Extreme temperatures can cause severe disruptions to society, from negative health consequences to infrastructure damage. Accurate and timely weather forecasts contribute to minimising these detrimental effects, by supporting early-warning systems. In this context, information on the expected performance of the forecasts is valuable. Here, we investigate whether there is a relationship between the persistence of atmospheric circulation patterns in the Euro-Atlantic sector and forecast skill for temperatures and temperature extremes in Europe. We first apply an objective method to compute the persistence of large-scale atmospheric patterns in European Centre for Medium-Range Weather Forecasts (ECMWF) subseasonal retrospective forecasts. We find that the forecasts successfully predict atmospheric persistence up to time-scales of approximately two weeks. We next investigate the relationship between the persistence of an atmospheric state and the practical predictability of temperature in terms of the error in surface temperature forecasts. The relationship between the two varies depending on season and location. Nonetheless, in a number of cases atmospheric persistence provides potentially valuable information on the practical predictability of temperature. We specifically highlight the cases of wintertime temperature forecasts up to three weeks lead time and wintertime cold spells up to roughly two weeks lead time.

In this study, the instability of extreme temperatures is defined as the degree of perturbation of the spatial and temporal distribution of extreme temperatures, which is to show the uncertainty of the intensity and occurrence of extreme temperatures in China. Based on identifying the extreme temperatures and by analyzing their variability, we refer to the entropy value in the entropy weight method to study the instability of extreme temperatures. The results show that TXx (annual maximum value of daily maximum temperature) and TNn (annual minimum value of daily minimum temperature) in China increased at 0.18 degrees C/10 year and 0.52 degrees C/10 year, respectively, from 1966 to 2015. The interannual data of TXx' occurrence (CTXx) and TNn' occurrence (CTNn), which are used to identify the timing of extreme temperatures, advance at 0.538 d/10 year and 1.02 d/10 year, respectively. In summary, extreme low-temperature changes are more sensitive to global warming. The results of extreme temperature instability show that the relative instability region of TXx is located in the middle and lower reaches of the Yangtze River basin, and the relative instability region of TNn is concentrated in the Yangtze River, Yellow River, Langtang River source area and parts of Tibet. The relative instability region of CTXx instability is distributed between 105 degrees E and 120 degrees E south of the 30 degrees N latitude line, while the distribution of CTNn instability region is more scattered; the TXx's instability intensity is higher than TNn's, and CTXx's instability intensity is higher than CTNn's. We further investigate the factors affecting extreme climate instability. We also find that the increase in mean temperature and the change in the intensity of the El Nino phenomenon has significant effects on extreme temperature instability.