Effect of various climate databases on the results of dendroclimatic analysis

Abstract. The paper deals with the comparison of the time series drawn from different climate databases. We compared the observed data with the modeled data of monthly and seasonal temperature means and precipitation totals. Reliable and longest available time series of such data represent the basic starting point of dendroclimatic analyses. We evaluated the differences in the growth response of spruce derived using different databases of the considered climatic variables. The stem cores used to derive the cross-correlation function were taken from Hårås locality situated in the boreal zone of the Swedish part of Lapland. We compared the observed records from the nearest weather stations situated 18, 40, 70 and 110 km away from the locality with the interpolated data from four modeled temperature databases and four modeled precipitation databases generated by KNMI Climate Explorer. The spatial resolution of the modeled databases was 0.5°  ×  0.5° of latitude and longitude or 1°  ×  1° respectively. The evaluation revealed that in all modeled databases systematic errors of different magnitudes occurred. We also found that the radial increments of spruce correlated more tightly with the temperature than with the precipitation in the area of interest. Hence, in the conditions of the boreal zone, temperature could be a more important factor with regard to tree-ring formation. Because of higher spatial variability seen in precipitation data when compared to temperature data, we conclude that the nearest weather station is the most suitable for dendroclimatic analysis leaning on precipitation. Drawing on these results we recommend that the modeled precipitation and temperature databases examined in our study are used for dendroclimatic analyses within areas featuring a sparse network of weather stations.