1 Introduction

According to the statement on the state of global climate in 2019, it was noted that the trend of global warming is intensifying and could break the record again in the next five years. It also noted that climate change and extreme weather and climate events have a significant impact on human socio-economic development, health, population migration, food security and terrestrial and marine ecosystems, making climate change a global hot issue (Ding et al. 2016). The Sixth Synthesis Report of the Intergovernmental Panel on Climate Change (IPCC) states that the global surface temperature is higher by 0.99 (0.84–1.10) °C in the last 20 years and 1.09 (0.95–1.20) °C higher in the last 10 years than in 1850–1900. CMIP6 projections indicate a continuous increase in temperature until at least 2050, and possibly over 1.5 °C and 2 °C in the twenty-first century. The global average precipitation over land is increasing, and the world will have 0–13% more precipitation by the end of the twenty-first century than in 1995–2014 (IPCC 2021). Wind speed shows a downward trend ("Global Stationary") (Wu et al. 2020; Minola et al. 2016; Pryor and Ledo-lter 2010). Moreover, there is a global trend of decreasing sunshine hours from the 1950s to the 1980s and increasing after the 1980s (Dumitrescu et al. 2017; Sanchez-Lorenzo 2015; Manara et al. 2015; Rahimzadeh et al. 2012; Bartoszek et al. 2021). However, different regions of the world have different sensitivities to climate change, and it is generally accepted that the northern hemisphere is more sensitive than the southern hemisphere, higher latitudes are more sensitive than lower latitudes, and mountainous regions are more sensitive than plains (Mountain Research Initiative EDW Working Group 2015; Karmalkar et al. 2017). Even at the same latitude, there are certain variations in temperature and precipitation between regions due to various factors such as topography, altitude and atmospheric circulation. Temperature change is an important indicator of climate change, and other climate factors can also change as a result of temperature change (Xu et al. 2013).

Climate changes in China over the last hundred years are similar to other global and regional trends, with rising temperatures, small changes in precipitation amplitude and significant regional differences. According to the National Assessment Report on Climate Change, it is indicated that the mean annual surface temperature in China has increased by 0.78 ± 0.27 °C over the last 100 years (Ding et al. 2006). Meanwhile, the Blue Book on Climate Change in China states that the mean annual surface temperature in China increased by an average of 0.24 °C per decade in the period 1951–2017, with a higher warming rate than the global average over the same period. Annual mean surface temperatures in China will continue to rise over the twenty-first century, and the warming amplitude will also increase with increasing radiative forcing due to greenhouse gases (Zhang et al. 2013; CMA 2018). The warming range is higher in winter than in summer and higher in the north than in the south of China. Tan et al. (2016) and Zhai et al. (2017) reached a similar conclusion that in the context of global warming of 1.5 °C there is a consistent trend of increasing surface temperatures in China, with a larger increase in the north than in the south of China. From a geographical perspective, maximum temperatures show a warming trend west of 95° E and north of the Yellow River, and a cooling trend south of the Yellow River in the east (Zhai and Ren 1997). In contrast to the temperature trend, the precipitation trend in China during the last 100 years is not obvious (about + 0.1%/100 years), and the low frequency variation of precipitation in China in recent decades may be mainly caused by interdecadal variability, not entirely due to climate change trends (Wang et al. 2000). However, scholars point out that future East Asian summer winds will increase significantly, with an average range of precipitation increases of 5–14% across the country by the end of the twenty-first century (Ding et al. 2016), which will also further contribute to increased precipitation as surface temperature increases (Zhai et al. 2017). National precipitation is characterized by large interdecadal fluctuations (Wang et al. 2000; Ding et al. 2006) and large regional differences. The Blue Book on Climate Change in China notes that since the beginning of the twenty-first century, mean annual precipitation has fluctuated increased in North, South and Northwest China, while interannual precipitation fluctuations have increased in Northeast and East China (CMA 2018). The results of studies on trends in climate elements such as sunshine hours and wind speed show that both are dominated by decreasing trends to varying degrees, but there is a degree of variability in climate trends across regions, provinces and watersheds (Gao et al. 2017; Cao et al. 2015).

Shandong Province belongs to the temperate monsoon climate region, which is a typical climate vulnerability zone and one of the areas that are most sensitive to the effects of global warming (Zhang et al. 2007; Lv et al. 2013). Numerous climatologists have conducted extensive research on climate factors in Shandong Province and pointed out an overall trend of increasing temperatures and a declining trend of precipitation and wind speed in the province (Yan et al. 2015; Chen et al. 2016). Cui et al. (2003) also pointed out that the general trend of temperature change in Shandong Province over the last 40 years has been fluctuating and warming, and the mean temperature increase in the province during the 1990s and 1960s was higher in coastal areas than in neighbouring inland areas. The mean annual precipitation in the province showed a general trend of gradual decline from 1961 to 2000, with the general trend of decreasing precipitation being lower in the west than in the east, lower in the south than in the north, lower in the mountains than in plains and lower inland than on the coast, while the temperature tended to decrease from the northeast to the southwest (Cui et al. 2003). Ren et al. (2012) used linear analysis, the Mann–Kendall (M–K) method and the R/S method to conclude that the annual temperature in the southwestern part of Shandong Province continued to increase, insufficient precipitation and sunshine hours decreased during the period 1961–2010. Some scholars also used the empirical orthogonal function (EOF) method to analyse the spatial and temporal variability of monthly and seasonal precipitation in Shandong Province over 38 years and concluded that droughts and floods occur at the same time in spring in Shandong Province. Drought occurs in the spring in Shandong Province, and precipitation has obvious interannual variation, with an oscillation period of 4.8–6 years (Gao et al. 2005). Hu et al. (2020) used rotated empirical orthogonal function (REOF) and extreme-point symmetric mode decomposition (ESMD) methods to analyse the spatial and temporal coupling characteristics of winter precipitation fields in Shandong Province over the past 68 years. They have concluded that winter precipitation in Shandong Province as a whole presents three main distribution fields and three spatial modes, i.e. southeastern, northern and western. These studies confirm the significant spatial variability of climate change in Shandong Province under the influence of topography, altitude and geographical location, but previous studies on climate change were mainly conducted for the whole province, with few reports on mutation characteristics and spatial and temporal patterns and future projections of different climate elements in different climate regions. The complex topographic characteristics of Shandong Province are such that the characteristics of climate change in different climatic regions are not identical to the whole Shandong Province, even in the context of global climate change. Therefore, it is of great theoretical and practical significance to reveal the characteristics and projected trends of climate change in different regions of Shandong Province, for the development of local agriculture, forestry and animal husbandry, as well as to formulate strategies to cope with regional climate change.

EOF is one of the most important tools for studying climate change. Lorenz first introduced it to meteorological and climate research in the 1950s (Wu et al. 2005). EOF can decompose a variable field that changes with time into a space function part that does not change with time and a time function part that only depends on time change. It decomposes irregularly distributed sites in a limited area, and the decomposed space structure has a clear physical meaning and can extract the main signal features in the meteorological variable field and convert the variables from multidimensional to low-dimensional, which facilitates more accurate identification of meteorological variable field characteristics (Ding et al. 1992), and thus provides effective reflection of spatial distribution characteristics and changes in meteorological variable field. Climate region division studies using the EOF method have been widely used (Lin 2019; Karmalkar et al. 2017; Krauzig et al. 2020; Kosaka et al. 2016; Dong et al. 2014; Deng et al. 2014; Li et al. 2015; Shen et al. 2017; Lu et al. 2015; Huang et al. 2013), which has played an important role in promoting climate change research.

In this study, daily temperature, precipitation, sunshine hours and wind speed in Shandong Province from 1984 to 2019 were selected, and the EOF method and Mann–Kendall mutation detection were used to comprehensively investigate spatial and temporal distribution patterns of climate elements in different climatic regions of Shandong Province. At the same time, R/S analysis was applied to predict the trends of each climate element in different climatic regions of Shandong Province. This study aims to develop strategies for local built environment, World Heritage protection, living environment, agriculture, forestry, animal husbandry in order to cope with regional climate change.

2 Study area and data sources

2.1 Study area

Shandong Province, or Lu, is a coastal province in East China and is located between 114° 21′ E and 122° 43′ E and 34° 22′ N and 38° 24′ N (Fig. 1). The province is divided into the Shandong Peninsula and the interior. Tai Shan, the world's natural and cultural double heritage site, and the Great Wall of Qi, a world cultural heritage site 2500 years old, are located in the study area. Tai Shan and the Great Wall of Qi have many natural wonders related to climate, with multiple scientific, cultural and aesthetic values affected by geographical location and geological conditions (Tong 2011). The terrain which is complex and diverse and dominated by mountains, hills and plains accounts for 14.59%, 15.39% and 65.56% of the area, respectively. The terrain is high in the centre of the province and low in the surroundings. The climate is a warm temperate semi-humid monsoon climate, influenced by the continental and oceanic monsoons, with cold and dry winters and hot and humid summers. It has obvious climatic characteristics of changing seasons and significant regional differences. Therefore, natural disasters are common and diverse in Shandong Province, which is one of the areas with the high incidence and most severely affected by meteorological disasters such as droughts, seasonal torrential rains, floods, hailstorms, lightning and typhoons, as well as geological disasters such as earthquakes and mudslides (Chen 2019; Wang et al. 2015; Yue 2013). Regional climatic differences lead to different distribution of vegetation types. The plains are dominated by vegetation types that protect from wind and sand and are tolerant of salinity and can improve saline soils. The Yellow River plain is the main agricultural production zone in Shandong Province. Both mountainous and coastal hilly areas are dominated by vegetation types used for ecological construction, soil and water conservation, afforestation and greening. Among them, the mountainous and hilly area is an important forestry base in Shandong Province. Coastal hills, however, are the best preserved and most species-rich area of natural vegetation.

Fig. 1
figure 1

Distribution of meteorological stations in Shandong Province

Full size image

According to Gao and Liu (2005) and Hu et al. (2020), Shandong Province is divided into five major climatic regions (Fig. 1), namely: (1) the east Lu climate region, located along the Yellow Sea coast, which is rainy and windy climatic region with frequent typhoons; (2) the northwest Lu climate region, located in the northern impact plain of the Yellow River, is a high-temperature, semi-humid and semi-arid climate region; (3) the climate region of southwest Lu, which is typical for the Huang Huai Plain region, with significant precipitation seasonality, high temperatures, abundant sunshine hours and numerous droughts and floods; (4) the central Lu climate region, which includes the central mountains, hills and valleys, is a cool and humid climate region; and (5) the climate region of south-central Lu is located on the eastern edge of the North China Plain, and the terrain is dominated by low hills and mountains and represents a mild, low wind, humid and semi-humid climate region.

2.2 Data sources

Daily meteorological data on temperature, sunshine hours, precipitation and wind speed for 20 meteorological stations in Shandong Province were obtained from the Greenhouse Data Sharing Platform (http://data.sheshiyuanyi.com/) for the period 1984–2019. The metadata of each climate station are shown in Table 1. Twenty climate stations used by the institute all belong to the national monitoring stations, and their data acquisition equipment includes ZQZ-TF wind speed and direction detector, DZZ4 and ZQZ-A automatic weather stations, SL3-1 tipping bucket type rainfall sensor, DSG1 precipitation phenomenon meter, DSC1 weighing rain sensor and DFC1 and DFC2 Photoelectric Digital Sunlight Meters. All devices are maintained and data transmission is every minute (Liu et al. 2018a, b). According to geographical location, level of economic development, administrative division, geographical conditions and China Weather Network (https://www.tianqi.com/toutiao/wenhua/1874.html), Shandong Province is divided into five regions, i.e. east Lu Undulating Hilly Area, northwest Lu Plain Area, southwest Lu Plain Lake Depression Area, central Lu Mountainous Hilly Area and south-central Lu Central Mountainous Hilly Area (Liu et al. 2020a, b, c; Zhao et al. 2019; Guo 2015a, b). The geographical location of each meteorological station is shown in Fig. 1.

Table 1 Metadata table of meteorological stations in Shandong Province
Full size table

2.3 Analysis methods

In this study, the EOF method was used to analyse the variability characteristics of the spatial and temporal fields of climate elements (Wu et al. 2005), which consists of several uncorrelated typical modes that contain as much information as possible about the original fields by decomposing the original variables. The first few modes that pass the North test have some physical significance and can effectively reflect the changing characteristics of the spatial and temporal fields of meteorological elements. The temporal coefficients can effectively reflect changes in mode weights over time (North et al. 1982; Zhao et al. 2017). Applying the linear tendency estimation method and estimating a one-dimensional linear function Y = aX + b of the climate element of time X, where 10a is the tendency rate of the climate element to change, so the positive (negative) value reflects the increasing (decreasing) trend of the climate element with time. The larger the value the larger the trend (Yang et al. 2014). Both Mann–Kendall trend analysis and mutation test methods are applied. The Mann–Kendall trend analysis is used to determine the trend and magnitude of significance by calculating Z values, and when Z > 0 and Z < 0, they indicate increasing and decreasing trends, respectively. |Z| ≥ 1.28, 1.64, 2.32 means that the test of reliability is 90%, 95% and 99%, respectively. In the Mann–Kendall mutation test, both UFi (red line) and UBi (blue line) are statistical series, where UFi is obtained in a positive time series order, and UBi is obtained in the inverse time series order. Generally, when α = 0.05, the critical linear value u0.05 =  ± 1.96 and α = 0.01, u0.01 = 2.56. When the value of UFi or UBi is greater than 0 (less than 0), it indicates that the series have an upward (downward) trend, and exceeding the critical value indicates a significant trend of change. If the two curves of UFi and UBi intersect within a critical straight line, it is the mutation time point. If there is more than one mutation point, the climate mutation point detection is confirmed by the cumulative distance level method and sliding t-test (Wang et al. 2018; Wei 1999). Quantitative reflection of the sustainability of time series of climate elements was performed using R/S analysis, determining the trend of climate sustainability and whether it is significant in the future period in terms of the magnitude of Hurst. When Hurst = 0.5, it means completely random and not persistent; when 0.5 < Hurst ≤ 1, it indicates a long-term correlation with positive persistence, and the greater the Hurst value, the stronger the persistence; when 0 ≤ Hurst < 0.5, it indicates a long-term correlation and opposite persistence, and the smaller the Hurst value, the stronger the persistence. By establishing a statistical V curve, it is possible to judge and determine the existence of periodic cycles and the length of the sequence cycle, i.e. the length of time that past trends affect future trends. When the curve has the first obvious turning point, it means that the influence of the past trend on the future trend disappears. At this time, the corresponding time (n) is the average cycle length of climate element, i.e. the duration of change (Wang et al. 2018; Wei 1999; Zhao 2017). The above analysis was performed in MATLAB R2018b and Origin 2018. Significance tests (P < 0.05) were performed using SPSS 20.0. Spatially modelled distribution trends of climate elements were plotted using Surfer 15 and ArcGIS 10.2.

All station data are first checked and corrected for mechanical and recording errors, and then processed for high and low outliers, missing values, spatial outliers and month-by-month difference corrections (Zhang et al. 2014). The missing values are ≤ 30% for all stations and this is in line with the Chinese standard for meteorological data sequence elimination (Wu et al. 2013; Guo et al. 2009; Liu et al. 2018a, b). Therefore, in order to optimize incomplete data of the stations, linear regression and linear interpolation are used and the arithmetic average of the contemporary data from adjacent 2–3 stations within 150 kms is selected (Meng et al. 2012; Zhao et al. 2020; Bi et al. 2015; Yang et al. 2012;  Tang et al. 2009; Song et al. 2014), In terms of regional characteristics, taking a national monitoring station as a basic data station can not only maximize the use of research resources, but also guarantee the representativeness of the regional average trend estimation, and can effectively reflect regional local climate changes (Guo and Ding 2008; Liu et al. 2018a, b). Therefore, the features of each climate element are expressed by the average features of climate data from multiple meteorological stations in the same region (Chen et al. 2006; Liu et al. 2020a, b, c; Zhu et al. 2013).

3 Results

3.1 Characterization of climate factors

Seven climate elements were statistically analysed for five climate regions in Shandong Province (Fig. 2). As shown in Fig. 2, there are differences in median and mean values of each climate element between the five climate regions. On the median, it can be seen that the skewness of the distribution of W-Smean and W-Smax is significantly higher than that of other climatic elements. Median and mean values show that both Tmean and Tmax of the east Lu and central Lu climatic regions are lower than median and mean values of Shandong Province, while the opposite is true for other climatic regions, indicating that the east Lu and central Lu climatic regions are the low-temperature regions. Tmin in northwest Lu and central Lu are lower than median and mean values of Shandong Province, indicating that high-value Tmin areas are mainly distributed in the other three climate regions. Pmean was significantly higher in the south-central Lu and central Lu than in Shandong Province, indicating that the mountain region is a rainy region. S-Hmean is significantly lower in southwest Lu and south-central Lu than in other climate regions, which indicates that southwest Lu and south-central Lu areas have low sunshine hours. W-Smean and W-Smax, on the other hand, show opposite results to Tmean and Tmax, i.e. the east Lu and central Lu are windier regions. For each climate element, the five climate regions show interannual fluctuations, but there is relative consistency between the five climate regions. At the same time, the east Lu and central Lu climate regions (Taishan station and its nearby areas) are influenced by the ocean, altitude, topography and heat island effect, leading to their greater variability in temperature and wind speed compared to other climate regions (Bian et al. 2011).

Fig. 2
figure 2

Summary of meteorological elements statistics by each station

Full size image

3.2 Regional trends in climate elements and detection of mutations

3.2.1 Temperature

As given in Table 2, on an interannual scale there is a significant upward trend (|Z| > 2.32) of Tmean, Tmax and Tmin in Shandong Province, among which the  highest warming rate of Tmax reached 0.710 °C per decade. The contribution to temperature rise is shown as Tmin (0.547 °C·per decade, P < 0.01) > Tmean (0.446 °C·per decade, P < 0.01) > Tmax (0.391 °C·per decade, P < 0.05). Regionally, Tmean of each climate region is spatially distributed as "low in the middle, high around it, high in the west, low in the east" (Fig. 3a1), with the highest value (14.10 °C) in southwest Lu and the lowest value (11.74 °C) in central Lu. The spatial layout of Tmax in each climate region is "high in the west and low in the east" (Fig. 3b1), with the highest value (19.84 °C) in southwest Lu and the lowest value (16.62 °C) in east Lu. However, the spatial distribution of Tmin in each climate region is "high in the southeast and low in the northwest" (Fig. 3c1), with the highest value (9.60 °C) in east Lu and the lowest value (7.56 °C) in central Lu.

Table 2 Analysis of temporal changes in temperature and precipitation in Shandong Province
Full size table
Fig. 3
figure 3

Spatial trends of Tmean (a1), Tmax (b1) and Tmin (c1) and their propensity rates (a2), (b2) and (c2) in Shandong Province

Full size image

Figure 4a–c shows that on the provincial scale there is variability in the timing of abrupt temperature changes in Shandong Province, with Tmean, Tmax and Tmin underwent temperature increase mutations in the late 1990s, i.e. Tmax mutated one year earlier (1996) than Tmean and Tmin. On a provincial scale, the timing of mutations in Tmean, Tmax and Tmin occurs earlier on the coast than in the inland and earlier in the east than in the west of the province. Regionally, the timing of mutations in the three types of mean annual temperature also differs in each climate region, with the Tmean mutation time occurring earlier in the east Lu climatic region (1989), followed by northwest and south-central Lu (1996), and later in southwest Lu and central Lu (1997). The timing of mutation in Tmax was earlier in east Lu (1988), with Tmax (1988) > Tmean (1989) > Tmin (1993), followed by south-central Lu (1993), and later in other climatic regions (all in 1997). Among them, the mutation times of mean annual temperatures in the central Lu climate region are relatively late, i.e. Tmean (1997) = Tmax (1997) > Tmin (2005). It was shown that the timing of the Tmin mutation was earlier (1993) in the east Lu and south-central Lu climate regions, followed by northwest (1997), southwest (1999) and central Lu (2005) climate regions. Although there is variability in the mutation time in temperatures across climatic regions of Shandong Province, they all occurred between the 1990s and the beginning of the twenty-first century. Combined with Table 2, it can be seen that the change rule of mutation time of three types of temperatures in all climatic regions of Shandong Province (except south-central Lu) is basically in line with the magnitude of the temperature increase, which generally shows late mutation time and large amplitudes of temperature increase.

Fig. 4
figure 4

a Discriminant trend of M–K mutation change in Tmean. b Discriminant trend of M–K mutation change in Tmax. c Discriminant trend of M–K mutation change in Tmin

Full size image

3.2.2 Precipitation

As given in Table 2 and Fig. 5, Pmean in Shandong Province is 679.27 mm, with an increasing trend (|Z| < 1.28) and large interannual fluctuations. However, it does not pass the significance test (P > 0.05), and the overall spatial distribution of Pmean shows a gradual decrease from southeast to northwest. The south-central Lu has the most precipitation of 782.00 mm, while the northwest Lu has the least precipitation of 546.90 mm. The increase in precipitation amplitude shows the spatial variations features in the opposite direction to the spatial distribution of precipitation. Except for the decreasing trend of Pmean (Z < 0) in the south-central Lu climate region, other regions show an increasing trend of Pmean (Z < 1.28).

Fig. 5
figure 5

Spatial trends in Pmean (a1) and its propensity rate (a2) in Shandong Province

Full size image

As shown in Fig. 6, the mutation time in Pmean in Shandong Province occurred in 1989, with a decreasing trend before the mutation and a tortuous change trend after the mutation. Precipitation shows a non-significant 3–7 years and > 7 years periods of precipitation changes. Regionally, Pmean in each climatic region shows a fluctuating trend with no obvious patterns, but the mutation occurs in each climatic region, with the earliest occurrence in the east Lu climatic region (1985) and the latest in the central Lu climatic region (2002). Each climatic region shows tortuous changes in precipitation after the mutation, but all show an increasing trend. From this, it is clear that the timing of mutations of Pmean in each climatic region in Shandong Province shows a gradual postponement from east to west.

Fig. 6
figure 6

Discriminant trend of M–K mutation of Pmean (a) in Shandong Province

Full size image

3.2.3 Sunshine hours

As given in Table 3 and Fig. 7a, S-Hmean in Shandong Province were 2383.01 h, with an overall significant downward trend (|Z| > 2.32) and a decreasing amplitude of 71.769 h per decade (P < 0.01). From a regional perspective, S-Hmean in each climate region is the highest in east Lu with 2525.92 h and the lowest in south-central Lu with 2251.97 h. At the same time, in terms of regional contribution rate, S-Hmean shows a significant decreasing trend in each climatic regions (|Z| > 1.64), with the most significant trend for northwest Lu (− 96.749 h per decade, P < 0.01), followed by east Lu (− 84.956 h per decade, P < 0.01), south-central Lu (− 83.573 h per decade, P < 0.01) and southwest Lu (− 43.391 h per decade, P < 0.01). This indicates that northwest Lu, east Lu and south-central Lu are the main regions contributing to the decline of S-Hmean in Shandong Province, which may be related to climate, topography, altitude, cloudiness, urban and human activities (Zhai et al. 2017).

Table 3 Analysis of S-Hmean and W-Smean and W-Smax in Shandong Province and climatic regions
Full size table
Fig. 7
figure 7

Spatial trends in S-Hmean (a1), W-Smean (b1), W-Smax (c1) and their propensity rates (a2), (b2) and (c2) in Shandong Province

Full size image

As shown in Fig. 8a, the mutation time of S-Hmean in Shandong Province occurred in 1998, with monotonous decrease of 173.0 h in S-Hmean after the mutation. Regionally, mutations occurred in each climatic regions, with the latest occurrence in northwest Lu climate region (2006), followed by south-central Lu (2001) and east Lu (2001), and with the earliest occurrence in southwest Lu climate region (1990). Each climatic region shows a significant decreasing trend in S-Hmean after the mutation, and the variation of mutation time remains consistent with the magnitude of declining S-Hmean. From this, it is clear that S-Hmean in Shandong Province have late mutation time and large decreasing amplitudes.

Fig. 8
figure 8

Trend of M–K mutation discrimination of S-Hmean (a), W-Smean (b) and W-Smax (c) by region in Shandong Province

Full size image

3.2.4 Wind speed

Table 3 and Fig. 7b–c show that W-Smean and W-Smax in Shandong Province are 2.987 m s−1 and 6.127 m·s−1, respectively. Both show a significant downward trend (|Z| > 2.32), with the most significant contribution from W-Smax that decreased by 0.483 m s−1 per decade (P < 0.01). In contrast, the contribution rate of W-Smean is relatively weak, with a decreasing amplitude of − 0.264 m s−1·per decade (P < 0.01). Regionally, there is also a significant downward trend in both mean annual wind speeds (|Z| > 2.32) in all climatic regions. The most significant decrease is in the east Lu climate region (− 0.381 m s−1 per decade, P < 0.01), while lower decrease is in the southwest Lu climate region (− 0.190 m·s−1 per decade, P < 0.01). The decreasing trend in W-Smax is most pronounced in the east Lu climate region (− 0.693 m s−1 per decade, P < 0.01) and lower in the south-central Lu climate region (− 0.400 m s−1 per decade, P < 0.01). It is clear that east Lu is the main area contributing to the decrease in wind speed, which is in line with the largest amplitude of decrease in S-Hmean.

Figure 8b–c shows that the mutation time in W-Smax in Shandong Province occurred in 2001, which is earlier than the mutation time in W-Smean (2007), but both showed a monotonic decreasing trend after the mutation. The mutation time of each climate region is different, of which the central Lu region had a later mutation time in W-Smean (2008) and W-Smax (2007) relative to the other climate regions. The mutation time of the south-central Lu and east Lu is earlier than other climatic regions and the mutations in W-Smean and W-Smax occurred in 2004 and 1993, respectively. From this, it is clear that W-Smean and W-Smax in each climate region in Shandong Province show late mutation time in the central area and early mutation time in the surroundings, but W-Smean has early mutation time in the south and late in the north, while W-Smax has late mutation time in the south and early in the north.

3.3 EOF analysis of climate elements

To further investigate the spatial and temporal variations of climate elements in Shandong Province, the EOF decomposition was done for 20 stations in the province to obtain characteristic vector fields and time coefficient series of climate elements (Table 4). Table 4 shows the first four modes of the distance field for each climate element. The North test (North et al. 1982) shows that the first two modes can be selected for the analysis of all climate elements, except the annual mean maximum and annual mean minimum temperatures, for which the first three modes can be selected.

Table 4 First 4 EOF analysis variance contributions and cumulative variance contributions for each climate element / %
Full size table

3.3.1 Temperature

Table 4 shows that the cumulative contribution rates of the first four modes of Tmean, Tmax and Tmin are all over 96.10%. The variance contributions of the first mode eigenvector are all above 84.46%, as the main spatial distribution type. Among them, as shown in Fig. 9a1, the values of the first mode eigenvectors of Tmean are all positive, indicating consistency in Tmean change, i.e. warming or cooling during the same period. The high value centre is located in the Juxian area in the south-central Lu region and the low-value centre is located in the Jinan area in the central Lu region, which reflects the maximum temperature difference between the two regions. The second modal eigenvector had a variance contribution of 3.99% (Table 4), as the typical distribution type. As shown in Fig. 9a2, the spatial boundary is Rizhao–Hui Minxian–Lingxian, with areas of positive values in the north and negative values in the south, i.e. high temperatures in the north and low temperatures in the south or high temperatures in the south and low temperatures in the north. The centres of positive and negative values are located in the Fushan and Weifang regions in east Lu and central Lu, respectively, and changes in the eigenvector values reflect increasing Tmean from south to north. This shows that the mean annual temperature in Shandong Province is mainly distributed in two modes: "consistent in the whole province" and "north–south anti-phase".

Fig. 9
figure 9figure 9

EOF analysis of Tmean, Tmax, Tmin and their time coefficient trends (b) in Shandong Province

Full size image

The first modal eigenvector values for both Tmax (Fig. 9a3) and Tmin (Fig. 9a6) are positive, which indicates the consistency of the type of spatial distribution, i.e. warming or cooling during the same period. However, there is variability between the low- and high-value centres of Tmax located in Lingxian and Fushan areas in northwest Lu and east Lu, respectively, which shows that the degree of change in the climate region of northwest Lu is substantially lower than in the climate region of east Lu, while other climate regions are transitional regions. The high-value centre of Tmin is located in the Juxian area in south-central Lu, which reflects that Tmin is decreasing from the centre of the province to its sides.

The variance contributions of the second and third modes of Tmax and Tmin were 3.42% and 1.77% and 6.64% and 3.10%, respectively, and they are all typical and occasional spatial distribution types, but with great variability. Among them, the second mode of Tmax (Fig. 9a4) shows the inverse phase distribution mode of "negative in southwest and positive in northeast". The centres of positive and negative values are located in the Dingtao and Cheng Shantuo regions in the southwest Lu and east Lu, respectively, i.e. with low temperatures in the southwest Lu climate region, while high temperatures are in the east Lu climate region or with high temperatures in the southwest Lu climate region and low temperatures in the east Lu climate region. Meanwhile, the law of change of characteristic vector values shows a gradual Tmax increase from southwest to northeast. The third eigenvector (Fig. 9a5) shows the inverse mode of phase distribution of "positive in the northwest and negative in the southeast" with the centres of positive and negative values located in the Jinan and Rizhao regions in the central Lu and south-central Lu, respectively. That is, the temperature in the central Lu climatic region is high, while the temperature in the south-central Lu is low or vice versa. The difference is that the second mode of Tmin (Fig. 9a7) shows a "positive–negative–positive–negative" distribution pattern from southwest to northeast, with the centre of positive values located in the Weifang region of central Lu, and the centre of negative values located in the Fushan region of east Lu, which reflects that these two climatic regions are sensitive areas to changes in Tmin. The third mode (Fig. 9a8) shows the distribution in the same direction as the second mode, but in anti-phase, with the centre of positive values located in the Jinan region of central Lu, indicating large temperature variations in this region. It can be seen that the second and third modes of Tmax mainly show a band-shaped distribution in space, while Tmin mainly shows a regional peak–valley distribution.

In terms of temporal distribution, Tmean, Tmax and Tmin have four, six and six manifestations, respectively. Statistical analysis of Tmean, Tmax and Tmin results (Table 5) shows that for Tmean, the first mode has 30a and accounts for 83.33%, while the second mode has 3a and accounts for 8.33%, indicating that Tmean change is dominated by the first mode. At the same time, from Figs. 9b1 and 10a1, the first mode of Tmean field shows a significant upward trend (P < 0.01) and exceeded the significance level of 0.05 in 1996, indicating that Shandong Province has a cyclic warming of about 10a and significant warming trend after 1996. At the same time, shown in Figs. 9b2 and 10a2, the second mode of Tmean field shows a significant decreasing trend (P < 0.05), with positive values prevailing from 1988 to 2009 and negative values prevailing after the mutation in 2009 and exceeding the significance level of 0.05 in 2018. This indicates that the second mode of the Tmean field has a significant tendency towards "high temperatures in the south area and low temperatures in the north area" since 2009.

Table 5 Distribution of Tmean, Tmax and Tmin performance types by years
Full size table
Fig. 10
figure 10

Plot of Tmean, Tmax and Tmin time coefficients M–K mutation test

Full size image

The six manifestations of Tmax (Table 5) show that belonging to the first mode has 32a, (88.89%), and belonging to the second mode has 2a (5.56%), indicating that the first mode is a dominant mode. At the same time, the first mode of the Tmax field shows a significant increasing trend (P < 0.01), with all positive value after 1986, and exceeded the significance level of 0.05 in 1997, as shown in Figs. 9b3 and 10a3, indicating that the increasing trend of Tmax in Shandong Province will become increasingly significant. From Figs. 9b4 and 10a4, the second mode of the Tmax field shows a non-significant decreasing trend (P > 0.05), with mutations in 1987 and 2018, and predominantly negative values after 2012 that did not reach the significance level of 0.05. From this, it is clear that the spatial distribution of the second mode of the Tmax field has a tendency towards "high temperatures in the southwest and low temperatures in the northeast". Figures 9b5 and 10a5 show that the third mode of the Tmax field showed a significant downward trend (P < 0.01), and the mutation occurred in 1994 and exceeded the significance level of 0.05 in 1996. This indicates that the third mode of the spatial distribution of the Tmax field has a significant tendency towards "low temperature in the northwest and high temperature in the southeast".

The six manifestations of Tmin (Table 5) belong to the first mode for 25a (69.44%), the second mode for 6a (16.67%) and the third mode for 5a (13.89%). Among them, the first mode of the Tmin field shows a significant warming trend (P < 0.01) as shown in Figs. 9b6 and 10a6, and exceeds the significance level of 0.05 in 1998, which indicates that Tmin warming trend in Shandong Province will become increasingly significant. From Figs. 9b7 and 10a7, the second mode of the Tmin field shows a non-significant warming trend (P > 0.05), dominated by negative values from 1988 to 2012 and positive values after 2012, indicating that the second mode of the Tmin field will continue its current distribution pattern. In contrast, Figs. 9b8 and 10a8 shows that the third mode of the Tmin field shows a non-significant decreasing trend (P > 0.05) with a mutation occurring in 1994 and exceeding the significance level of 0.05 in 2001, but did not reach the significance level of 0.05 in 2018. This indicates a non-significant tendency towards the reverse development of the spatial distribution of the third mode of the Tmin field.

3.3.2 Precipitation

As given in Table 4, the cumulative variance contribution of the first two modes of Pmean reached 66.95%, and among them, the first modal eigenvector with a variance contribution of 53.87% is the dominant spatial distribution type. All values of the first modal eigenvector are positive (Fig. 11a1), which indicates an overall spatial consistency in precipitation change, i.e. precipitation increases or decreases during the same period. At the same time, the high-value centre is located in the Taishan region in central Lu and the low-value centre is located in the Longkou region in east Lu. This indicates that the Taishan region is more sensitive to precipitation changes and the change is large. Pmean in Shandong Province is influenced by topography, with a spatial distribution characterized by higher precipitation in the centre of the province and lower precipitation around the centre, as well as higher precipitation in the southeast and lower in the northwest of the province. The second modal eigenvector has a variance contribution of 13.07%, which is typical for this type of distribution. As shown in Fig. 11a2, the modal pattern of distribution is roughly divided by the boundary of Juxian–Weifang–Feixian, with negative values in the west and positive values in the east. The centres of positive and negative values are Taishan in central Lu and Weihai in east Lu, respectively, showing an inverse phase distribution with "negative values in the northwest and positive values in northeast". That is, there is little rain in the southwest and plenty in the northeast or there is plenty rain in the southwest and little in the northeast. The distribution of eigenvector value reflects that precipitation also increases from west to east. It can be seen that Pmean has two spatial types of distribution: "overall consistency" and "northwest–northeast" anti-phase.

Fig. 11
figure 11

EOF analysis of Pmean and S-Hmean in Shandong Province and its time factor trend

Full size image

Regarding the temporal distribution, Pmean has four manifestations, and statistical analysis of the results (Table 6) shows that the first mode has 20a (55.56%) and the second mode has 8a (22.22%), which shows that the first mode is the decisive mode. Figures 11b1 and 12a1 show that the first mode of the Pmean field showed an insignificant increasing trend (P > 0.05). It had mutations in 1985, 1990 and 1992 and generally had a positive value after 1993. However, none of them reached the significance level of 0.05. This shows that Pmean will have an insignificant increasing trend. Meanwhile, the second mode of the Pmean field shows an insignificant downward trend (P > 0.05) as shown in Figs. 11b2 and 12a2, with many mutation points, but none reached the significance level of 0.05. From this, it is clear that the second mode of the Pmean field has a non-significant tendency towards lower precipitation in the centre of the province and more around it, as well as lower precipitation in the southeast and more precipitation in the northwest of the province.

Table 6 Distribution of Pmean, S-Hmean, W-Smean and W-Smax performance types by year
Full size table
Fig. 12
figure 12

Pmean and S-Hmean time coefficients M–K mutation test plots

Full size image

3.3.3 Sunshine hours

As given in Table 4, the cumulative variance contribution of the first two modes of S-Hmean reached 70.58%. The variance contribution of the first modal eigenvector is 58.63%, making it the main type of spatial distribution. As shown in Fig. 11a3, the first modal eigenvector values are all positive, indicating the overall consistency in the spatial variation of S-Hmean in Shandong Province, i.e. an increase or decrease in the number of sunshine hours during the same period. The centre of the high value is in Qingdao area in east Lu, and the centre of the low value is in Fushan area in east Lu. This reflects that the degree of change in the Rizhao region in east Lu is much lower than that in the Qingdao region in east Lu, and the amount of change in sunshine hours is much higher in the Qingdao area than in other areas. The second modal eigenvector has a variance contribution of 11.95%, which is typical for this type of spatial distribution. As shown in Fig. 11a4, the distribution pattern is bounded by Qingdao–Lingxian–Juxian, with negative values in the southwest and positive values in the northeast. The negative value centre is in the Yanzhou area in southwest Lu, and the positive value centre is in the Weihai area in east Lu, showing the northeast–southwest inverse phase distribution mode with fewer sunshine hours in southwest Lu and more in east Lu or more sunshine hours in southwest Lu and fewer in east Lu. This indicates that the northeast–southwest distribution is an important form of the distribution pattern.

In terms of temporal distribution, S-Hmean have four manifestations. Statistical analysis of the results (Table 6) shows that the first mode has 20a (55.56%) and the second mode has 9a (25%). This shows that other modal changes have relatively large impact on S-Hmean. At the same time, the first mode of the S-Hmean field shows a significant decreasing trend (P < 0.01) (Figs. 11b3 and 12a3), with a predominantly negative values after 1989 and a mutation in 1998, and exceeded the significance level of 0.05 in 2003, which indicates a significant and continuous decreasing trend in S-Hmean. As shown in Figs. 11b4 and 12a4, the second mode of the S-Hmean field shows a non-significant decreasing trend (P > 0.05), dominated by positive values from 1987 to 2015, with mutation in 2015, and predominantly negative values that did not reach the significance level of 0.05 after the mutation. From this, it is clear that the second mode of S-Hmean has a phase variation with a period of 5–8 years, and their spatial distribution type has a non-significant tendency towards "low sunshine in the northeast and high sunshine in the southwest".

3.3.4 Wind speed

Table 4 shows that the cumulative variance contribution of the first two modes of W-Smean and W-Smax reached 81.21% and 82.36%, respectively. Both are effectively reflecting the variability of the two annual mean wind speeds, but with differences. Among them, the variance contribution rate of the first mode eigenvector in W-Smean reaches 71.11%, making it the main spatial distribution. The eigenvector values of the first mode (Fig. 13a1) are all positive, which indicates that changes in W-Smean have overall spatial consistency. The centres of high values are mainly located in Qingdao in east Lu and Weifang in central Lu, indicating that the two regions are more sensitive to changes in wind speed. At the same time, the eigenvector value reflects that W-Smean is increasing from southwest to northeast. Similarly, the eigenvector values of the first mode (Fig. 13a3) of W-Smax are all positive. The centre of high value is located in Haiyang area in east Lu, and the centre of low value is located in Dingtao area in southwest Lu. This indicates that W-Smax in the east Lu climate region is highly variable and much more variable than in the southwest Lu climate region, while other climate regions are transitional regions.

Fig. 13
figure 13

W-Smean and W-Smax EOF analysis and its time factor trend in Shandong Province

Full size image

The variance contribution of the second modal eigenvectors of W-Smean and W-Smax were 10.10% and 9.08%, respectively, both typical for this type of spatial distribution. Among them, the second mode of W-Smean (Fig. 13a2) shows an approximately tripolar spatial distribution mode of "positive, negative, positive". Areas of low values are mainly located in Qingdao in east Lu, Kenli in northwest Lu, Taishan and Weifang in central Lu and Feixian in south-central Lu, indicating that variations in W-Smean in these areas are smaller than in other climatic regions. The difference is that the second mode of W-Smax (Fig. 13a4) is spatially bounded by Qingdao–Weifang–Hui Minxian, with a region of positive values in the north and a region of negative values in the south, showing a "positive north and negative south" mode of spatial distribution. The positive value centre is located in the Cheng Shantou region in east Lu, while the negative value centre is located in the Rizhao region in south-central Lu, indicating that the central Lu climate region is more sensitive to changes in maximum wind speed, and its eigenvector values approximately show an increase from south to northeast, which reflects the same law of changes in W-Smax.

In terms of temporal distribution, both W-Smean and W-Smax have four manifestations. Statistical analysis of the results (Table 6) shows that for W-Smean there is 33a in the first mode (91.67%), while the other modes have 3a (8.33%). For W-Smax, there are 27a in the first mode (75%) and there are 4a in the second mode (11.11%). It can be seen that the first mode is the dominant mode. Figures 13b1 and 14a1 show that the first mode of the W-Smean field showed a significant downward trend (P < 0.01). After 2000, it was mostly negative and exceeded the significance level of 0.05 in 2005. This indicated a more significant downward trend in W-Smean in Shandong Province. As shown in Figs. 13b2 and 14a2, the second mode of the W-Smean field shows a significant increasing trend (P < 0.05), with mutations in 1989 and 2012, and predominantly positive values after 1989. It can be seen that the second mode of the W-Smean field has a cyclic law of variation with a base value of 10a and will maintain its current spatial distribution mode. Figures 13b3 and 14a3 show that the first mode of the W-Smax field had a significant downward trend (P < 0.01), which was mostly negative after 1985, and exceeded the significance level of 0.05 in 1993, which indicates that the downward trend of W-Smax will continue to intensify and that its contribution ratio to the decrease in wind speed will be more significant. Figures 13b4 and 14a4 show that the second mode of the W-Smax field showed an insignificant downward trend (P > 0.05), with negative values in 1989–2006 and mutations in 1989, 2002 and 2014. However, it did not reach the significance level of 0.05. This indicates that the second mode of the W-Smax field in Shandong Province has significant periodic changes and its spatial distribution type has a non-significant tendency towards low wind speeds in the north and high wind speeds in the south.

Fig. 14
figure 14

W-Smean and W-Smax time coefficients M–K mutation test plots

Full size image

3.4 R/S projections of climate elements

3.4.1 Analysis of future climate trends

After analysis by the EOF method, there is an in-depth study of regional changes in different climate elements, but future trends are still not clear. Therefore, based on the principles of R/S analysis, the five climate regions of Shandong Province were analysed to determine the future trend of changes of each climate element in the province (Table 7).

Table 7 Hurst indices for mutations of climatic elements by regional stations in Shandong Province, 1984–2019
Full size table

Table 7 shows that the Hurst index is above 0.75 for all three types of mean annual temperature, indicating their high relevance. The future has a consistent trend of change with the past, i.e. temperatures will continue to rise with the most significant increase in Tmin, which is a further response to the global temperature increase.

In the same way as the temperature trend, the series of Pmean in Shandong Province are also correlated, but relatively weak. This suggests that precipitation trends in the past and in the future will remain consistent, i.e. there is still an overall increasing trend, but with lower persistence and significant regional differences. Northwest Lu region has the reverse persistence (Hurst < 0.5), i.e. a decreasing trend in precipitation, while the other climatic regions have positive persistence with consistency between future and past increasing trends. Among them, east Lu and southwest Lu have higher persistence, while central Lu and south-central Lu have weaker persistence.

In contrast to the precipitation trend, the Hurst values for S-Hmean, W-Smean and W-Smax are all above 0.81. This indicates that they all have a very high degree of correlation, i.e. a consistent future trend with the past and a very high degree of persistence. Among wind speeds, the persistence of W-Smean is more significant, followed by W-Smax. The trends of W-Smax and W-Smean are highly similar, and the future W-Smean and W-Smean follow the same trend as in the past 36 years (Hurst > 0.9), with a significant decreasing trend. Although the degree of correlation is weaker for S-Hmean, it is still significant and has a similar trend of change as the wind speed, i.e. the number of sunshine hours will continue to decrease in the future.

3.4.2 Analysis of the average cycle length of climate elements

As shown in Fig. 15, the future trends of each climate element and the magnitude of the climate propensity rate in Shandong Province and the five climate regions are consistent with the results of the Hurst index (Table 7). From Fig. 15d, the first obvious turning point of Pmean in Shandong Province occurs at ln(n) = 2.079, which corresponds to a time length of n = e, 2.079 ≈ 10a, indicating that the effective impact of past trends on future trends is 10 years. The first obvious turning point of W-Smax (Fig. 15g) is located at ln(n) = 2.197, whose corresponding time length is n = e, 2.197 ≈ 12a, indicating that the past trend will have an effective impact on the future trend for 12 years. From Fig. 15, we also know that the cycle lengths of the time series of Tmean, Tmax, Tmin, S-Hmean and W-Smean are 12a, 10a, 12a, 10a and 10a, respectively. The cycle lengths of different climatic elements in the climatic regions of east Lu, northwest Lu, southwest Lu, central Lu and south-central Lu are shown in Table 8. Each climate element shows a positive correlation with each other, i.e. the stronger the persistence, the longer the cyclic period.

Fig. 15
figure 15

Variation curves of V statistic values vs ln(n) based on each climatic element in Shandong Province

Full size image
Table 8 Cycle length of each climate element time series/a
Full size table

4 Discussion and conclusions

Overall, in the context of global warming, Tmean, Tmax and Tmin showed increasing trends in Shandong Province. Among them, trend in Tmin is the most significant at 0.547 °C per decade, followed by Tmean trend of 0.446 °C per decade, and the weakest trend in Tmax of 0.391 °C per decade. Meanwhile, the Tmean, Tmax and Tmin mutation times in Shandong Province were concentrated in the late 1980s, mid-late 1990s and early twenty-first century. The research results are similar to the research conclusions of Liu (2015). that sudden temperature changes in China occurred in the mid to late 1990s, but later than the mutation time in the late 1980s and early 1990s obtained by Meng et al. (2012) and Wang et al. (2013). This may be related to differences in study period, geographical environment, site selection and data processing, as well as the Blue Book on Climate Change in China, which pointed out that in 2017, the Asian average surface temperature was 0.74 °C higher than the average value from 1981 to 2010 (CMA 2018). On a regional scale, Tmean, Tmin and Tmax warming trends show maximum values in the south-central Lu. However, Tmean and Tmin warming trends show minimum values in east Lu, while Tmax warming trends show minimum values in central Lu. This may be related to factors such as sea and land differences, altitude and other factors (Wang et al. 2013). The results of the study were similar to those of Han et al. (2013), but in terms of mutation time, with the exception of east Lu climate region, all other climate regions are not in line with the research conclusions of Yin et al. (2009). The reason for this regional difference may be that, although temperature change in China is mainly controlled by large influencing factors (e.g. atmospheric circulations, ocean currents), the local natural environment and human activities in different regions also influence the spatial and temporal distribution of temperature change (Liu 2015). Pmean in Shandong Province shows a non-significant increasing trend and has a band-like distribution characterized by a gradual decrease from southeast to northwest. The results are basically consistent with the results of Xu et al. (2018) and Dong et al. (2014), but in contrast to those of Ma et al. (2015) and Lu et al. (2021), Furthermore, the mutation time is earlier than that of Xu et al. (2018), and this may be related to the number of selected stations, study period and different data interpolation methods. The Blue Book on Climate Change in China pointed out that since the beginning of the twenty-first century, the mean annual precipitation in North, South and Northwest China has fluctuated and increased, and the precipitation has continued to be higher since 2012 (CMA 2018). There is an overall consistent and significant mutation downward trend in S-Hmean, W-Smean and W-Smax. Among them, the mutation time of S-Hmean in each climate region occurred approximately in the late 1990s and early twenty-first century. The spatial distribution of S-Hmean in Shandong Province shows higher values in the north and lower in the south, as well as higher values in the northwest and lower in the southeast (Ma et al. 2020), but this is not in line with the Wang et al. (2011) findings that S-Hmean in Shandong Province have a distinct geographical distribution of "high values on both sides and low values in the middle". This may be related to the difference in the number of stations and study period, as well as the influence of the increasing trend of cloud cover. W-Smean and W-Smax both show a significant decreasing trend, with the mutation time occurring in the 1990s and early twenty-first century. W-Smax contributes the most to the decreasing trend, with maximum wind speeds higher in coastal and mountainous areas in central Lu and lower in southern and southwest Lu (Dong et al. 2018). This is the later mutation time compared to the findings of Guo (2015a, b) (early to late 1980s), which may be related to differences in environment, altitude, longitude and latitude. Meanwhile, Yang et al. (2009) showed that a decrease in wind speed can accelerate the reduction in sunshine hours and indicated that wind speed has a positive correlation with sunshine hours and significantly affects them. Satheesh et al. (2005) have shown that wind speed makes a significant contribution to global radiation, and therefore, wind speed can be an important driver of changes in sunshine hours.

The EOF method shows that the spatial variability of each climate element in Shandong Province is dominated by the overall consistent distribution pattern, but with variability between each climatic element. The spatial variability rate of Tmean generally has two distribution types: "overall consistent" type and "north–south anti-phase" type. The spatial variability rate of Tmax generally has three distribution types, i.e. "overall consistent" type, "southwest–northeast" anti-phase type and "northwest–southeast" anti-phase type. Furthermore, the spatial variability rate of Tmin generally has three distributed types, i.e. "overall consistent" type, "positive, negative, positive, negative" type from southwest to northeast and "negative, positive, negative, positive" type from southwest to northeast. This is related to the wider environmental factors of the global temperature rise, the landscape features, the influence of polar cold air and the enhanced atmospheric greenhouse effect (Yuan et al. 2009; Zhang and Song 2018; Zhao et al. 2005). The results of this study are similar to those of Li et al. (2003) and Jia (2008), The spatial variability rate of Pmean mainly consists of two distribution types, i.e. "overall consistent" type and "northwest–northeast" anti-phase type, which is related to the topography of the Shandong Peninsula and the influence of the southwesterly flow from the west side of the sub-high and anomalous anticyclonic circulation over East Asia in summer and winter, respectively (Ma et al. 2006). The results are similar to those of Gao et al. (2005) and Yu et al. (2011), The spatial variability rate of W-Smean generally has two distribution types, namely the "overall consistent" type and the opposite phase of the "northeast–southwest" type. This may be related to the influence of the unique landform of Shandong Province and its altitude, frequent human and urban construction activities on cloud density and aerosol concentration (Guo et al. 2010; Xiao et al. 2020), increase in relative humidity and light fog (Sun et al. 2013), and the interaction between wind speed and aerosol load that can drive changes in sunshine hours (Yang et al. 2009). The spatial variability rates of both W-Smean and W-Smax have two types of distribution, i.e. the "overall consistent" type and the "positive, negative, positive" polar type, and the "overall consistent" type and "north–south anti-phase" type, respectively. This is closely related to the topographic complexity of the Shandong Peninsula, the development of urbanization and the reduction of air pressure (Ren et al. 2005).

In terms of time distribution characteristics, each climate element field is dominated by the first mode, with six forms of distribution for Tmax and Tmin, but four forms of distribution for every other climate element. At the same time, the first modes of Tmean, Tmax and Tmin fields show that the temperatures have a significant upward trend. The second mode shows that the other two temperatures, except for Tmin that keeps the same distribution direction, have the reverse development trend, but the significance is different. The third mode of Tmax and Tmin fields shows that these temperatures have the opposite development trend, but the significance is different. The first mode of the Pmean field shows that the precipitation has an insignificant upward trend, and the second mode shows that the precipitation have the opposite direction with an insignificant trend. The first mode of the S-Hmean, W-Smean and W-Smax fields all show a significant downward trend. The second mode of the S-Hmean and W-Smax fields shows that there is an insignificant trend with reverse development, but the second mode of the W-Smean field shows a significant trend and maintains the current spatial distribution. Each of the climate elements underwent a significant turnaround around the 1990s, reflecting relative uniformity in the timing of mutation of the climate elements in Shandong Province.

R/S analysis was performed in relation to temperature, precipitation, duration of insolation and wind speed in Shandong Province. The results are similar to those of Ren et al. (2012), which showed that temperature will continue to increase and sunshine hours will continue to decrease in southwest Lu in the future. The four climate elements have significant Hurst value (H > 0.5) in Shandong Province, and both future trends will remain in line with those of the past 36a, dominated by warm and wet trends and low wind speed and low sunshine hours trends. There are regional differences in the H-values of each climate element, and the trends of change are also different, which creates differences in future trends. Among them, the increase in precipitation in northwest Lu may be related to the cold phase of "Ramadere" in Shandong Province after 2003 (Chen et al. 2016), but the future trend in northwest Lu is reversed (H < 0.5), i.e. a decreasing trend may be related to the constant disturbances of the large topography and land and sea under atmospheric circulation (Yu et al. 2011) and the blocking effect of the Taishan Mountains (Yan 2013). Therefore, the climate region of northwest Lu will be dominated by a warm and dry trend and low wind speed and low sunshine hours trends in the future, while the other climate regions will be dominated by warm and wet trend and low wind speed and low sunshine hours trends in the future. Each climate element shows a positive correlation with each other, i.e. the stronger the persistence, the longer the cyclic period.

Climate change further aggravates the frequency and intensity of natural disasters in Shandong Province and has a significant impact on buildings, World Heritage protection, agriculture, forestry and human living environment. It also affects sustainable development, alleviation of ecological vulnerability, increase in ecological resilience and improvement of human habitat (Feng 2017; Wang 2020; Guo and Wang  2010; Duan et al. 2014). With increasing precipitation, urban drainage pressure and risk of urban flooding will increase, and the overall rain island effect in Shandong Province increases from inland to the coast (Lu et al. 2018; Qiao et al. 2014). The increase in temperatures and precipitation will lead to reduced summer and winter fires in cities, respectively (Liu et al. 2013). It also leads to a downward trend in the total number of heating and cooling days in Shandong Province, which has a positive effect on reducing energy consumption in residential buildings (Shi et al. 2011). The urban heat island effect in Shandong Province will further increase as wind speed continues to decline (Gao et al. 2014). Rapid urbanization has resulted in lower warming trends (except in summer) and lower rates of wind and precipitation increase in cities compared to rural areas. It also aggravated hazy weather, which further led to a decreasing trend in the sunshine hours in Shandong Province (Liu et al. 2015; Ji et al. 2020). Increase in temperature and decrease in sunshine hours have led to a decrease in the biodiversity of World Heritage Sites, an increase in plant endangerment levels and a chaotic distribution of species (Baral et al. 2017; Song 2017; Allan et al. 2017; Xu 2010). The increase in precipitation will also reduce the growth and stress resistance of the ancient Chinese pines in Tai Shan, and will also increase the degree of corrosion and damage to the stone carvings and walls of cultural sites (Qi 2008; Wang 2016). The increase in temperature causes sea level rise, and increase in precipitation triggers flooding. This can also increase the incidence of chronic diseases, respiratory diseases, mental trauma and infectious diseases (Shen and Wang 2013; Gao et al. 2013). Increasing the frequency and intensity of heat waves in the North China Plain due to higher temperatures and lower wind speeds will increase morbidity and mortality from heat-related diseases (Liu et al. 2008). However, increased precipitation can effectively improve the air in Shandong Province. Warming and decrease in sunshine hours also contribute to increasing climate comfort period and comfort levels in Shandong Province (Zhu et al. 2021). The increase in temperature has led to an increase in the use of pesticides, which leads to increased pollution of water resources. Warming and reduction in sunshine hours will lead to a northward shift of species boundaries, early greening and maturity of winter wheat in Shandong Province, resulting in reduced crop yields, as well as increased land erosion and desertification (Xu et al. 2010; Xiao et al. 2012; Liu et al. 2020a, b, c). The future "warm and humid" climatic conditions in Shandong Province will be conducive to the increase in the climate productivity. However, by increasing the intensity of regional precipitation, this will lead to more difficult development and use of surface water resources, and will increase the pressure on agricultural water security (Liu et al. 2019; Liu et al. 2018a, b; Pang et al. 2021; Shen et al. 2021). As climate warming increases, the frequency and intensity of global forest fires are also increasing. The level of fire risk in Shandong Province is increasing due to the increase in temperatures (Wei et al. 2020; Huang et al. 2014).