Ongoing Scientific Assessment of the 2010 Western Russia Heatwave


Draft - Last Update: 3 November 2011

Disclaimer: This draft is an evolving research assessment and not a final report. Comments are welcome. For more information, contact Dr. Martin Hoerling (martin.hoerling@noaa.gov)

Western Russian July Temperature Departures
Figure 4: (Top) Observed time series of western Russian July temperature anomalies for the period 1880-2011. Shown are the time series using 5 different gridded analyses, and the 1880-2009 linear trends for each is indicated. The domain for averaging is 50°N-60°N,30°E-55°E. (Bottom) Same as top, but for the period 1979-2011 together with linear trends for the period 1979-2009.
CMIP3 and Observed July Temperature Departures
Figure 5: (Top) Observed (CRU unadjusted, blue) and simulated (ensemble mean of 22 CMIP3 model simulations, black) time series of western Russian July temperature anomalies for the period 1880-2011. The 1880-2009 linear trends for each is indicated. The shaded area represents the envelope of positive and negative monthly mean temperature extremes based on 22 CMIP3 model simulations. The domain for averaging is 50°N-60°N,30°E-55°E. (Bottom) Same as top, but for the period 1979 to 2011 and linear trends for the period 1979-2009.

Detection of Warming over Western Russia

Two different questions have been raised in the peer-reviewed literature regarding the July 2010 heat wave over western Russia. One concerns the physical causes for the extreme magnitude of the heat wave, and whether its intensity could have been anticipated based on prior trends or prior knowledge of specific boundary conditions a month or more in advance (e.g., Dole et al. 2011; Barriopedro et al. 2011). The other question concerns the changing probability of achieving a record-breaking heat wave (whatever its magnitude) associated with a warming trend (e.g., Rahmstorf and Coumou 2011; Barriopedro et al. 2011). Both questions include consideration of the potential role of anthropogenic warming, at least indirectly --- on the one hand the possible contribution of GHG warming to the magnitude of the 2010 Russian heat wave, and on the other hand the effect of GHG warming in producing a regional warming trend that increased the probability of a record-breaking heat wave in 2010.

The principal conclusion by Dole et al. is that the extreme magnitude of the 2010 Russian heat wave was mainly due to internal dynamical processes (associated with atmospheric blocking), and that it was very unlikely that warming attributable to GHG forcing contributed substantially to the heat wave's magnitude. Rahmstorf and Coumou concluded that a strong warming over western Russia (which they attribute primarily to GHG forcing) multiplied the likelihood of a record heat wave. They estimated an 80% probability that the 2010 July heat records in Moscow would not have occurred without climate warming. Barriopedro et al., on the other hand, conclude that the magnitude of the 2010 event was so extreme that despite GHG warming, the likelihood of an analog over the same region remains fairly low at this time. This is consistent with estimates in Dole et al. (2011), which showed a very low probability of an event of this magnitude in 2010, but a rapidly increasing likelihood of crossing given thresholds in future climate, based on results from CMIP3 model runs.

Toward attempting to reconcile these conclusions, Figure 4 presents various time series of western Russia July surface temperature. The top panel repeats the observed time series for 1880-2011, and also shows in an inset graph the post-1979 time series. The linear trends in surface temperature for 1979-2009 are also calculated and shown. Note that whereas the time series are now extended to update conditions thru July 2011, we calculate the trends thru 2009 only. This is done in order to meaningfully address whether one could have, from available antecedent observations, anticipated the 2010 heat wave.

Also presented in Figure 5 are the time series of July western Russia surface temperature from the 22 models of the IPCC Fourth Assessment (AR4) that were subjected to the time evolving GHG, aerosol, and in many cases solar and volcanic forcing. The black curve, representing the 22 model ensemble average, is our best estimate of the externally forced signal. The gray band denotes the extreme occurrences of July conditions among the 22 members, and illustrates the effects of internal climate variability (noise). Plotted in blue is the observed (CRU-unadjusted) July surface temperature time series.

These additional analyses provide insight on the following questions:

  1. What is the GHG warming signal over western Russia during July?
  2. Has a change in the region's July temperatures been detected?
  3. What role did GHG warming play in the 2010 Russian heat wave?

Figure 4 indicates that all the available data sets are consistent in indicating a positive surface temperature trend over western Russia, with magnitudes ranging from +0.20°C/decade to +0.49°C/decade for 1979-2009. These values are appreciably larger than the temperature trends computed using the entire period 1880-2009. These 1979-2009 values are consistent with the area-averaged warming rate of +0.45°C/decade for this same period based on satellite remote sensing (see Fig. 5 of Rahmstorf and Coumou 2011). For all data sets, the area-averaged warming rate is appreciably less than that at the single station site of Moscow, indicating the lack of representativeness of a single station for inferring the evolving conditions over western Russia as a whole, as we previously noted.

An important issue is assessing to what extent this increased warming rate after 1979 is attributable to noise and statistical sampling (since similar magnitude 30-yr warming trends occurred in the early 20th Century (see Figure 3)), and to what extent the GHG warming signal has increased in magnitude during the same period. Clearly, the magnitude of the GHG warming signal is one factor very relevant to assessing the extent to which the 2010 heat intensity resulted from anthropogenic warming. The signal strength is also very relevant in assessing the statistical likelihood for extremes exceeding a particular threshold, especially for the tails of a frequency distribution of July temperatures whose probabilities will be most sensitive to a mean warming.

Regarding the posed questions above, we find the following:

  1. The surface temperature trends from the ensemble mean model simulations are +0.06°C/decade for the 1880-2009 period, and +0.40°C/decade for the recent 1979-2009 period. These are our best estimates of the anthropogenic warming signals at this time, and the AR4 simulations indicate an accelerated warming signal over western Russia in the recent decades.
  2. Despite a strengthened GHG warming signal, the intrinsic variability of observed western Russia temperature during July is too large to permit detection of a change in temperatures, with high confidence, at this time. Therefore, attribution of these regional July trends to human influences is premature at this time. Nonetheless, a signal of warming due to GHG forcing likely exists, though likely having appreciably smaller magnitude than the estimated intrinsic variability.
  3. Figure 5 indicates that the 2010 observed heat wave magnitude is greatly different from that expected by GHG forcing alone (compare black and blue curves). While recognizing the presence of a moderate amplitude GHG signal, the majority of the event's magnitude was very likely due to internal variability.

This is also consistent with the fact that the extreme values of western Russia temperatures occurring in individual model runs were, from time to time, as large as the observed 2010 case. However, there is no indication of a trend in the frequency of occurrence of such extreme model events, at least thru 2009. These analyses support the conclusions of Dole et al. and Barriopedro et al., and would not be especially sensitive to physical and model-based estimated ranges in uncertainty in the GHG signal magnitude. The analysis of Figs. 4 and 5 also would support the notion of an increased probability for the exceedance of a threshold in the region's July temperatures as argued in Rahmstorf and Coumou. However, the quantification of that change in probability, is very sensitive to the magnitude of inferred background warming.