Rt Hon Lord Lilley

    Lord Lilley: 1. To ask Her Majesty’s Government, further to the Written Answer by Lord Callanan on 6 October (HL8378), what reasons they have for assuming that the set of residuals is stationary.

    2. To ask Her Majesty’s Government, further to the Written Answer by Lord Callanan on 6 October (HL8377), whether they will place a copy of a quantile-quantile plot of a GEV distribution against the distribution of June maximum temperatures at Verkhoyansk during 1926–2020, that includes data from 2020 in particular, in the Library of the House.

    Lord Callanan:

    1.

    The assumption of stationarity of the residuals is well established in the peer reviewed literature. Recent examples of this can be found in the following papers:

    Kew, Sarah F., Sjoukje Y. Philip, Geert Jan van Oldenborgh, Gerard van der Schrier, Friederike EL Otto, and Robert Vautard. “The exceptional summer heat wave in southern Europe 2017.” Bulletin of the American Meteorological Society 100, no. 1 (2019): S49-S53.

    Yiou, Pascal, Julien Cattiaux, Davide Faranda, Nikolay Kadygrov, Aglae Jézéquel, Philippe Naveau, Aurelien Ribes et al. “Analyses of the Northern European summer heatwave of 2018.” Bulletin of the American Meteorological Society 101, no. 1 (2020): S35-S40.

    Leach, Nicholas J., Sihan Li, Sarah Sparrow, Geert Jan van Oldenborgh, Fraser C. Lott, Antje Weisheimer, and Myles R. Allen. “Anthropogenic influence on the 2018 summer warm spell in Europe: the impact of different spatio-temporal scales.” Bulletin of the American Meteorological Society 101, no. 1 (2020): S41-S46.

    Further to the written answer provided on 6 October (HL8377), the plot placed in the Library of the House shows that the GEV distribution is a good fit to the data, which supports the assumption that the distribution of residuals may be adequately modelled as stationary.

    2.

    Further to the written answer provided on 6 October (HL8377), the plot placed in the Library of the House uses data over 1926 – 2019. The 2020 value is not included in the fit of the GEV itself, as is standard scientific practice to avoid biasing the fit by the extreme value of interest. It is well established that the selection of a series in which a very large extreme has occurred means that the return time of this value is likely to be much larger than the length of the dataset itself. This means the time series would give an artificially low view of the return time compared to reality. However, as requested, we have provided an additional quantile-quantile plot that does include the 2020 value in the fit – which we provide subject to these caveats.

    HL9546 Plot (pdf, 456.1KB)

    Share.