| 英文摘要: | To the Editor —
Ji et al.1 present a methodology to analyse global (excluding Antarctica) spatiotemporal patterns of temperature change, using mean monthly temperatures obtained from the updated Climate Research Unit (CRU) high-resolution gridded climate database2, 3. Their analysis fails to take into account several key characteristics of the CRU database, seriously compromising the conclusions regarding the spatiotemporal patterns of global warming during the twentieth century.
Climatic data comes from thousands of stations scattered non-randomly across Earth, with much higher densities at mid-latitudes than in the tropics or the Arctic, creating spatial bias. A distance-weighted interpolation from available meteorological stations was implemented to fill spatial gaps in the CRU database2, 3, 4. Land pixels outside a search radius of 1,200 km from the closest meteorological station were given the corresponding CRU 0.5° 1961–1990 mean monthly climatology4, 5 (Supplementary Fig. 1; other search radii apply to other variables in the CRU database).
In terms of temporal bias, the CRU dataset logically contains many fewer observations in the early part of its record. This is particularly prevalent in remote tropical and Arctic regions, where temperature records abound with long-term climatological averages. Consequently, the temporal autocorrelation of such time series is artificially high, and the climatic variability they portray for the early decades of the record is meaningless (Fig. 1). |