Quantitative response relationships between annual precipitation in China from 1951 to 2018 and its influencing factors

Affected by anomalous climate changes, the problem of an uneven spatial distribution of precipitation on a global scale has been aggravated (Chen et al. 2014; Sun et al. 2016; Diffenbaugh et al. 2017; Vousdoukas et al. 2017). Research on the causes of climate change deems the degree of change in precipitation to be caused by influencing factors (Vaishnavi et al. 2021). Precipitation is a key meteorological factor, and the qualitative and quantitative response relationships between precipitation and its influencing factors are a key issue in climate change research.

The main tools for measuring and explaining global climate change are climate system models and their related mathematical statistical methods (IPCC 2007). The tools’ advantages are that each possible factor can be examined one by one, or the impact of multiple factors can be considered at the same time, to quantitatively analyse their respective relative contributions. In the past, most model experiments classified the influencing factors into two categories: natural forcing (solar activities, etc.) and human activities (greenhouse gases, etc.), and some model experiments set different human emission scenarios, such as those in the International Panel on Climate Change's second report on emission scenarios (Ding & Ren 2008). The relationship between precipitation and its influencing factors has been analysed using various methods, including fuzzy C-means clustering algorithm (Huang et al. 2017), empirical orthogonal function (EOF) (Liu et al. 2019), and NCEP/NCAR reanalysis (Yue 2020) and multivariate regression (Chen et al. 2014), and these relationships have been studied on the global (Lau et al. 2013) or the regional scale (Northern Hemisphere (Zhang et al. 2013), Europe (Liu & He 2020), North America (Whan & Zwiers 2017), and Canada (Nalley et al. 2019)). When carbon dioxide (CO₂) increases by 1% each year, global heavy precipitation events increase, moderately heavy precipitation events decrease, and the period with little or no precipitation increases. In the Northern Hemisphere, the influence of anthropogenic forcing was detected in the observation of extreme precipitation from 1951 to 2005, and the influence of human activities on precipitation grew by an average of 3.3% in this period (Zhang et al. 2013). The EOF1 of summer and winter precipitation in Europe is related to the North Atlantic oscillation (NAO), the EOF2 of summer precipitation is related to the 500-hPa height model, and the EOF2 of winter precipitation is related to the East Atlantic teleconnection patterns (Zveryaev 2006). The NAO has the greatest impact on extreme precipitation in eastern North America, and the probability of extreme precipitation events is higher in the north than in the south (Whan & Zwiers 2017). The low-precipitation and high-precipitation periods in the Canadian Basin are correlated with the positive and negative phases (El Niño–southern oscillation (ENSO) and NAO), respectively, and the neutral and negative phases have similar trends (Nalley et al. 2019).

China is located in eastern Asia on the western coast of the Pacific Ocean. The complex and diverse climate and the geographical environment lead to significant temporal and spatial differences in precipitation across China. The precipitation and its influencing factors in eastern China (Huang et al. 2011), North China (Hao et al. 2011), the Loess Plateau and the Jianghuai River Basin (Zhou 2018) are studied. For example, the annual precipitation in eastern China is closely correlated with the summer water vapour transport flux over East Asia (Huang et al. 2011). The increase in intensity of strong precipitation events has been attributed to anthropogenic forcing factors, among which the increase in greenhouse gases plays a dominant role (Ma et al. 2017). In the northern region, the annual precipitation gradually increases from the northwest to the southeast and northeast, and the summer precipitation is more affected by the sea-surface temperature (SST) of the Pacific Ocean and the SST of the Indian Ocean (Zhang 2007). In North China, the seasonal precipitation is mainly controlled by the evolution of the ENSO and the Indian Ocean dipole (IOD), and the SST anomaly in the Indian Ocean has a more pronounced effect on the seasonal precipitation than the ENSO evolution (Hao et al. 2011). The precipitation in Southwest China is low during a weak Arctic oscillation (AO) and high during a strong Northern Hemisphere annular mode and in El Niño, the precipitation in the central part of Southwest China is high in La Niña years, and the precipitation in the eastern and western parts of Southwest China is low (Jiang & Li 2010). The IOD, NAO, and ENSO, respectively, have large impacts on the annual precipitation, wet season precipitation, and dry season precipitation in the Pearl River Basin (Huang et al. 2017).

Previous studies on the quantitative response relationships between precipitation and its influencing factors have certain limitations. Some of them only evaluated the relationship between precipitation and its influencing factors by correlation analysis and could not quantify the strength of the response of precipitation to its influencing factors. Some applied models to analyse the quantitative response relationship between precipitation and a single influencing factor in China. Few studies have analysed the quantitative response relationships between precipitation and multiple influencing factors or investigated the ability of CMIP6 precipitation data to simulate this relationship in China. We qualitatively and quantitatively studied the response relationships between precipitation and its influencing factors by analysing the monthly precipitation data of 619 meteorological stations in China from 1951 to 2018, the precipitation data of multiple CMIP6 models, and the data of 10 influencing factors that are closely related to precipitation, including the Pacific decadal oscillation (PDO), Atlantic multidecadal oscillation (AMO), multivariate El Niño southern oscillation index (MEI), AO, and solar radiation (SR). We applied multiple methods, including EOF decomposition, detrending, and redundancy analysis. The results provide a reference for understanding of the response characteristics of precipitation to climate change on a global scale and better equip us to cope with climate change.

Located in the mid-latitudes of the Northern Hemisphere (Figure 1), China has a vast territory, with a land area of approximately 9.6 million square kilometres. Mountains, plateaus, and hills account for approximately 67% of its land area, and basins and plains account for approximately 33% of its land area. The land is higher in the west and lower in the east, with various types of landforms and climates, including monsoon climates, temperate continental climates, and plateau mountain climates. Therefore, the ecological environment systems in China are complex and diverse.

In this study, the monthly precipitation data of 683 stations in China from 1951 to 2018 were obtained from the China Meteorological Data Network (http://data.cma.cn/). The data source is accurate, the coverage is wide, the credibility is high, and can fully represent the regional precipitation situation. Considering the reliability of the meteorological data, 64 stations were excluded after the precipitation data at each station were preprocessed before the study. Finally, 619 meteorological stations (Figure 1) were selected to represent the precipitation in the study area. To analyse the response relationships between precipitation and its influencing factors in China, climatic factors that are closely related to precipitation were selected as the driving factors of precipitation changes. The time-series data of global CO₂ radiative forcing (CO₂) and annual greenhouse gas radiative forcing (Annual Greenhouse Gas radiative forcing (AGG)) from 1979 to 2018 were obtained from the US National Oceanic and Atmospheric Administration (NOAA) Earth System Research Laboratory (Global Monitoring Division). The time-series data of PDO, AMO, MEI, and AO from 1951 to 2018 were all obtained from the US NOAA Earth System Research Laboratory (Physical Sciences Division). The monthly data of atmospheric pressure (AP), wind speed (WS), and relative humidity (RH) from 1951 to 2018 were obtained from the 619 meteorological stations where the precipitation data were obtained. The monthly time-series SR data of 59 meteorological stations, which covered the entire study area, from 1959 to 2018 were also obtained from the China Meteorological Data Network. Eight CMIP6 models were selected, namely BCC-CSM2-MR, CanESM5, CMCC-CM2, EC-Earth3, FGOALS-f3-L, MIROC6, MPI-ESM1-2-HR, and MPI-ESM1-2-LR.

The following two types of experiments were conducted using these models. The first type of experiment was the historical climate simulation experiment, which integrated the data from the middle and late-19th centuries to December 2005. The second type of experiment was the 21st-century climate prediction experiment, which integrated the climate forecast data under the future shared socioeconomic path (SSP2-4.5) scenario from January 2006 to December 2100. The historical and forecast data of monthly precipitation output by the two types of experiments were concatenated, forming the time series from 1951 to 2018.

• (1)

  Correlation analysis and regression analysis were done for interpolation and extension of the missing data.

• (2)

  Correlation analysis was done to analyse the response relationships between precipitation and changes in its influencing factors.

• (3)
  The precipitation series attributed to natural variability was obtained by removing the contribution of human activities to the changes in precipitation using the CMIP6 data through detrending (Dai et al. 2015). The steps are as follows: First, rank the spatio-temporal simulation effects of the eight modes mentioned above, calculate the weight of each mode, and make a weighted set of the model results. Assuming that the precipitation T(n, i) of site i in the NTH year can be expressed as:

  

Type is to use regression observed precipitation regression coefficient. The precipitation anomaly field (relative to 1951–1990) is used in the above formulas to eliminate the model's deviation from the baseline period simulation and reduce the error as much as possible.

• (4)

  The precipitation field was decomposed using EOF decomposition (Ullah et al. 2021). The combinations of the resulting space vectors and the time coefficient can basically reflect the information about the original meteorological factor field. The spatial modes obtained from the analysis need to be subjected to the North test.

• (5)

  The time coefficient of EOF decomposition was processed using the z-score standardization method.

• (6)

  Ensemble empirical mode decomposition was applied to separate interdecadal signals from the precipitation series to more stably decompose the nonlinear and nonstationary series so that the true climate change signals could be extracted (Wu & Huang 2009).

• (7)

  To accurately estimate the relative contributions of various external forcing factors to climate change, the CMIP6 data were used to remove the human influence from the observation data. It is assumed that after the removal of human influence, the precipitation observation field will respond to various natural variabilities independently and linearly. Stepwise regression was used to make sure the sum of the contributions of different influencing factors to precipitation was smaller than the change in the observed results.

• (8)

  Precipitation was set as the response variable, and the selected influencing factors were the explanatory variables. Through multivariate statistical analysis, principal component analysis was used to perform dimensionality reduction analysis on the influencing factors, and partial redundancy analysis was used to perform variance separation of the influencing factors to quantify the contribution rates of different influencing factors to precipitation.

In this paper, the time series of precipitation from 1951 to 2018 at each station was decomposed by EOF. The first three EOF modes, namely EOF1, EOF2, and EOF3, can largely explain the variability of annual precipitation (passes the North test). The cumulative variance contribution rate of EOF1, EOF2, and EOF3 to annual precipitation reached 39.9%. The other modes of EOF each contributed less than 5% to the annual precipitation, and their relationships with atmospheric circulation are still not clear. Therefore, this paper only analyses the spatial patterns and the time series of the principal components of the first three EOF modes of precipitation and the correlations between the first three EOF modes and the influencing factors of precipitation (Figures 3–5 and Table 1).

EOF1 accounts for 20.1% of the total variance in annual precipitation. Figures 2 and 3 show that the eigenvectors of EOF1 basically reveal the whole spatial distribution of annual precipitation in China. Most regions north of the Yangtze River Basin have negative EOF1 values, while most regions south of the Yangtze River Basin have positive EOF1 values. The centre of positive EOF1 values appears in the southern part of East China, and the overall distributions of annual precipitation in the northwest and southeast have opposite polarity. In addition, the precipitation north of the Yangtze River Basin changes much less than that south of the Yangtze River Basin, with the middle and upper reaches of the Yangtze River and Northeast China as the transition zone. The time coefficient represents the temporal variation characteristics of the spatial distribution mode of the corresponding eigenvector. The sign of the time coefficient determines the direction of the mode. Specifically, a positive sign indicates the same direction as the mode, while a negative sign indicates the opposite direction to the mode. The time series of EOF1 basically reveals the interannual variability of precipitation in China, and its negative slope indicates that the annual precipitation increased in the regions north of the Yangtze River Basin and decreased in the regions south of the Yangtze River Basin over the 68 years from 1951 to 2018. The annual precipitation in the regions north of the Yangtze River Basin showed a decreasing trend before the 1960s, an increasing trend from then to 1997, a rapid decline for the next 6 years, and then an increasing trend through 2018. The annual precipitation in the regions south of the Yangtze River Basin showed an opposite pattern to that in the regions north of the Yangtze River Basin. According to the analysis of the correlations between the first three modes of EOF and the influencing factors (Table 1), the correlations of PC1 (the first principal component) with MEI, RH, and PDO pass the 95% significance test, indicating that the EOF1 of annual precipitation in China was significantly correlated with RH, MEI, and PDO.

EOF2 accounts for 13.5% of the total variance in annual precipitation. It is positive in the northern part of Central South China and the northern part of Northeast China and negative in other areas. The centre of negative EOF2 values is on the edge of southern China, while the centre of positive EOF2 values is on the middle–lower Yangtze plains. The degree of change in precipitation is much higher in the south than in the north (Figure 4). The positive slope of the time series of EOF2 indicates that from 1951 to 2018, the precipitation increased in the northern part of Central South China and the northern part of East China and decreased the other areas. The correlations of PC2 with SR and AMO pass the 95% significance test, indicating that EOF2 was mainly affected by SR and AMO.

EOF3 exhibits a single precipitation pattern, accounting for 6.3% of the total variance in annual precipitation (Figure 5). The centre of negative EOF3 values is on the North China Plain. The degree of change in precipitation is higher in the east than in the west. The negative slope of the time series of EOF3 indicates that the precipitation in China increased before 2002 and then began to decrease. EOF3 is greatly affected by WS and MEI.

As shown in Table 1, the annual precipitation in China is significantly correlated with such influencing factors as the PDO and AMO. Although the correlations of annual precipitation with AO and CO₂ do not pass the significance test, AO and CO₂ still affect the precipitation in some regions of China (e.g., AO has a large impact on precipitation in northern China (Thompson et al. 2003)). Therefore, the spatial distribution of the correlations between precipitation and its influencing factors was analysed in this study through correlation analysis (Figures 6–11). The overall response of precipitation to the PDO and AMO was strong (Figure 6), followed by the response of precipitation to the MEI. The correlation between precipitation and the PDO is significantly positive in Northwest and East China, significantly negative in the coastal area of the Bohai Sea, and weak on the Loess Plateau. The correlation between precipitation and the AMO is positive at 90% of the stations and negative only on the Yunnan–Guizhou Plateau. The correlation between precipitation and the MEI is positive at half of the stations and negative at the other half of the stations, significantly positive in Northeast and East China, significantly negative in the Yellow River Basin and the regions south of the Tibetan Plateau, and not significant in the other areas. As shown in Figure 7, the interannual oscillation between annual precipitation and the PDO and the interannual oscillation between annual precipitation and MEI is overall in phase, while the interannual oscillation between annual precipitation and the AMO was out of phase before 1996 and in phase after 1996.

The coefficients of correlation of precipitation with WS, RH, and AP follow the descending order of WS > RH > AP (Figure 8). The correlation coefficient between precipitation and WS exhibits a distribution pattern of gradually decreasing from north to south, while the correlation coefficient between precipitation and RH has an opposite distribution pattern. The correlation between precipitation and WS is significantly positive in Northwest China, North China, and the northern part of East China and significantly negative in the marginal region of Southwest China. The correlation between precipitation and RH is significantly positive in southern China (passes the 95% significance test at half of the stations) and negative in the regions north of the Tianshan Mountains. The correlation between precipitation and AP is weak and does not pass the significance test at most stations, but it is significantly positive in the northern part of the Great Khingan Range and at the southern marginal region of Central South China and significantly negative in the north of the Qilian Mountains in China. As shown in Figure 9, the interannual oscillation between annual precipitation and WS is overall out of phase, the interannual oscillation between annual precipitation and RH is in phase, and the interannual oscillation between annual precipitation and AP was out of phase before 1998 and in phase after 1998.

The spatial distribution of correlation between precipitation and SR in China is > AGG > AO > CO₂, which showed a significant negative correlation with SR, and a positive correlation with AGG, but a poor correlation with CO₂ (Figure 10). The correlation between precipitation and SR is significantly negative in Northwest China, the central part of North China, and the regions north of the middle and lower reaches of the Yangtze River (the negative correlations have a concentrated distribution), significantly positive only in Northeast China, and nonsignificantly positive at the marginal region of Southwest China. The correlation between precipitation and AO is significantly negative in the high-altitude Tibetan Plateau, the Tarim Basin, and the Jungar Basin in Northwest China as well as in Northeast China, and is significantly positive in the regions south of the Yellow River Basin and the Yangtze River Basin. It is overall positive at half of the stations and negative at half. The negative correlation between precipitation and AGG gradually weakens from northwest to southeast until the dividing line between the second and third steps of topography in China, then strengthens, and became significant in Northwest China, the north part of southwest China, and the coastal areas of East China. There is no positive correlation between precipitation and CO₂ in the entire study area; the correlation between precipitation and CO₂ is nonsignificantly positive across the study area. As shown in Figure 11, the interannual oscillations between annual precipitation and SR, AO, AGG, and CO₂ were in phase.

Overall, precipitation had strong response relationships with the AMO and MEI across China (Figure 12). PDO and AGG mainly affected precipitation in western China. SR and AO mainly affected precipitation in areas north of 25 °N. AP affected precipitation in Northeast China, and WS affected precipitation in North China and western China. The response relationship between precipitation and RH in the region south of 25 °N was strong, and the response relationship between precipitation in China and CO₂ was weak.

The model data were extracted, corresponding to 619 meteorological stations in China, and the correlation between the observation data of all stations and the model prediction was analysed. As shown in Figure 13, although the simulated precipitation series of different models deviates from the observed precipitation series in China, their overall fluctuation characteristics are consistent with each other, and the correlation between the two passes the significance test (0.702). Therefore, the models can well simulate the trend of precipitation in China. The contribution of human activities to precipitation at these 619 stations in China was removed using the model series to obtain the change in precipitation attributed to natural variability.

To quantify the contributions of the selected influencing factors to precipitation, the eigenmode function on the time scale was obtained by ensemble empirical mode decomposition of the precipitation observation field series and the influencing factor time series in China. Then, linear model-based redundancy analysis was applied to the eigenmode function to quantitatively analyse the contribution rates of the influencing factors to precipitation. The precipitation at 619 meteorological stations in China was calculated using the above method. The detailed calculation process at one station is used as an example. The selected influencing factors are classified into two categories. The first category is composed of the atmospheric circulation factors, including PDO, AMO, MEI, and AO. The second category is composed of the regional influencing factors, including RH, SR, WS, and AP. As shown in Figure 14, the contribution rate of atmospheric circulation factors to precipitation is 39.5%, and the contribution rate of regional influencing factors to precipitation is 21.7%.

According to the above calculations, the spatial distribution of the contribution rate of each influencing factor to precipitation in China is obtained (Figure 15). The overall contribution rates of the atmospheric circulation factors AMO, PDO, MEI, and AO to precipitation follow the descending order of AMO > PDO > MEI > AO, and the overall contribution rates of the regional influencing factors SR, WS, RH, and AP follow the descending order of SR > WS > RH > AP. The high contribution rates of the PDO to precipitation are concentrated in the northern Tarim Basin (where its contribution rate reaches 14.6%), and the contribution rate of the PDO to precipitation gradually decreases towards the east, drops to approximately 0.8% on the Loess Plateau, then increases gradually further down in the east, reaching 12% in the area near the Yellow Sea. The contribution rate of the AMO to precipitation is large in western China, Northeast China, and the central part of East China and smallest in the Sichuan Basin and the North China Plain. The contribution rate of the MEI to precipitation is smallest at the zone including the sources of the Yangtze, Yellow, and Lancang Rivers and the regions south of the Sichuan Basin, becomes larger in regions east or west of this zone, and reaches its maximum (13.1%) at the coastal region of the Bohai Sea. The AO is mainly active in the Arctic. As shown in Figure 15, the AO contributes to the precipitation in Northeast China and the Qaidam Basin, and its contribution rate to precipitation lessens from north to south. The overall contribution rate of the AO to the precipitation in China is low, accounting for less than 5% of the natural variability. The high contribution rates of RH to precipitation are concentrated in the regions south of the Yangtze River, with contribution rates of up to 15.0%, and the contribution rates of RH to precipitation gradually decrease towards the north, drop to 1.5–3.0% in western China, and become essentially zero in Northeast and North China. The contribution rate of SR to precipitation is large in western China, peaking in the Qaidam Basin, and small in the southern part of East China and the southern part of Northeast China. The contribution rate of WS to precipitation is large in the Tarim Basin and the central part of North China but is essentially zero in Northeast China and the regions south of the middle and lower reaches of the Yangtze River. The overall contribution rate of AP to precipitation in China is less than 5%, and the contribution rate of AP to precipitation in the regions north of Qilian Mountains and the northern part of the Great Khingan Range is essentially zero.

Taken together, the causes of precipitation are very complex. Both natural factors and human activities affect precipitation. Moreover, the climate system itself has an inherent movement pattern. The influencing factors selected in this study differ greatly in the spatial distribution of their contribution rates. Some of the influencing factors make almost no contribution to precipitation in some areas of China. Except for RH, the influencing factors selected in this study have high contribution rates to precipitation in the regions west of 100 °E and in Northeast China. The AMO, MEI, RH, and SR all contribute to the precipitation in regions south of the Yangtze River, especially RH (rate ≈ 12%). The contribution rates of the AO and AP to precipitation in China are small, both approximately 5%.

Overall, the total contribution rate of the influencing factors to the annual precipitation in China is higher in the west and lower in the east (Figure 16). In the regions west of 100 °E, the total contribution rate of the influencing factors to annual precipitation increases from south to north and exceeds 38% north of the Tarim Basin. Among these influencing factors, the contribution rates of the AMO, SR, and WS each exceed 8%, and the contribution rate of the PDO exceeds 6.5%. In the regions east of 100 °E, the total contribution rate of the influencing factors to annual precipitation decreases from north and south to the central area. The influencing factors contribute least (approximately 10%) to the annual precipitation on the Inner Mongolia Plateau.

The AMO plays an important role in the precipitation variation in the mid-high latitude regions of the Northern Hemisphere (Gu & Adler 2015; Sang et al. 2020). The significant periodicity of annual precipitation is attributed to the variability of the PDO and AMO. Annual precipitation is highly correlated with the AMO in Northwest China, the Songhua-Liao River Basin, the Huang-Huai-Hai Plain, and most areas in the southern margin of the Yangtze River Basin, which is consistent with our results. The ENSO and PDO are important factors that affect precipitation in Asia (Gu & Adler 2015). When El Niño occurs, precipitation in the Xinjiang area decreases (Wen et al. 2020), cyclones appear over the Indian Ocean, and the South Asian high increases, resulting in more precipitation on the Tibetan Plateau (Chen et al. 2021). The high SST of the tropical eastern Pacific leads to the weakening of the Walker circulation and the Hadley circulation, and the enhanced Western Pacific subtropical high spreads southward, transporting water vapour to Northeast and North China (Chang et al. 2019), which is consistent with the response relationship between annual precipitation and MEI detected in this study. The PDO index represents the thermal difference over the North Pacific Ocean (Hurrell & Van Loon 1997), and its impact on precipitation is mainly attributed to the changes in the intensity of the East Asian summer monsoon and the westerly circulation (He & Jiang 2011; Wei et al. 2021). Since the 1970s, the interdecadal PDO has been in the positive phase, the Mascarene high and the Australian high have increased, the cross-equatorial flow has intensified, and the precipitation in the Xinjiang region has increased (Wang & Yang 2008). The westerly jet and the western Pacific subtropical high move northwards. Under the influence of westerly circulation, the abundant water vapour over the Bay of Bengal, the South China Sea and the west Pacific can be transported to the Hetao region and to Inner Mongolia (Li et al. 2011). When a positive PDO phase is combined with El Niño, an anticyclonic circulation is formed over the northwestern Pacific Ocean, which increases the southward transport of water vapour (Li et al. 2010; Cheng et al. 2016). Therefore, the PDO greatly affects precipitation in China. The different phases of the AO correspond to the redistribution of atmospheric mass over the Northern Hemisphere. This redistribution process can change the meridional gradient of the pressure field and cause the zonal westerly anomaly at middle latitudes, thereby affecting the climate in the Northern Hemisphere (Thompson et al. 2003; Kryjov 2021). The AO is mainly active in the Arctic, so its impact on precipitation in China also weakens from north to south, which is consistent with the results of this study.

The greenhouse gases CO₂ and AGG affect the zonal mean distribution of precipitation through two basic mechanisms. One is that when there is more greenhouse gas in the atmosphere, the potential significant decrease in the frequency of tropical cyclones and the increase in the downward longwave flux of the Earth's surface will raise the surface temperature and thus increase the atmospheric temperature and water vapour, and the increased water vapour further augments the downward longwave flux (Sugi & Yoshimura 2004), which in turn enhances the hydrological cycle. The other mechanism is that the change in the atmospheric circulation pattern will cause the storm track and the subtropical arid region to shift towards the polar regions, leading to the expansion of the tropical zone (Marvel & Bonfils 2013). The increase of greenhouse gas leads to enhanced east Asia summer monsoon, the summer precipitation increase, which is mainly due to the increase of greenhouse gases caused by atmospheric heating uneven enhancement on land, atmospheric stability and lower convection activities strengthen, convective clouds, cloud cover and the thickness increase, thus making the convective precipitation enhancement, and convection process is the main cause of a total increased precipitation (Peng 2021).

The incident SR affects the energy budget and climate system of the Earth. The atmospheric and surface characteristics of the Earth affect the SR and are affected by climate feedback. The scattering and absorption of SR reduce the surface solar flux, surface temperature, and surface moisture flux, thereby increasing the atmospheric stability and reducing precipitation (Liu et al. 2019). These factors all have important effects on precipitation in different regions. In addition, radiation changes caused by volcanic and solar activities (Hu & Fedorov 2017; Liu et al. 2019) superimposed on the natural evolution of the El Niño phenomenon (Schmidt et al. 2014), the Interdecadal Pacific oscillation (Dai et al. 2015), and the multidecadal variability of the NAO (Li et al. 2013), WS, and AP may all be important causes of changes in global precipitation.

In this paper, 10 influencing factors that are closely related to changes in precipitation were selected to qualitatively and quantitatively analyse the response relationships between precipitation across China and its influencing factors. Precipitation in China has overall strong response relationships with the AMO and MEI. The PDO and AGG significantly affect the precipitation in western China. The precipitation in the regions north of 25 °N has strong response relationships with SR and AO. AP affects the precipitation in Northeast China, and WS affected the precipitation in North China and western China. RH affects the precipitation in the regions south of 25 °N. Precipitation in China has the weakest response relationship with CO₂.

Overall, the contribution rates of the influencing factors to the total annual precipitation in China are higher in the west than in the east. In the regions west of 100 °E, the total contribution rate of the influencing factors to annual precipitation increases from south to north and exceeds 38% north of the Tarim Basin. Among these influencing factors, the AMO, SR, and WS have contribution rates greater than 8%, the PDO has a rate greater than 6.5%, and RH is the only influencing factor that makes no contribution to annual precipitation. In the regions east of 100 °E, the total contribution rate of the influencing factors to annual precipitation decreases from north and south to the central area, and the factors other than RH contribute to approximately 20% of the annual precipitation in Northeast China. The annual precipitation in the area south of the Yangtze River is mainly attributed to the AMO, MEI, RH, and total SR, among which RH has the highest contribution rate (approximately 12%). The influencing factors contribute least (approximately 10%) to the annual precipitation on the Inner Mongolia Plateau. The contribution rates of AO and AP to the precipitation are small across China, both below 5%.

This research was supported by the National Natural Science Foundation of China (Grant No. 51869016 and 52069019) and the Inner Mongolia Autonomous Region ‘Grassland talents’ project. We are grateful for their support.

Chang Lu carried out the Conceptualization, Data curation, Formal analysis and Writing – original draft. Xing Huang, Tingxi Liu, and Guohua Sun helped in the Investigation and Software. Long Ma provided guidance in the Funding acquisition, Investigation, Methodology and Writing – review & editing.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data cannot be made publicly available; readers should contact the corresponding author for details.