资源环境科技发展态势分析平台

The Berkeley Earth surface temperature (BETP) project provides gridded global temperature anomaly products using an automated geostatistical approach to adjust station data for systematic biases. Despite its widespread usage, the BETP data set has not been evaluated at the national-scale, especially in data-sparse high latitude environments. This study provides an evaluation of the BETP product across all of Canada using 333 climate stations made available from the homogenized Environment Canada station network (HTcan). Comparison between co-located monthly air temperature anomalies for the two data sets suggests small differences between the two products for mean surface (approximate to 2 m) air temperature. However, the relatively minimal bias in mean temperature is a consequence of contrasting cold and warm biases in minimum and maximum air temperatures, respectively, that are larger but effectively even out when averaged together. The BETP product is shown to exhibit systematic underestimation of recent regional warming in northern Canada which when combined with an overestimation of warmth earlier in the record results in an observable reduction in warming rates for minimum and mean temperature anomalies since 1950. The temporal evolution and spatial pattern of the observed biases suggest that the BETP-automated adjustments to station data in northern Canada miss some inhomogeneities in the raw station data. These results highlight the need for enhanced data recovery and homogenization efforts in data-sparse high latitude regions and emphasize the importance of national-scale climate data sets for evaluating global gridded products. We also recommend caution when using the BETP minimum and maximum monthly air temperature products for long-term trend analyses.