for estimates of monthly accumulations, and (b) probable errors for area‐average amounts with station densities of 1 per 54 km2 to 1 per 707 km2 of 21 to 85% for estimates of daily values and 6 to 24% for estimates of monthly totals. The practice of using data from a single point observation site, the AES site, to approximate Winnipeg's monthly area‐average rainfalls may provide estimates with tolerable errors. However, basing daily area‐average rainfall estimates on data from just one observation site will likely result in large errors and this practice should be avoided. Extrapolating either daily or monthly point measurements over 10 km or more from the AES site to ungauged points should also be avoided owing to the high probability of unacceptably large estimation errors.