Due to a mistake in the calculation of the warm spell duration index (WSDI) and cold spell duration index (CSDI) for the ERAInterim data, the magnitudes of the trends presented in our study are too small. Both indices were calculated using daily mean temperatures instead of daily minimum (for CSDI) and daily maximum (for WSDI) temperatures in comparison with the percentiles TN10 (for CSDI) and TX90 (for WSDI), which leads to much smaller values for the indices and therefore smaller trends. We corrected the calculation and plotted figure 1 again. The most prominent changes occur for warm spell duration index in summer. However, the main results and conclusions of our study are not affected by the error.
Results3.1. Spatial patterns of trends in spells 3.1.1. Warm spells (WSDI) The paragraph on the spatial patterns of warm spell duration index (WSDI) should read: ERA-Interim dataderived trends (shading in figure 1(d)) suggest that warm spells have changed most noticeably over Greenland (up to 4 days/decade). The station-derived (colored circles in figure 1(d)) significant positive trends indicating an increase of warm spells over southern Siberia are reproduced by ERA-Interim, in contrast to the significant positive station based trends over Scandinavia. Both data sets indicate an increase in warm spells over western Russia and parts of the Canadian Archipelago.
AbstractIn the Arctic, climate change manifests with the strongest warming trends on the globe, especially in the cold season. It is under debate if climate extremes change similarly strong. Our study provides detailed regional information about two selected temperature extreme indices in the Arctic, namely warm and cold spells in winter and summer. We analyze their temporal evolution and variability from 1979-2013, based on daily station data and ERA-Interim reanalysis. Calculated trends from both datasets suggest a widespread decrease of cold spells in winter and summer of up to −4 days/decade, with regional patches where trends are statistically significant throughout the Arctic. Winter trends are spatially heterogeneous, the reanalysis also shows small areas with statistically significant increases of cold spells throughout Siberia. Calculated changes in warm spells from both datasets are mostly small throughout the Arctic (less than±1 day/decade) and statistically not significant. Remarkable exceptions are the Lena river basin in winter with a statistically significant decrease of up to −1.5 days/decade and areas in Scandinavia with statistically significant increases of up to 2.5 days/decade in winter and summer (again from both datasets). From the analysis of spell lengths, we find that there are no shifts from longer to shorter spells or vice versa with time, but long cold spells (events lasting for more than 15 days) disappear almost completely after the year 2000. There is a distinct inter-annual and decadal variability in the spells, which hinders the detection of significant trends for all spell categories in all regions.