Removing the relocation bias from the 155‐year Haparanda temperature record in Northern Europe

Abstract

The village Haparanda in northern Sweden hosts one of the longest meteorological station records in Europe depicting climate conditions in the subarctic. Since the station was relocated several times, moving gradually from urbanized to more rural areas, the record is likely biased by anthropogenic influences. We here assess these influences and demonstrate that even in villages urban heat island biases might affect the temperature readings. We detail a method to quantify this bias and remove it from the long Haparanda station record running since 1859. The correction is based on parallel temperature measurements at previous station locations in Haparanda. These measurements revealed a distinct urban warming pattern, largest in minimum temperatures during summer, since urban geometry and its heat capacity amplify additional warming and night‐time heat release, especially during a period of high insolation and calm conditions. Due to the station movement from the village centre to the outskirts, the net correction results in an additional warming trend over the past 155 years. The trend increase is most substantial for minimum temperatures (+0.03 °C 10 years−1). Maximum and mean temperatures are less affected (+0.01 °C 10 years−1). An increase in trend is even more severe if the 20th century is regarded exclusively, displaying a rise in annual mean temperature trend by +0.03 °C 10 years−1 and +0.07 °C 10 years−1 in annual minimum temperatures, respectively. Our approach of using actual temperature differences between historical station locations did not take into account other factors that might have influenced the data such as changes in instruments or observers. The presented adjustment for temperature residuals caused by a change in historical station locations might be considered as a minimum approach to improve long‐term temperature readings. The adjustment of the Haparanda station record results in an increased warming trend, thereby adding critical information to the evaluation and explanation of 20th century anthropogenic warming.

A 1286‐year hydro‐climate reconstruction for the Balkan Peninsula

“…Potential reasons for this discrepancy in the low-frequency domain include (i) a nonlinearity or decoupling of the growth/climate response on longer time scales (Fritts 1976); (ii) the more white precipitation spectrum compared to the red temperature spectrum (Ault & St. George 2010); (iii) potential biases in longterm trends inherent to instrumental data, e.g. caused by station relocations or changes in their operating system (Dienst et al 2017); and (iv) the difficulty in preserving low-frequency variance in TRW (Esper et al 2002). As also the reliable instrumental target extends only over 55 years back to 1961, the full spectrum of potential SPI variability, and particularly the low-frequency end of that spectrum over longer time scales remains unaddressed.…”

Quality control and homogenization of monthly extreme air temperature of Ukraine

“…HOMER has been widely used for the homogenization of different climatological variables. More than 30 published papers (to the best of our knowledge) have mentioned the use of this software for the homogenization of different variables such as air temperature (e.g., Mamara et al ., ; Dienst et al ., ), precipitation (e.g., Coll et al ., ; Noone et al ., ; Prohom et al ., ) and wind speed (e.g., Laapas and Venäläinen, ). In our study, we applied a two‐cycle process, as suggested in Mestre and Aguilar ().…”

Determination of the urban heat island intensity in villages and its connection to land cover in three European climate zones

“…A network of temperature sensors was established in each village and its surroundings ( Fig. 1), originally to detect the urban warming effect on historical temperature measurements (Lindén et al 2015b, Dienst et al 2017. We closely followed the WMO guidelines for sensor installation in urban climate studies (WMO 2008), though a completely standardized placement, e.g.…”

“…Therefore, ap proaches to homogenize temperature data have be come vital to ensure their reliability (Brunet et al 2006, Venema et al 2012. Albeit aiming at re moving all non-climatic impact in the temperature data, homo genization and correction of time series might as well have inadvertent implications, like the recovery of urban warming bias (Zhang et al 2014, Dienst et al 2017, emphasizing the need to adjust records with special diligence. Cities contain the highest potential to affect the representativity of meteorological station measurements due to their large extension, but even smaller towns and villages influence the observations.…”