Close Window

Oregon State University: "The use of synoptic map-derived climatologies to identify analogs for forecasting situations"

Final Report

The use of synoptic climatologies for describing weather map types has been common among forecasters for decades. Until the advent of high-speed, readily-available computers, however, most such climatologies were very subjective. The Oregon Climate Service at Oregon State University recently developed a map-type climatology for 500 mb maps for a 46-year period. Using a 12 x 12 subset of the NMC 500 mb height octagonal grid, gridded data were divided into monthly data sets. Correlation coefficients were computed for each pair of maps in each monthly data set. A map type was then defined as the day which has the most cross correlations with other days greater than or equal to an R value of 0.80. The days which met the criteria were then designated as map type 1 and removed from the same. Successive map types were defined using the same technique until there were no correlations which met the R value cutoff criteria.

Using the map types obtained in this manner, Oregon State University's Partners Project consisted of a comparison of the current day's data with those of other days throughout the 46-year period using a 12 x 12 subset of the NMC 500 mb height octagonal grid. Later the grid was reduced to a smaller grid to limit the analysis to areas in the vicinity of Oregon. Gridded data for a given date were correlated with all previous days whose Julian day was within 45 days of the given day. Correlation coefficients (R values) with the given day's map were computed for every such day. This process identifies the shape of the map, but not the magnitude. To obtain magnitude correlations as well, the 20 days with the highest correlations were saved and then sorted by mean 500 mb height. The best analog days were those which matched well in both mean height and correlation value. In general, correlation values exceeding .90 and means within 50 meters were considered desirable.