How to select climate data for calculating growth-climate correlation

We identified the climate zones where the climate has highest variation similarity to aid to climate data selection. The calculation of climate-growth correlations is the analytical foundation to study climate change influence on tree growth in dendrochronology. However, the majority of climate data used in climate-growth correlation analyses are not directly recorded on the sample sites, but obtained from nearby weather stations. We used a sample site in Saihanba region as a case study to address how correlation bias may occur if nearby climate products have no high correlation with the climate in the sample site. Temperatures in the sample site and from other data resources were highly correlated, suggesting that small potential bias in growth-temperature correlations when using temperatures from nearby climate stations. However, precipitation had large spatial variability, resulting in low correlation between precipitation of the sample site and precipitation from other resources. Large biases in growth-precipitation analysis would be expected when using precipitation from nearby stations, suggesting that precipitation records should be carefully chosen. To aid in this selection, we used a cluster analysis and multiple data-products across China to identify regions where station climate do and do not reflect accurately site conditions, and classified temperature and precipitation zones where climate has high correlation among grid cells of the same climate zone based on similarity of the macroclimate using a ~ 2.5 km resolution gridded climate dataset. Using climate stations located in the same cluster as the sample sites would help to prevent or reduce correlation biases in growth-climate analyses. The generated temperature and precipitation zones are freely available to download as GeoTIFF files in the online supplementary materials (Fig. 1S and Fig. 2S).