In the article „Climatic warming above the Arctic Circle: are there trends in timing and length of the thermal growing season in Murmansk Region (Russia) between 1951 and 2012?“, written by I. Blinova and F.-M. Chmielewski, 2015 (Int J Biometeorol (2015) 59:693–705; see https://www.researchgate.net/publication/265054773_Climatic_warming_above_the_Arctic_Circle_are_there_trends_in_timing_and_length_of_the_thermal_growing_season_in_Murmansk_Region_%28Russia%29_between_1951_and_2012) trends in timing (beginning B5, end E5 and length L5) of the thermal growing season in the Murmansk region (see black frame in Figure 4) between 1951-2012 are compared for 9 different land cover units. B5 and E5 are very sensitive functions of daily mean air temperature. (B5 is defined as the first day of the year on which the mean daily air temperature is greater or equal 5.0 °C with the condition that the mean temperature of the following 30 days is also greater 5.0 °C. E5 is defined similar but must lie after B5 and „greater“ must be replaced by „less than“. L5=E5-B5.)

They found some significant trends (see, e.g., Figure 1; here the trend of the anomaly of the end of the growing season is shown). The coefficient of determination R2 is only 0.104 (not mentioned in the article). I assume that this long term trend is mainly caused by a breakpoint in the year 2000. Here are my reasons for this assumption: The authors used daily mean air temperatures from the E-OBS datset (version 9.0) (http://www.ecad.eu/download/ensembles/ensembles.php) at 0.25° resolution to calculate B5 and E5. These temperatures were derived by Universal Kriging from station data. The problem here is, that the station density in the Murmansk region is low and the number of stations used for E-OBS data shows a jump downwards at the year 2000 (see Figure 2). Some of the used stations were found to be „non homogeneous“ (see Figure 3, graphs on the right side). Because of the varying and pretty low station density in the Murmansk region, there are many days with large kriging errors (also referred to as standard deviations) (these errors are not mentioned in the article above, but they are part of the E-OBS data)  (see Figure 4 for „onset of growing season“ and Figure 5 for „end of growing season“; please note that the 95% confidence intervals are two times larger than these standard errors which have values up to 5 K in the eastern part of the region).

Therefore I am convinced that the significant slopes (which were found mainly in the middle and eastern part of the region) are caused by inhomogeneities and interpolation errrors in the E-OBS data. The authors should have mentioned this pitfall and should have given some error assessments!

Does anybody agree with me? 

There are a lot of other shortcomings in this article. You can find the discussion of these deficiencies and a more comprehensive discussion of the „E-OBS data issue“ in my Open Review about this article.

Article Climatic warming above the Arctic Circle: are there trends i...

Similar questions and discussions