Changes in the mean and variability of temperature and precipitation over global land areas

The frequency and intensity of extreme weather and climate events may change in response to shifts in the mean and variability of climate, which pose high risks to societies and natural ecosystems. Gridded near-surface temperature, precipitation, and the number of wet days from the Climatic Research Unit dataset were analyzed for two 30 year periods to explore changes in the mean and variability of temperature and precipitation over global land areas in the recent period (1991–2020) compared to the reference period (1951–1980). Global land areas are characterized by warmer and slightly wetter conditions in the recent period, while the variability of temperature and precipitation has remained nearly unchanged. Changes in the mean and variability of both temperature and precipitation are also analyzed over tropical, subtropical, and midlatitude land areas. The annual mean temperature over all these three latitudinal regions has increased in the recent period compared to the reference period, with the highest increase in subtropical and midlatitude land areas (0.7 ∘C), followed by tropical land areas (0.5 ∘C), while temperature variability has remained nearly unchanged. The annual precipitation has decreased over tropical, subtropical, and midlatitude land areas in the recent period compared to the reference period. Precipitation variability has not changed considerably over subtropical land areas. However, it has substantially increased over tropical land areas, which indicates a higher risk of droughts and periods of excess water in the recent period. In contrast, precipitation variability has decreased over midlatitude land areas, indicating narrower swings between wet and dry conditions, which decrease the risk of droughts and periods of excess water in the recent period.


Introduction
Shifts in the mean and variability of temperature and precipitation largely influence the frequency and intensity of extreme weather and climate events (e.g. Schneider et al 2015, van der Wiel and Bintanja 2021, Zhang et al 2021. For instance, the rise in temperature has been associated with an increase in the frequency of heatwaves over the terrestrial Arctic in the last decades (Dobricic et al 2020). The higher mean temperature is associated with higher atmospheric moisture content (Santer et al 2007) according to the Clausius-Clapeyron relation. The higher moisture content contributes to accelerating the overall hydrological cycle, intensifying extreme precipitation in wet regions (Allan andSoden 2008, Li et al 2018) and dry extremes in arid regions (e.g. Alizadeh-Choobari and Najafi 2018). Indeed, some evidence suggests that the intensity of extreme precipitation events (Hegerl et al 2004, O'Gorman 2012 and the risk of both flooding (e.g. Tabari 2020) and drought (Balting et al 2021) increase with global warming.
It is argued that more frequent heatwaves and high-temperature events will occur in the future climate (e.g. Guo et al 2022) mainly in response to an increase in the mean temperature. Changes in the frequency of short-term (daily and sub-daily) extreme precipitation are also arguably largely related to changes in the mean temperature (Westra et al 2013). On the other hand, future changes in the longer-term extreme precipitation such as droughts or periods of excess water (to some extent the Clausius-Clapeyron relation and general intensification of the hydrological cycle, but also shifting atmospheric currents) are mainly governed by changes in climate variability (van der Wiel and Bintanja 2021). An increase in the occurrence of extreme weather and climate events poses high risks to societies (e.g. Clarke et al 2021) and natural ecosystems (e.g. Ummenhofer and Meehl 2017). Therefore, exploring changes in the mean and variability of temperature and precipitation is important.
Climate models suggest that global precipitation's mean and variability increase under future global warming (e.g. Held and Soden 2006, Pendergrass et al 2017, Zhang et al 2021. There is also evidence of changes in temperature variability in response to future global warming (Suarez-Gutierrez et al 2020, Kotz et al 2021). However, the outputs of these models cannot be directly and continuously evaluated against observational data because the time horizon is too long (Baumberger et al 2017). Instead, climate model projections are generally evaluated indirectly based on the performance of these models in simulating important features of the current and past climate. The confidence of this approach, however, could be artificially high if the performance of a climate model in the past is assessed by comparison against observational data that were initially used to develop the model (Knutti et al 2010). Another indirect way to evaluate the projection of a climate model is to assess how well model results agree with the results of other models (Lloyd 2010). However, such an assessment of the robustness of the results of climate models may not be a good measure to determine the reliability of climate projections because, as outlined by Baumberger et al (2017), similar common biases in different climate models may lead to overall agreement in some important features. For instance, most current climate models simulate higher warming in the eastern tropical South Pacific during recent decades (e.g. Marjani et al 2019), while observational data indicate that higher warming has occurred in the western tropical Pacific, and the eastern tropical South Pacific has experienced cooling (Clement et al 1996, Lian et al 2018, Seager et al 2019, Alizadeh 2022. Despite these shortcomings, there is considerable evidence of an accurate representation of many well-understood physical processes by climate models, e.g. for short-term rainfall extremes (Fischer and Knutti 2016) or other rainfall characteristics (Allen and Ingram 2002). On the other hand, while observed changes do give a good indication of how future aspects of climate may develop under anthropogenic warming, the total reliance on such an approach is limited by the non-linearity in the climate system (Good et al 2016).
To better rely on projections of changes in the mean and variability of temperature and precipitation under future global warming, we first need to explore changes that have already occurred in recent decades under global warming. Indeed, past climatic changes that can be attributed to recent global warming can be used as evidence as to what may happen in the future, although increasing temperatures may manifest more at some periods. In this study, the Climatic Research Unit (CRU) gridded dataset was analyzed for two 30 year periods to explore changes in the mean and variability of temperature and precipitation over global land areas in the recent period (1991-2020) relative to a climatological reference period . These changes are also analyzed over tropical, subtropical, and midlatitude land areas. The analysis is only conducted over land areas where the impact of extreme weather and climate events is more felt by humans. CRU is also a land-only dataset.

Data description and methodology
Gridded near-surface temperature, precipitation, and the number of wet days from version 4 of the monthly CRU dataset (Harris et al 2020) with a horizontal resolution of 0.5 • × 0.5 • were obtained and analyzed in the reference period and the recent period. A wet day in the CRU data is defined as a day receiving ⩾ 0.1 mm of precipitation. The CRU dataset is based on the interpolation of the observed values from extensive networks of weather stations (Harris et al 2020). Based on these data (without detrending), changes in temperature, precipitation, and the number of wet days were analyzed over global land areas (except Antarctica) in the recent period relative to the climatological reference period. Note that based on temperature data from the National Oceanic and Atmospheric Administration, the reference period largely preceded the rapid global warming trend that started in 1981. In addition, statistical characteristics of temperature and precipitation and their changes were explored over global land areas in recent decades based on the analysis of the probability density function (PDF). To this end, time series of both global land temperature and precipitation were obtained for both the reference period and the recent period, and then their PDFs were analyzed.
In this study, the global land averaged values of temperature and precipitation were derived first, and then their PDFs were calculated, implying that the PDFs show the distribution of the average land temperature and precipitation. The linear trend was first removed from the original time series of temperature and precipitation, and then the PDFs were obtained and analyzed because the analysis indicates that the larger trend in temperature/precipitation in the recent period is significantly contributing to the increased standard deviation in this period. To calculate the PDF of temperature and precipitation from monthly data, at first 12 month running averages of the time series of both global land temperature and precipitation were taken separately for the reference and recent periods. The PDFs of temperature and precipitation were then calculated based on the obtained 12 month running averages. This implies that the PDF analysis in this study is based on annual means of temperature and precipitation.
The PDF of temperature and precipitation was also analyzed based on the same approach over tropical, subtropical, and midlatitude land areas for both the reference and recent periods. In this study, the tropics are defined as the area between 23.25 • S and 23.25 • N, the subtropics cover the latitudes from 23.25 • to approximately 33.25 • in the Northern and Southern hemispheres, and midlatitudes cover the latitudes from 33.25 • to approximately 66.25 • in the Northern and Southern hemispheres. Note that the width of the PDF indicates the range of variability, such that an expansion of the width indicates an increase in variability and vice versa. Figure 1 shows the geographical distribution of near-surface temperature changes in the recent period relative to the reference period over global land areas. The spatial pattern of changes in temperature shows warming over virtually all land areas. However, the rate of warming is not uniform across different land areas. The level of warming is particularly much higher over high-latitude land areas of the Northern Hemisphere with values exceeding 1.8 • C, which is partly due to the retreat of ice and snow in recent decades and the involvement of the surface albedo positive feedback (e.g. Alizadeh and Lin 2021). The much higher warming over high-latitude land areas of the Northern Hemisphere could partly explain a substantial increase in the frequency of heatwaves in some parts of the terrestrial Arctic in recent decades (Dobricic et al 2020), although it can also be related to an increase in temperature variability in recent decades. Despite the warming of most land areas in the recent period, the temperature has not changed over central South America (figure 1), which is attributed to an eastward shift of the Southeast Pacific Subtropical Anticyclone (Zou and Xi 2021). Figure 2 shows the geographical distribution of precipitation changes in the recent period relative to the reference period. Note substantial regional differences in precipitation changes, such that some regions experienced larger increases, while decreases in precipitation have also been detected, implying that precipitation changes are not solely caused by thermodynamic processes. Instead, differences in the pattern of regional precipitation changes are partly caused by a forced dynamic change in the atmospheric circulation (Fu 2015), which may lead to changes in weather patterns (Hu et al 2013, Smeed et al 2018. For example, as argued by Fu (2015), variability in the intensity and width of the Hadley circulation is central in determining regionally different precipitation changes in the margins of tropics-subtropics and subtropics-midlatitudes. In addition, regional intrinsic variability partly contributes to regional differences in precipitation changes.

Temperature and precipitation changes over global land areas
A decrease in precipitation over parts of the semi-arid to hyperarid regions of subtropical North Africa and southeastern Iran is notable, exacerbating the already dry climate of these regions, which impose a particular concern because water resources are scarce in regions with the semi-arid to hyperarid climate (Sowers et al 2011). Precipitation has also largely decreased over Baffin Island, lying between Greenland and the Canadian mainland. A precipitation deficit in the recent period has also occurred in many parts of Africa. Although the overall rising temperature is associated with an increase in atmospheric moisture, precipitation has decreased over these land areas due to changes in atmospheric circulation and ocean currents that drive the global climate (Hu et al 2013, Smeed et al 2018. By contrast, an increase in precipitation in the recent period can be particularly observed in Scandinavia, parts of Asia, parts of northern Australia, the eastern USA, Alaska, Iceland, and parts of Greenland. Figure 3 shows changes in the number of wet days in the recent period relative to the reference period. Changes in the number of wet days as the climate warms can either reinforce or counteract an increase in the intensity of daily precipitation. More frequent wet days in the recent period are particularly larger over the Indian subcontinent, parts of western Australia, and the margins of southwestern Greenland. Despite an increase in annual precipitation over parts of Asia in the recent period (figure 2), the frequency of wet days has largely decreased in many parts of Asia and some parts of Europe, except in most parts of the Indian subcontinent where the monsoon system dominates ( figure 3). This implies that the identified increase in the annual precipitation over parts of Asia (figure 2) can be ascribed to increases in the intensity of precipitation during wet days. In the Indian subcontinent, however, more frequent wet days in the recent period (figure 3) probably contributed to an increase in the annual precipitation ( figure 2). Indeed, the Indian subcontinent experienced up to 20% more wet days in the recent period relative to the reference period, while most of the other regions of Asia experienced up to 20% fewer wet days in the recent period. Less frequent wet days are also notable over parts of tropical and subtropical North Africa, southeastern Iran, Canada, the western margins of Peru, and parts of Southern Africa. Annual precipitation also decreased in these regions in the recent period, except over Canada (figure 2), implying that the decline in annual precipitation over these regions is primarily caused by less frequent wet days. Note that in subtropical North Africa where   figure 1, but for annual precipitation changes (%). Only statistically significant changes at the 95% confidence level are shown. A t-test was applied to determine if the changes are significant. precipitation is extremely low and infrequent, the identified decrease in the frequency of wet days contributed to exacerbating the already dry climate of the region. By contrast, in addition to the Indian subcontinent, there are more frequent wet days in parts of Western Australia, the eastern USA, and parts of Greenland in the recent period ( figure 3). Therefore, the increased annual precipitation over these regions (figure 2) is primarily caused by more frequent wet days. Figure 4 shows the PDF of the annual mean global land temperature and precipitation for both the reference and recent periods. Note that, as explained in section 2, the PDF analysis in this study is based on annual means of temperature and precipitation. It is not then surprising that for temperature, the low value for the later epoch is higher than the top of the PDF for the earlier epoch. The analysis of the CRU dataset indicates that the annual mean near-surface temperature and annual precipitation over global land areas are  1961-1980 and 1991-2020. Both temperature and precipitation over global land areas are characterized by an increase in the mean in the recent period, with increasing values of 1.0 • C and 1.2 mm (0.2%), respectively. The magnitude of global land warming is large, suggesting a need for prompt actions to reduce greenhouse gas emissions, which are the principal cause of global warming in recent decades. An increase in the mean of the global land temperature in the recent period contributes to the increased occurrence of monthly high-temperature events (van der Wiel and Bintanja 2021), which intensifies droughts (Kelley et al 2015, Mazdiyasni andAghaKouchak 2015) and may increase the global risk of heat-related mortality (Mora et al 2017). The magnitude of precipitation enhancement over global land areas, however, is too small.
At the global scale, an overall increase in annual precipitation in figure 4(b) is due to changes in the energy budget of the atmosphere, because changes in average precipitation are related to the availability of energy (Allen and Ingram 2002). Any increase in global precipitation extremes on the other hand is related to an increase in the moisture content of the atmosphere according to the Clausius-Clapeyron relation because precipitation extremes are related to the availability of moisture. This mechanism is known as the thermodynamic contribution to precipitation changes (Emori and Brown 2005) and can be understood as the wet-get-wetter mechanism (Held and Soden 2006), and the warmer-get-wetter mechanism (Xie et al 2010). However, global warming also raises the level of free convection, particularly over relatively dry land areas (Fasullo 2010), which inhibits or weakens convective precipitation. Changes in atmospheric dynamics in response to global warming are also highly complex (Emori and Brown 2005) and may regionally contribute to the weakening  or strengthening (Li et al 2019) of the thermodynamic effect. Therefore, global warming is generally associated with both regional increase and decrease in precipitation (see figure 2), such that only a slight change in precipitation is identified when averaged over global land areas ( figure 4(b)). On the other hand, at the regional scale, the thermodynamic contribution is more important for extreme precipitation, such that the results of previous studies indicate that under global warming, there is an increase in extreme precipitation in many regions (Pfahl et al 2017).
The variability of temperature and precipitation is also analyzed over global land areas. Temperature variability over global land areas has not changed significantly, with a standard deviation of 0.2 • C both in the reference and recent periods ( figure 4(a)). Similarly, precipitation variability over global land areas has not significantly changed on the interannual timescale in the recent period, with standard deviations of 10.3 mm and 9.8 mm in the reference and recent periods, respectively ( figure 4(b)). Figure 5 shows the PDF of annual mean temperature and precipitation over tropical, subtropical, and midlatitude land areas for both the reference and recent periods. The annual mean temperatures in all these three regions are characterized by an increase in the recent period, with the highest increase over subtropical and midlatitude land areas (0.7 • C), followed by tropical land areas (0.5 • C). Considering natural subtropical  (1951-1980, black lines) and the recent period (1991-2020, red lines). Vertical dashed lines denote climatological means of temperature and precipitation in the reference period (black dashed lines) and the recent period (red dashed lines). The global land averaged values of temperature and precipitation were derived first and then the PDFs were calculated. The linear trend was first removed from the original time series of temperature and precipitation and then the PDFs were obtained. aridity, this large warming over subtropical land areas in the recent period might have already contributed to drought intensification (Minetti et al 2019), such as the most severe 3 year drought in the instrumental record that occurred in Syria in recent years, which is attributed to human-induced climate change (Kelley et al 2015). Drought intensification is also projected over subtropical areas under future global warming (Balting et al 2021). Despite the warming of the examined areas, the annual precipitation has decreased in these areas, with the highest decrease over tropical land areas (−15.4 mm), followed in decreasing order over midlatitude (−10.8 mm) and subtropical land areas (−5.3 mm). In terms of changes in precipitation in percent, however, the highest decrease has occurred in midlatitude land areas (−1.2%), followed in decreasing order over tropical (−1%) and subtropical land areas (−0.7%).

Temperature and precipitation changes over tropical, subtropical, and midlatitude land areas
Temperature variability has remained nearly unchanged (slightly declined) over tropical, subtropical, and midlatitude land areas (figures 5(a), (c) and (e)). However, the bell curve of temperature over midlatitude land areas has become slightly asymmetric in the recent period with an increasingly long tail on the cold side (figure 5(e)). This indicates an increase in the number of cold extremes over midlatitude land areas in the recent period relative to the reference period, in which cold extremes in each period are defined with respect to their contemporary climatology. This finding agrees with the conclusion of Cohen et al (2014) who pointed out that continental winter temperature trends exhibit cooling over the midlatitudes since 1990.
Precipitation variability has not changed considerably over subtropical land areas (figure 5(d)). However, it has substantially increased over tropical land areas, as can be inferred from a much wider distribution of the PDF in the recent period, with standard deviations of 27 and 46 mm in the reference and recent periods, respectively ( figure 5(b)). In contrast, precipitation variability has decreased over midlatitude land areas, with standard deviations of 22.3 and 18.9 mm in the reference and recent periods, respectively (figure 5(f)). Figure 5(b) suggests that precipitation variability over tropical land areas has increased more rapidly than the mean precipitation in this region. This is similar to the results of van der Wiel and Bintanja (2021), which gives additional confidence to the obtained results in this study. The increase in precipitation variability over tropical land areas in the recent period in response to global warming has been attributed to different factors such as higher atmospheric moisture content in a warmer climate (Pendergrass et al 2017, Akinsanola et al 2020 and changes in the spatial extent and intensity of atmospheric convection (Lochbihler et al 2019).
The increase in precipitation variability over tropical land areas on the interannual timescale indicates wider swings between wet and dry conditions, indicating an increase in the risk of droughts and periods of excess water with important socioeconomic implications. In contrast, a decrease in precipitation variability over midlatitude land areas on the interannual timescale indicates narrower swings between wet and dry conditions, which decrease the risk of droughts and periods of excess water. This is consistent with the results of Kim et al (2020) who pointed out that global warming aggravates (alleviates) severe droughts in regions with warm (cold) climate.

Conclusions
Changes in the mean and variability of temperature and precipitation were examined over global land areas in the recent period (1991-2020) relative to the reference period  using the CRU dataset. Global land areas are characterized by warmer (1 • C) and slightly wetter (1.2 mm) conditions in the recent period relative to the reference period, while the variability of temperature and precipitation over global land areas has remained nearly unchanged Changes in the mean and variability of both temperature and precipitation were also analyzed over tropical, subtropical, and midlatitude land areas. The annual mean temperatures in all these three regions are characterized by an increase in the recent period relative to the reference period, with the highest increase over both subtropical and midlatitude land areas (0.7 • C), followed by tropical land areas (0.5 • C). Temperature variability has remained nearly unchanged over tropical, subtropical, and midlatitude land areas in the recent period, but temperature over midlatitude land areas has become negatively skewed in the recent period.
The annual precipitation has decreased over tropical (−15.4 mm or −1%), subtropical (−5.3 mm or −0.7%), and midlatitude (−10.8 mm or −1.2%) land areas. Precipitation variability has not changed considerably over subtropical land areas. However, it has substantially increased over tropical land areas, which indicates an increase in the risk of droughts and periods of excess water in the recent period. In contrast, precipitation variability has decreased over midlatitude land areas, indicating narrower swings between wet and dry conditions, which decrease the risk of droughts and periods of excess water in the recent period.
The identified changes in the mean and variability of temperature and precipitation might have contributed to changes in the frequency of weather and climate extremes, which need to be investigated in future studies. In particular, the relative contributions of the changes in the mean and variability of temperature and precipitation to extreme weather and climate events need to be explored in future studies. Further investigation is also required to understand the processes involved in regionally different changes in the mean and variability of temperature and precipitation.