Asymmetric Changes in Heating and Cooling Degree-Day Climates Across China in the Context of Faster Minimum-Temperature Warming
Ming Shang, Yanan Yao, Qian Li, Yihang Xing, Jingying Xu, Lei BaiDegree days are outdoor-temperature indicators relevant to heating and cooling. We compare land-area, fixed 2023 population raster, and fixed 2015 GDP raster summaries; no measures observed building energy use. Using daily CN05.1 temperature for 15,247 Chinese land cells during 1961–2022, we quantify Tmin/Tmax trends, HDD/CDD changes, sensitivity to degree-day bases and periods, and spatially dependent uncertainty. Area-weighted Tmin warmed at 0.0398 °C yr−1 (95% bootstrap CI: 0.0357–0.0438), faster than Tmax at 0.0245 °C yr−1 (0.0189–0.0300); the Tmin-minus-Tmax trend was positive over 79.5% of weighted area after FDR control. Between 1961–1990 and 1991–2020, HDD18 declined from 4714.6 to 4432.8 °C d, whereas CDD26 increased from 29.1 to 37.3 °C d. CDD26 growth was spatially heterogeneous and FDR-significant over 38.5% of weighted area, while HDD18 decline was widespread. The cooling-share metric R increased from 0.0197 to 0.0264, but R > 0.10 occupied only 5.02% and 6.96% of baseline and recent weighted area, respectively. The signs of national HDD retreat and CDD growth persisted across nine base-temperature combinations and a 1993–2022 recent-window sensitivity. A four-group k-medoids synthesis was stable under 1000 spatial-block bootstrap refits (median ARI 0.874). The unadjusted Tmin-minus-Tmax trend was associated with ΔHDD18 (Pearson’s r = −0.640) but was near zero for ΔCDD26 (r = −0.037); after controlling for the TM trend, both partial associations were near zero. Tmin-dominant warming is therefore interpreted as climatic context rather than an independent explanation for either index. The comparison with GB 50176-2016 identifies correspondence and mismatch with its temperature-based design-climate concepts. China thus shows a robust heating-side retreat alongside geographically differentiated cooling growth. Under fixed 2023 population and 2015 GDP raster distributions, HDD18 changed by −206.9/−217.6 °C d and CDD26 by +20.9/+26.1 °C d, showing that magnitude depends on the spatial denominator.