Extreme weather events have become a dominant feature of the narrative surrounding changes in global climate. with large impacts on ecosystem stability, functioning and resilience, however, understanding of their risk of co-occurrence at the regional scale is lacking. Based on the UK Met Office’s long-term temperature and rainfall records, we present the first evidence demonstrating significant increases in the magnitude, direction of change and spatial co-localization of extreme weather events since 1961. Combining this new understanding with land use datasets allowed us to assess the likely consequences on future agricultural production and conservation priority areas. All land uses are impacted by the increasing risk of at least one extreme event and conservation areas were identified as hotspots of risk for the co-occurrence of multiple event types. Our findings provide a basis to regionally guide land use optimisation, land management practices and regulatory actions preserving ecosystem services against multiple climate threats.
Recent large flood and drought events have received global media attention. For example, unprecedented winter rainfall across the UK in 2013/14 resulted in extreme flooding and storm surges with large areas of agricultural land under water for more than 80 days (ADAS, 2014), while over 60 % of the state of California’s land area was under varying severity of drought from 2011 to 2017 (U.S. Government, 2018). Flooding and drought can have large economic impacts; the World Economic Forum has rated extreme weather events as the most significant risk facing humanity (World Economic Forum, 2018). Losses to the UK agricultural sector of £180 million were reported as a result of the 1995 drought and associated heatwave (Environment Agency, 2006), while the 2013/14 flood led to losses of over £20 million (ADAS, 2014). Similarly, the total economic impact of the European heatwave in 2013 was estimated at 11 billion Euros (Environment Agency, 2006), while extreme snow was estimated to cost the US economy up to $3 billion in 2016 (Lee and Lee, 2019). Natural ecosystems are also vulnerable, for example, record heat and dry conditions in 2010/2011 led to a sudden collapse of large areas of Australian eucalypt forest previously considered to be resilient to drought (Matusick et al., 2013). Furthermore, the hot and dry conditions of 2018-19 in the UK resulted in unprecedented wildfires in the globally rare moorland habitat with 135 individual fires burning 29,334 ha of land (E.F.F.I.S., 2019). In 2019, hot and dry conditions in Australia resulted in the generation of mega-fires with unprecedented sizes and number of fires covering at least 3.8 million ha of temperate forest (Nolan et al. 2020)
While there is a wealth of evidence that temperatures are increasing, the pattern for rainfall is uncertain (IPCC, 2013) but predicted to become temporally uneven with the majority of annual precipitation totals occurring in a small number of intense events (Pendergrass and Knutti, 2018). For many regions of the UK, climate models and historical observations indicate that the frequency, intensity (Thompson et al. 2017; Kendon et al. 2014) and duration (Roudier et al. 2016) of winter rainfall has increased, along with the incidence and intensity of short burst summer downpours (ADAS, 2014) and the kinetic energy of autumn rainfall (Alfieri et al., 2015). Models also predict an increase in the frequency of short-term droughts of three to six months in duration (Burke et al. 2010). These all have implications for agriculture, conservation and human health.
To date, the majority of studies investigating the risk of extreme weather events have focused on the global or national scale, and often only on a single event type. There is greater uncertainty in changes at the regional scale where the immediate impacts will be felt (Marotzke et al. 2017). Spatial variation in weather patterns can be large and analysis at the national scale masks regional differences in the risk of occurrence and the expected event type (Zampieri et al., 2017). Furthermore, extreme events might not occur in isolation and there are an increasing number of examples of direct transitions from one extreme weather regime to another (e.g. flood to drought or vice versa ) (Mahony and Cannon, 2018; Swain et al. 2018; Loecke et al., 2017). In the UK, heavy spring rainfall in 2012 led to 78 days of flooding, while 98 days of official drought were declared the following summer which the media dubbed ‘the wettest drought on record’ (Channel 4 News, 2012). In 2019 there were 5,600 flood warnings across England while groundwater reserves were depleted in 25 areas (The Guardian, 2019)( Such events have highlighted the need for stakeholders, including farmers, water companies, forestry and environmental protection and conservation bodies to prepare for the possibility of both flooding and drought within the same year. So called ‘compound events’ have been identified by the World Climate Research Program as a research priority (Alexander et al. 2016) and are likely to have disproportionately severe impacts on ecosystems, potentially tipping ecosystem functions into new trajectories (Johnstone et al. 2016).
To safeguard vulnerable ecosystems and the services they provide, adaption in management may be required. However, the specific strategy employed will vary depending on the event type. For example, the re-introduction of grazing livestock to moorland could reduce fire risk during dry, hot summers but could also increase the risk of compaction during wet periods increasing subsequent flood risk. Similarly, planting trees to sequester carbon may increase fire risk under dry conditions leading to a potential reduction in air quality, water quality and human health if planted in the wrong place (Lui et al. 2017).
To advise stakeholders and guide policy we need to understand the regional risk posed by different (single and multiple) extreme events and identify where they might impact delivery of ecosystem services (e.g. food security, biodiversity, carbon storage) by different land use types. In this study, we utilised the historical UK weather record held by the UK Met Office National Climate Information Centre to examine, for the first time, the change in frequency and distribution of, and interaction between, indicators of four weather extremes; extreme heat, extreme cold, high rainfall and low rainfall, based on thresholds indicative of heatwaves, cold snaps, floods and droughts, between two time periods 1961-1988 and 1989-2016. We integrated the results from this analysis with national land cover data to identify extreme weather hotspots in relation to ecosystem type and their ability to deliver different ecosystem services.
These datasets were statistically interrogated to answer four key questions: (1) Has the frequency of extreme events in the UK increased between the two time periods? (2) Are there hotspots where the annual risk of occurrence for two or more event types has increased? (3) Are there areas of the UK where the probability of occurrence of two or more types of event within the same year has increased? and (4) Are some vulnerable ecosystems more exposed to changes in risk of increased numbers of events occurring than others?
Through this analysis, we provide evidence for the perceived increase in the frequency of extreme events across the UK. To date, most studies of this nature have focused on the incidence, or impact, at the national scale. Our results show strong regional variation in the direction and magnitude of change enabling the production of national risk maps which can be used by stakeholders to guide land management and policy that promotes adaptation to protect the delivery of ecosystem services from different land uses.
Our analysis shows that between the two 28-year periods of high resolution meteorological records there has been a notable change in the frequency of threshold exceedance across the UK with strong regional response patterns (Fig. 1). Temperature metrics showed the largest and most widespread response but the direction of change varied. For extreme heat events, there was a significant increase in the mean number of events during the last 28 years, with the south east of England experiencing the largest change, corresponding to on average 1.87 additional events each year. Significant increases (0.68–1.36) in the mean number of extreme events also occurred across most of England, except the north west and across the east of Northern Ireland, and the far north of Scotland. Concurrently, the frequency of extreme cold events decreased across all regions except for much of Wales and small regions of south west England and northern Scotland. The magnitude of change was greater than that for heat extremes, ranging from 1–2.3 fewer events each year. Response patterns in rainfall extremes were weaker than for temperature; this is consistent with the large body of research showing mixed results for predicted changes in rainfall patterns across the globe (Alfieri et al. 2015). However, the results show a significant increase in wet extremes ranging from 1.0 - 1.6 additional events each year in western Scotland to 0.8 - 1.0 additional events in the Welsh border region, along parts of the south coast of England and East Anglia, and in western Northern Ireland. The change in extreme dry events was small with no significant increase overall and a decrease of 0.9 events in the far north for Scotland. However, a strong spatial pattern in extreme dry events did emerge, reflecting the changes in heat events with an increase of up to 0.5 events in south east England.
These changes in threshold exceedances for temperature and rainfall provide statistical evidence underpinning the perceived increase in UK heatwaves, floods and droughts over the past decade and provide insight into which regions are most at risk. While the changes in temperature drivers relate directly to heat waves or cold snaps, the use of precipitation as a proxy for flood or drought events is less robust. However, an increase in extremely wet periods in Scotland, parts of southern England and Wales and Northern Ireland will heighten flood risk. Furthermore, runoff extremes have been shown to increase more quickly than precipitation extremes in a warming climate, and increases in rainfall are likely to underestimate the risk of flash flood events (Yin et al., 2018). These results corroborate the recent analysis of observed river discharge trends between 1960 and 2010 which found the largest increase in flood discharge in these areas (Yin et al., 2018). Similarly, drought risk is a function of both rainfall and temperature with prolonged high temperatures exacerbating soil dryness and providing feedback loops further reducing rainfall, increasing surface temperatures and promoting fire risk(Teuling, 2018). Seasonal analysis of changes in extreme dry events revealed that the greatest change occurs during spring (Fig. S1) when new season growth begins, and is a vital period for sufficient soil moisture supply for agricultural crops. Spring drought has been shown to be more detrimental to plant production compared to summer drought conditions across a range of ecosystems (Song et al., 2019). Increases in dry spring events may be exacerbated by a spatially coupled increase in the number of periods of suitable winter growing conditions utilising water reserves built up during preceding wetter seasons (Fig. 2). Whilst not statistically significant (atp < 0.05), the indicative combination of i) increased dry events with ii) an increase in heat events, and iii) increased winter growing periods, points towards a heightened drought risk in the future, especially in the south east of England where these metrics showed the greatest increase. Furthermore, the probability that a heat event and a dry event will occur within the same year was high and ranged from 0.80 to 0.98 in this area (Fig. S2). Although the evidence for increased extreme dry events, from this analysis is weak, it corroborates recent modelling indicating high drought vulnerability in East of England based on reported historical agricultural impacts (Parsons et al., 2019)[30]
The environmental impact of this increased frequency in extreme events depends on the land use and the biodiversity and ecosystem services it is expected to deliver. The response may vary, in magnitude and direction, based on the type of ecosystem and the dominant services it provides (Table 1, Table S1). We grouped the UK land cover categories (Rowland et al. 2015) into four broad classes each providing specific ecosystem services and levels of biodiversity: (1) Agriculture, incorporating arable/horticultural and improved grasslands (provisioning), (2) Woodlands, incorporating broadleaf and coniferous woodlands (provisioning, regulating and biodiversity), (3) Conservation, incorporating National Parks and Sites of Special Scientific Interest (SSSIs) (supporting regulating and biodiversity), (4) Carbon stores, incorporating heathland, heath grasslands and bogs (regulating).
The reduction in frequency of cold events (i.e. less frosts and snow) shows an impact across all ecosystem types, ranging from 64% of all the land in SSSIs to >80% of the total area under arable land use, respectively. Simplistically, it might be assumed that a reduction in winter cold events would be beneficial. However, many plants rely on low winter temperatures for vernalisation and warmer winters can cause increased pest and disease risk, loss of cold acclimation, asynchronicity of biological lifecycles and increased runoff (Table 1).
Agricultural systems and broadleaf forests represented the largest proportion of the total land area at increased risk of extreme heat events and the arable sector in particular appears to the most affected with 83% of the total area at risk (Fig. 3a). This reflects the large dominance of arable land use in the East of England. Furthermore, recent research suggests that heat extremes have a larger impact on grain yields than extremes in precipitation, highlighting the risk to arable systems (Vogel et al., 2019) and, hot dry spells can influence agricultural water use, especially under cropping. In the period between 2000 and 2017, the highest 2 years for abstraction for the purpose of spray irrigation correspond with the lowest 2 years of annual levels of rainfall (DEFRA, 2019a). Temperature extremes also dominated in improved grasslands, with 56% of the total area exposed to increase risk of extreme heat which directly impacts on livestock production. However, the proportion of grassland exposed to increases in extreme rainfall, and therefore flooding, was greater than in arable systems. Soil carbon (C) stores and coniferous forests appear to be most at risk of extreme rain and flooding, with increased frequency of events occurring across 35–55% of the total area. Forests are commonly proposed as mitigation strategies to reduce flood risk through interception of rainfall and increased soil infiltration (Stratford et al., 2017). However, extreme rainfall events often override this increased infiltration capacity and the potential to reduce the severity of major floods is limited. When flooding does occur, the impact can be severe in commercial forestry operations with largescale erosion and damage downstream from woody debris. For soil C stores, reduced extreme cold and extreme rainfall present the largest risk. This has large implications for the C cycle and is likely to increase the release of soil C and decrease sequestration through and increase in wet-drying cycles and microbial respiration and increased erosion losses (Petrakis et al., 2017; Reichstein et al., 2013; Kim et al., 2012; Schimel et al., 2007) (Table 1). Our analysis also indicates that large expanses of upland bog or lowland fen peat are located in regions experiencing higher temperatures, droughts and therefore potential fire risk. These events threaten to exacerbate greenhouse gas emissions and destabilization of terrestrial C stores.
Specific regions of the UK show a significant increase in frequency of more than one extreme event type (Fig. 4). Risk hotspots, with significant increased frequency of three threshold exceedances are identified in the south coast of England, areas in the Welsh borders and the north east of England, highlighting areas most at risk of unexpected ecosystem response and largescale impacts on function (Table 1). Land of high nature value appears to be at most risk of multiple extreme event types with all three stress indicators increasing in frequency in 24 and 21% of the total area covered by National Parks and SSSIs (Fig. 3b). Due to the importance of these sites as niche habitats for rare or endangered species this could have severe impacts on biodiversity and genetic resources. This was seen following the 1995 UK drought which led to a shift in butterfly communities from vulnerable specialised species to widespread generalist species (De Palma et al., 2017).
Exposure to an extreme event can make ecosystems more susceptible to a subsequent stress, magnifying impacts (Hohner et al., 2019; Ksudhal et al., 2018; Mazdiyasni and AghaKouchak, 2015; Zscheischler et al., 2014) with the potential to decrease the threshold by which climatic metrics, such as precipitation amount, generate an extreme event, such as flood or drought (DEFRA, 2019). Our results show that the overall UK mean increase in the probability of all four event types occurring with the same year low at 0.275. However, the impact on ecosystem function would likely be extreme. The increase in frequency of extreme heat events was the dominant driver of the response pattern, with the highest probabilities in the south east of the UK and the lowest probabilities in Scotland, Wales, Northern Ireland and north-west England (Table S2; Fig. S2).
To illustrate the impact on agriculture, we have taken the UK arable sector as a case study since the combination of adverse weather conditions can magnify the impacts on agricultural production. In particular, the combination of extreme wet spells and extreme dry spells within the same year has been shown to be particularly detrimental for crops. In 2017, there was an 8.3%, 17% and 19% reduction in income in England from key three crops, wheat, sugar beet and potatoes respectively. This was attributed, in part, due to reduced yields caused by wet spring conditions, hot dry summer and heavy autumn rains during harvest[45]. Reductions in yields reduced the export value of wheat by 73% and 84% in 2017 and 2018 respectively, and increased the import expenditure by 38% and 79% respectively (DEFRA, 2018). The majority of the UK’s arable and horticultural land area is in the East of England, with 28% of total wheat production and 62% of sugar beet production located in the South East, and East Anglia accounting for one third of England’s potato crop (DEFRA, 2018). The probability that extreme hot, dry and wet events will occur within the same year is highest for this region of the country and ranges from 0.69–0.99 (Fig. S2) highlighting the vulnerability of this sector to future climatic risk.
Globally, societies are facing unprecedented and complex threats to food and water security, infrastructure and well-being due to climate change. The increased frequency of multiple extreme events across different land uses identified by our analysis is likely to have detrimental impacts on the ecosystem service provision. While some benefits to service provision have been identified, these are likely to be out-weighed by the negative impacts (Table 1). In May 2019, the UK government declared a state of climate emergency that was swiftly followed by Ireland, France and Canada. Furthermore, large-scale land use change has been identified as a strategy for the UK to meet its emission reductions in the Paris Agreement (Committee on Climate Change, 2018), and its recent target of net zero emissions by 2050. The evidence herein provides vital information on the vulnerability of different areas and economic sectors to climate extremes and should be used by UK policy makers, farm advisers and environmental agencies to develop adaption strategies and land use change policy tailored to the specific extreme event threat, based on location and ecosystem type. This research highlights the importance of considering the change in exposure of land to (combinations of) extreme weather at the regional scale and adoption of a similar approach in other countries could inform the safeguarding of the vital ecosystem services on which society depends.