Past and Projected Weather Pattern Persistence with Associated Multi-Hazards in the British Isles

Hazards such as heatwaves, droughts and floods are often associated with persistent weather patterns. Atmosphere-Ocean General Circulation Models (AOGCMs) are important tools for evaluating projected changes in extreme weather. Here, we demonstrate that 2-day weather pattern persistence, derived from the Lamb Weather Types (LWTs) objective scheme, is a useful concept for both investigating climate risks from multi-hazard events as well as for assessing AOGCM realism. This study evaluates the ability of a Coupled Model Intercomparison Project Phase 5 (CMIP5) multi-model sub-ensemble of 10 AOGCMs at reproducing seasonal LWTs persistence and frequencies over the British Isles (BI). Changes in persistence are investigated under two Representative Concentration Pathways (RCP8.5 and RCP4.5) up to 2100. The ensemble broadly replicates historical LWTs persistence observed in reanalyses (1971–2000). Future persistence and frequency of summer anticyclonic LWT are found to increase, implying heightened risk of drought and heatwaves. On the other hand, the cyclonic LWT decreases in autumn suggesting reduced likelihood of flooding and severe gales. During winter, AOGCMs point to increased risk of concurrent fluvial flooding-wind hazards by 2100, however, they also tend to over-estimate such risks when compared to reanalyses. In summer, the strength of the nocturnal Urban Heat Island (UHI) of London could intensify, enhancing the likelihood of combined heatwave-poor air quality events. Further research is needed to explore other multi-hazards in relation to changing weather pattern persistence and how best to communicate such threats to vulnerable communities.


165
2. Lamb pure directional weather types (e.g. N, S, or E-types) correspond to an essentially 166 straight flow, when |Z| is less than F (Eq. 6);

174
5. An unclassified (U) type is obtained when F and |Z| are less than 6, with the choice of 6 175 depending on grid spacing, meaning that if using a grid resolution finer than 5° by 10° latitude-

181
Numbers refer to those points used in Equations 1 to 5.

183 184
The objective classification scheme yields 27 LWTs comprised of two synoptic types 185 (anticyclonic A and cyclonic C), five purely directional types (westerly W, north-westerly NW, 186 easterly E, northerly N, and southerly S) , 19 hybrid combinations of synoptic and directional types 187 (e.g. CNW, CSE and AE), and 1 unclassified (U) type (

256
To detect both linear and non-linear annual changes in the total seasonal counts of LWTs MMEM where represents the single AOGCM, 20CR, NCEP and Lamb's subjective datasets within the 282 relative time periods of 1980s, 2020s, 2050s, 2080s and is the given LWT considered from the 10 283 types mentioned above. The higher the F-Score, the greater the likelihood of concurrent multi-basin 284 fluvial flooding and wind hazards within winter, over the specified time horizon and RCP scenario.

286
As a proxy for combined heatwave and poor air quality hazards occurring during summer, we

320
Figure 2 provides a synthesis of the data and methodological framework.    (Table S1).

360 361
Results for RCP4.5 show similar changes in persistence compared to RCP8.5, although they are 362 smaller ( Figure S1). In particular, the C-type is found to change significantly (p<0.1) only in summer 363 by the 2080s; the E-type in winter by the 2080s; the N-type only in spring by the 2080s; and the S-type 364 in summer by the 2050s and spring also by the 2020s (Table S2).

366
3.2 Persistence of weather patterns (by model) 367 Figure 4 shows persistence for the same future periods but for each AOGCM in the MME 368 compared with the reanalyses and Lamb's catalogue, for impactful weather types and seasons.
12 type MMEM persistence increases during summer (Figures 3a and 4a); C-type persistence decreases 372 in all seasons, most markedly in summer and autumn (Figures 3 and 4b); W-type persistence does 373 not change in winter but increases in autumn and decreases in spring (Figures 3b-d and 4c).

375
Amongst the other weather types, we note only a decrease in C-and E-types during summer, an 376 increase in N-type in spring, and S-type persistence decreases in all seasons (Figures 3 and 4d). The

395 396
Analysis of RCP4.5 output shows similar, though less marked, results when compared to RCP8.5

397
( Figure S2). Under the lower emission scenario, we find that most AOGCMs project persistence that 398 falls outside the 95% confidence intervals of the 1980s. A-type MMEM persistence in summer could 399 increase slightly, in particular during the 2080s (Figures S1a-S2a), C-type in autumn may decrease 13 ( Figures S1b-S2b), W-type during winter is projected to remain stable across the three future periods 401 (Figures S1c-S2c) and S-type persistence in spring decreases by 2100 ( Figures S1d-S2d). The C-type 402 in summer and A-type in autumn exhibit decreased persistence, whereas the E-type shows a marked 403 increase in persistence during winter; findings that differ from RCP8.5 ( Figure S1).

418
increase in A-type during spring, which all reflect the changes in persistence (Figure 3a and 3d).  Figure S3 and Table S3. Results for RCP4.5 reflect the scenarios of RCP8.5 although the Sen's slopes 432 are less extreme and statistically significant. The A-type frequency is projected to increase 433 significantly (p<0.01, Figure S3a and Table S3) during summer, C-type in autumn is set to decrease 434 (p<0.05, Figure S3b), W-type frequency in winter shows no significant trend ( Figure S3c), and the S-435 type during spring decreases significantly (p<0.05, Figure S3d). As per RCP8.5, we also observe (not 436 shown) a significant decrease in C-type frequencies during summer (p<0.01) and spring (p<0.05) and 437 an increase in the A-type during spring (p<0.05), matching the relative changes in persistence ( Figure   438 S1a and S1d). northward, enhancing such effects [1][2][3]90,91,[93][94][95] Figure   580 S1: As per Figure 3 but for RCP4.5, Figure S2: As per Figure 4 but for RCP4.5, Figure S3: As per Figure 5 but for 581 RCP4.5, Figure S4: As per Figure 6 but for RCP4.5, Figure S5: As per Figure 7 but for RCP4.5. Supplementary   582 Tables, Table S1: MME statistical significance of LWTs persistence for RCP8.5, Table S2: The same as Table S1 but 583 for RCP4.5,  synoptic weather on UK surface ozone and implications for premature mortality. Environ