While the authors claim to perform extensive validation, the comparisons are primarily with datasets that utilize the same underlying station data. This limits the ability to truly assess the impact of their chosen methodology on the results and identify any potential biases. For instance, comparing CRU TS v4 with CRUTEM4.6 (which is not spatially interpolated and also uses CRU TS data) provides limited insights.

Changes in station coverage over time introduce potential biases in trend analysis, particularly at high spatial resolution. The reliance on climatological values to fill data gaps further exacerbates this issue. Despite providing metadata on station coverage, the dataset remains susceptible to artifacts in trend analysis. Users need to carefully mask regions or acknowledge potential inaccuracies stemming from these limitations.

The lack of homogeneity adjustments in CRU TS v4, while acknowledged by the authors, remains a significant concern. Though NMAs may homogenize individual station data before contributing it to CRU TS, combining these datasets could introduce new inhomogeneities. Users need to remain cautious when interpreting long-term changes and ideally consider other homogenized datasets for comparison or perform some homogenization.