A Modified Persistent Anomaly Index Applied to Present and Future Climates

Persistent anomalies (PAs) in tropospheric geopotential heights are associated with high impact weather events such as heat waves, cold air outbreaks, droughts, and floods. Thus, it is important to study the frequency of PAs as well as how their occurrence and magnitude will change with climate.

Dole and Gordon (1983, hereafter DG) define a PA as a height anomaly that exceeds a fixed threshold that persists for a specified duration. An analysis of the present day climatology, following DG, using a fixed height threshold reveals a strong seasonal and regional dependence in the occurrence of PAs, with an abundance of events in the winter but a near absence of summer events. Summertime events are societally important, due to their association with droughts and heat waves, yet a PA analysis using a fixed threshold misses most events in the warm seasons, when the variability of geopotential heights is reduced. In order to capture impactful warm-season PAs, we propose to identify PAs using a threshold that varies with the season. This new method differs from that of DG by employing a height-magnitude threshold determined from the seasonally dependent standard deviation of 500-hPa heights. This modification allows anomalous events to be identified relative to their season and region of interest.

We apply this revised index to the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-Interim reanalysis dataset. These data comprise 6-hourly, 500-hPa geopotential height fields from January 1, 1979 to December 31, 2016. Height anomalies are defined as the departure from a climatological annual cycle, calculated as the 8-week low-pass filtered, 38 year average daily mean calculated at each grid point. PAs are identified from anomalies with an inverse sine-of-latitude weighting (following DG). Heights are de-trended over the period of the record, to remove contamination from climatic trends.

Regions of maximum PA frequency, consistent with those described by DG, are the northeast Pacific over the Aleutian Islands, the northeast Atlantic, and northwest Russia. These regions vary depending on the PA criteria and the method of identification (e.g. those used by Tibaldi and Molteni, 1990, and Pelly and Hoskins, 2003). The resulting winter PA climatology of the ERA-Interim, using the unmodified DG index, identifies regions of maximum and minimum occurrence that closely resemble the results of DG (their Figure 7). In addition, the summer PA climatology identified very few events (not discussed in DG).

We use this modified DG method to investigate changes in current and future PAs across the seasons using analyzed data and climate model simulations of current and future climate. The climatologies are compared in order to identify changes in PA frequency, intensity, duration, and geographical location between models and analysis and between simulated current and future climates.