How can we accurately trace China’s climate change over the past more than 60 years? Instrument upgrades, station changes, and gradual urbanization may leave “non-climatic” signals in conventional meteorological observations, seriously hindering our understanding of climate change in China.
Recently, a research team led by Professor Kaicun Wang from the College of Urban and Environmental Sciences at Peking University published a paper titled “A Homogenized Gridded Climate Dataset for China” in Science China Earth Sciences . At the same time, the team publicly released a daily homogenized gridded climate dataset for China (HCD01) covering 1961-2022 through the National Tibetan Plateau Data Center (https://doi.org/10.11888/Atmos.tpdc.302712). The paper systematically summarizes the evolution of China’s national surface meteorological station network and observing instruments. Building on homogenized raw observations from about 2,400 national meteorological stations and reanalysis data, the team developed a gridded climate dataset at 0.1° spatial resolution, providing high-quality data for studies of regional climate change and its impacts in China.
In-situ observations from surface meteorological stations are widely used for climate-change detection, attribution, and impact studies, as well as for evaluating satellite remote-sensing retrievals and model simulations. However, as observing instruments are continuously upgraded and replaced, different instruments may introduce different biases, causing artificial changes in observations, namely inhomogeneity in climate data.
In general, observational data are expected to represent large-scale climate-change signals of 100 km or more. Yet changes in the local environment around stations, such as urbanization, may be superimposed on large-scale climate signals. Therefore, to accurately detect and attribute climate change, it is necessary to remove, as much as possible, the effects of non-climatic factors and local environmental changes.
Among these effects, abrupt changes caused by station relocations and instrument replacements are usually large and relatively easy to detect and correct through homogenization. By contrast, changes in the observing environment around stations are often gradual, and each individual change has only a small effect on observations, making them difficult to identify and correct. Over time, however, such effects can accumulate and produce systematic biases, significantly affecting estimates of climate-change trends and even the reliability of climate-change detection and attribution.
Traditional homogenization methods generally rely on comparisons with neighboring stations. Specifically, reference stations with similar elevation and reliable data quality are selected around a target station, and possible inhomogeneities are identified by comparing their time series. However, when many stations in a region are simultaneously affected by instrument sensitivity drift, instrument replacement, urbanization, and other factors, the reference stations themselves may also contain inhomogeneities. As a result, traditional methods can miss such signals and have limited ability to detect both gradual and abrupt inhomogeneities under these conditions.
To address this problem, after more than a decade of research, Wang’s team proposed a homogenization method based on same-station comparison. This method overcomes key difficulties in detecting and correcting gradual inhomogeneities and enables the detection and correction of both gradual and abrupt inhomogeneities in China’s land-surface climate observations, including surface solar radiation, wind speed, relative humidity, air temperature, ground temperature, and precipitation. It therefore provides important technical support for building high-quality climate datasets for China.
The results show that, because strict calibration instruments and procedures were lacking before 1990, gradual inhomogeneity caused by instrument sensitivity drift affected China’s surface solar radiation observations. As a result, the decreasing trend in surface solar radiation during 1960-1990 was seriously overestimated, while instrument updates during 1990-1993 caused a sudden increase in the observations.
In the 21st century, China’s meteorological observations began shifting from manual to automatic observation, with the greatest impact on relative humidity. Under low near-surface wind-speed conditions, the dry- and wet-bulb thermometers used in manual observations tended to overestimate relative humidity, whereas the capacitive sensors used in automatic observations did not have this problem. This difference produced a false decreasing trend in relative humidity observations in the 21st century.
In addition, urbanization around meteorological stations amplified the warming trend of daily minimum temperature, while increased surface roughness caused by urbanization reduced near-surface wind speed. At the same time, relocations of meteorological stations from urban to rural areas caused abrupt increases in wind speed. Although the decline in station-observed near-surface wind speed does not represent large-scale change, it reflects real changes in near-surface wind speed at observing sites and affects station precipitation observations.
However, homogenized station observations alone are still insufficient for climate-change research. Because observing stations are sparse and unevenly distributed, and because observation periods differ among stations, it is difficult to conduct direct spatiotemporal analysis at the national scale. Therefore, based on homogenized station observations and ERA5-Land reanalysis, the team constructed the Homogenized Gridded Climate Dataset for China (HCD01), providing a high-quality data foundation with complete spatial coverage for regional climate-change research. The results show that HCD01 clearly improves the representation of long-term trends compared with reanalysis data and raw observations.
The study was led by Professor Kaicun Wang from the College of Urban and Environmental Sciences at Peking University, who served as the first author and corresponding author. The research team included current graduate students Hongze Cai, Yun Li, Hanmeng Xia, and Changjian Yin from Wang’s group, as well as former graduate students including Professor Chunlue Zhou from Sun Yat-sen University, Associate Professor Yanyi He from Sun Yat-sen University, Young Researcher Zhengtai Zhang from Lanzhou University, and Dr. Runze Zhao from the National Satellite Meteorological Center of the China Meteorological Administration. The study was supported by the National Key Research and Development Program of China (2022YFF0801302) and the Science and Technology Program of Guizhou Province (Qian Ke He Ren Cai XKBF[2025]012).
See the article
Wang K, Cai H, He Y, Li Y, Xia H, Yin C, Yu H, Zhao R, Zhang Z, Zhou C. (2026). A homogenized gridded climate dataset for China. Science China Earth Sciences, 69(8), 2856-2878. https://doi.org/10.1007/s11430-025-1889-0
Science China Earth Sciences