The Effect of Elevation Bias in Interpolated Air Temperature Data Sets on Surface Warming in China During 1951–2015

Although gridded air temperature data sets share much of the same observations, different rates of warming can be detected due to different approaches employed for considering elevation signatures in the interpolation processes. Here we examine the influence of varying spatiotemporal distribution of sites on surface warming in the long‐term trend and over the recent warming hiatus period in China during 1951–2015. A suspicious cooling trend in raw interpolated air temperature time series is found in the 1950s, and 91% of which can be explained by the artificial elevation changes introduced by the interpolation process. We define the regression slope relating temperature difference and elevation difference as the bulk lapse rate of −5.6°C/km, which tends to be higher (−8.7°C/km) in dry regions but lower (−2.4°C/km) in wet regions. Compared to independent experimental observations, we find that the estimated monthly bulk lapse rates work well to capture the elevation bias. Significant improvement can be achieved in adjusting the interpolated original temperature time series using the bulk lapse rate. The results highlight that the developed bulk lapse rate is useful to account for the elevation signature in the interpolation of site‐based surface air temperature to gridded data sets and is necessary for avoiding elevation bias in climate change studies.