Cluster analysis is utilized to define the circulation regimes. Regimes are identified from the k-means algorithm, applied to running 5-day means of the combined anomalies of 500 hPa geopotential height and the 200 hPa zonal wind over the extended Pacific – North America (PNA) region. We show that the circulation regimes provide statistically significant modulation of observed precipitation over the continental US, as well as shifts in storm tracks (measured at 300 hPa and 850 hPa) and atmospheric rivers over the extended PNA region.
For the precipitation, we examine regime-specific geographical patterns of shifts in the mean, the normalized mean, and the occurrence of extreme values. We identify regions in which the ratio (R) of frequency of occurrence of extremes (5th, 10th, 90th and 95th percentile) to climatological frequency of occurrence in each regime is significantly high. Storm tracks are measured by covariances of high-pass (periods less than 10 days) fields: meridional wind and temperature at 850 hPa and meridional wind variance at 300 hPa. Atmospheric rivers are determined from an algorithm that identifies intense, narrow plumes of moisture flux. Patterns of vertically integrated moisture flux over the wider region are also used to interpret the results.
We find that a few of the circulation regimes (cluster centroids) affect extreme precipitation over multiple regions of the US. The Alaskan Ridge regime significantly increases the likelihood of extreme precipitation not only over the Southwest but also the upper Midwest, while the Pacific Trough increases this likelihood over the west coast and over southern Florida.
Data used in this work include output from CPC Unified Gauge-Based Analysis of Daily Precipitation over CONUS, as well as 850 hPa temperature, 250 hPa horizontal wind, and 500 hPa geopotential height (Z) from ERA Interim reanalysis. Our analysis has been applied for December-February (daily) over the time period of 1980-2014.