Drivers behind the summer 2010 wave train leading to Russian heatwave and Pakistan flooding

In late July to early August 2010, two severe and concurrent extremes occurred: a devastating heatwave over western Russia and catastrophic flooding in Pakistan. The Pakistan floods resulted from several weeks of intense rainfall (up to ~300% above climatology), influenced by a mid-tropospheric trough bringing dry, cold air from mid-latitudes, high SSTs in the Arabian Sea, and La Niña conditions that enhanced moisture advection. Western Russia experienced a prolonged heatwave from late June to mid-August, characterized by a blocking anticyclone, subsidence, adiabatic warming, reduced cloudiness, and soil desiccation, which amplified heat through land–atmosphere feedbacks. The two events were dynamically linked by an unusually wavy jet stream and a circumglobal Rossby wave train, illustrating how synoptic-scale stationary waves can connect remote extremes and potentially cause concurrent hazards across the Northern Hemisphere mid-latitudes. Thermodynamic warming intensifies heatwaves and heavy rainfall, and observations show increased extreme rainfall and heat events. However, dynamic contributions, including blocking and amplified Rossby waves, are less well constrained. Some studies suggest that the mid-latitude wave pattern like that in summer 2010 has increased in frequency, though results depend on metrics and regions. Tropical SSTs and convection strongly modulate summer mid-latitude circulation via circumglobal wave trains. In 2010, La Niña-like Pacific SST anomalies co-occurred with a warm Indian Ocean and a negative Indian Ocean Dipole, implicated in circulation anomalies fostering Pakistan rainfall and possibly Russian heat. Emerging work also links high-latitude land warming and quasi-resonant amplification (QRA) of stationary waves to persistent summer extremes. Additionally, soil moisture anomalies can force Rossby waves, amplify heat extremes, and may modulate downstream rainfall. This study quantifies the relative roles of (i) SST patterns and concurrent radiative forcings, (ii) high-latitude land warming around ~65°N, and (iii) early-summer soil moisture deficits over western Russia in producing the 2010 concurrent extremes. Using a very large weather@home/climateprediction.net ensemble with a global model and a nested 50-km regional model over South Asia, we assess how these drivers alter the probability of the Russian heatwave, Pakistan flooding, and their concurrence.

Prior work has established that summer mid-latitude circulation anomalies and blocking contribute significantly to heat extremes beyond thermodynamic warming alone. Amplified quasi-stationary Rossby waves and QRA have been linked to persistent heatwaves, including 2010 in Russia. Tropical SSTs, particularly ENSO and Indian Ocean anomalies, can excite circumglobal teleconnections that influence Euro-Atlantic and Asian summer circulation. In 2010, La Niña-like Pacific SSTs and a warm Indian Ocean were implicated in circulation patterns conducive to Pakistan rainfall. Soil moisture–atmosphere feedbacks intensify heatwaves locally and can act as a Rossby wave source affecting remote regions. Climate change is projected to increase Indian monsoon rainfall, decrease extratropical summer soil moisture, and enhance high-latitude land warming—each potentially modifying waveguides, persistence of wave trains, and extreme event probabilities. However, the precise dynamical response of summer wave trains to anthropogenic forcing remains uncertain, and attribution results can be sensitive to metrics and regions considered.

Modeling framework: The study uses the weather@home (W@H) system on climateprediction.net, employing the HadAM3P atmospheric GCM (1.25° × 1.875°) forced by observed SSTs, sea ice, well-mixed GHGs, volcanic and solar forcing, and aerosols (SO2 and DMS; no black carbon). A regional model, HadRM3P (0.44° × 0.44°), is one-way nested for South Asia to better resolve mesoscale topography and precipitation. Both global (Glob2010) and regional (Reg2010) ensembles comprise 649 members each, initialized in December 2008 and run through September 2010. Variables analyzed include daily SAT (1.5 m), Z300, V300, U300, rainfall (both models), and monthly soil moisture (global only). To reduce precipitation noise, 3-day running means are applied. Climatology and ensembles: Climatologies (GlobClim and RegClim) span 1987–2015, with 170 (global) and 99 (regional) members per year, totaling 4930 and 2871 years, respectively. RegClim is driven by a different set of global simulations than those comprising GlobClim; thus, joint-event metrics combining global SAT and regional rainfall are only computed for the global model in climatology. Anomalies in 2010 are computed relative to the corresponding model climatology means to account for biases; standardized indices are constructed by removing climatological means and dividing by standard deviations. Indices and period: The period of interest is 24 July–08 August (2010), aligned with the most extreme anomalies. Indices: WRussia SAT is the spatial mean SAT over 45°–65°N, 25°–60°E; Pakistan Rainfall is the mean rainfall over 25°–40°N, 65°–85°E (western Himalayan foothills/Indus basin). Extreme events are defined as exceedances of the climatological 90th percentile of the standardized indices. Concurrent events are those exceeding both thresholds simultaneously in an ensemble member. Driver-condition selections: Two targeted sub-selections are used in the 2010 ensembles to quantify specific drivers.

• High-latitude land warming (T65N): Following Mann et al. (2018), a QRA fingerprint based on zonal-mean SAT peaks around ~65°N is correlated (lead zero) with the zonal SAT profile for each 2010 ensemble member. Members with correlation above the 90th percentile (r ≥ 0.72; rmin = -0.61, rmax = 0.90) are selected (Glob2010 | T65N; Reg2010 | T65N).
• Early-summer dry soils (soilM): For western Russia (45°–65°N, 25°–60°E), the monthly soil moisture index is computed from Glob2010. Members with June soil moisture below the 10th percentile are selected (Glob2010 soilM; Reg2010 soilM), providing a one-month lead relative to the late-July/early-August target period. Diagnostics and statistics: Probability density functions (PDFs) of standardized indices are estimated with Gaussian kernel smoothing. Exceedance probabilities of 90th-percentile thresholds and 2010-event thresholds are computed for climatology versus 2010 and for the driver-conditioned sub-selections. Joint probabilities are compared against independence (0.1 × 0.1 = 1%). Significance of PDF differences is tested using Kolmogorov–Smirnov tests (α = 0.05). Composites of SAT, rainfall, Z300, and V300 are constructed for concurrent events and for driver-conditioned selections, with stippling indicating significance from Student’s t-tests (α = 0.05). Observations: ERA-Interim SAT (2 m), Z300, V300 at 1.5° × 1.5°, and CPC/NCEP 0.25° rainfall are used for the same period, with anomalies relative to 1987–2015.

• Observed anomalies and rarity: Both the WRussia SAT and Pakistan Rainfall indices in 2010 were >3 standard deviations above their respective 1987–2015 means. Observed WRussia SAT mean: 293.1 K (20.1 °C), s.d. 1.7 K; Pakistan rainfall mean: 2.7 mm/day, s.d. 0.9 mm/day. The model climatology exhibits bias (e.g., WRussia SAT mean 296.4 K; s.d. 2.4 K; rainfall 3.50 mm/day, s.d. 0.59 mm/day), addressed by standardization.
• Recurrent wave train: A circumglobal, quasi-stationary wave-5 pattern linked the Russian heat and Pakistan rainfall, seen in observations (Z300, V300) and reproduced in the 2010 ensemble mean. The ensemble mean wave pattern resembles observations over the Atlantic–Eurasian sector, confirming the dynamical linkage.
• Increased probabilities in 2010 forcing: Relative to climatology, in the 2010 ensemble (SST + radiative forcings) the probability of exceeding the 90th percentile increases from ~10% to ~17% for WRussia SAT, and from ~10% to ~34% (global) and ~43% (regional) for Pakistan rainfall—approximately a factor 2 for heat and 3–4 for rainfall. Concurrent event probability increases ~4-fold from ~2% in climatology to ~8–9% in 2010. Even so, the model rarely reproduces the observed extreme Russian SAT 2010 threshold; its return time is >500 years (>0.2% occurrence) in the ensemble for the 2-week window considered.
• Threshold exceedance relative to 2010 extremes: The fraction of members exceeding the observed 2010 Pakistan rainfall threshold rises from ~0.2% (climatology) to ~0.8% (Glob2010) and ~1.4% (Reg2010), a 4–7-fold increase.
• Effect of high-latitude land warming (T65N): Conditioning on high-latitude land warming (top 10% of the T65N metric) further elevates extremes: WRussia SAT extreme probability increases from ~17% (2010) to ~26% (Glob2010 | T65N), and Pakistan rainfall from ~43% (Reg2010) to ~54% (Reg2010 | T65N). Concurrent event probability rises to ~12.3% in Glob2010 | T65N (a 30–50% increase over the full 2010 ensemble).
• Effect of early-summer dry soils (soilM): Conditioning on June soil moisture below the 10th percentile over western Russia raises WRussia SAT extremes from ~17% to ~25% and Pakistan rainfall extremes from ~43% to ~48% (regional). Concurrent event probability increases to ~10.8% in Glob2010 | soilM.
• Climatological dependence: Even in the model climatology, the concurrent extreme occurs at ~2%, twice the independent expectation (1%), indicating a baseline dynamical linkage via recurrent wave trains.
• Mechanistic interpretation: Dry soils enhance local heating and blocking over western Russia, strengthening the wave train and downstream cold-air advection toward Pakistan, which supports extreme rainfall with about one-month lead time. High-latitude land warming likely favors waveguide states, persistence, and amplification of quasi-stationary waves, compounding SST-forced teleconnections. All reported PDF pair differences (2010 vs. climatology; driver-conditioned vs. unconditioned 2010) are significant at α = 0.05 via Kolmogorov–Smirnov tests.

The study demonstrates that the 2010 Russian heatwave and Pakistan floods were dynamically linked by a recurrent circumglobal Rossby wave train, rendering their concurrence substantially more likely than if they were independent. The weather@home large ensemble reproduces key circulation features (wave-5 in Z300/V300) seen in observations. Three drivers jointly elevate risk: (i) 2010 SST anomalies and radiative forcings substantially increase probabilities of both extremes and their concurrence; (ii) early-summer soil moisture deficits over western Russia intensify local heat and, through a strengthened wave train, enhance extreme rainfall downstream over Pakistan; and (iii) high-latitude land warming favors the occurrence and persistence of wave trains, further increasing both heat and rainfall extremes. These results reconcile differing emphases in previous studies by quantifying the synergistic interactions among SST forcing, land–atmosphere feedbacks, and high-latitude warming. While SST anomalies primarily set up the circumglobal wave response and thermodynamic moisture availability, land processes modulate the amplitude and persistence of the wave train. The enhanced joint probability in climatology (relative to independence) underscores that such wave patterns are intrinsic features of boreal-summer dynamics, onto which external forcings and land feedbacks superimpose. However, separating direct GHG versus SST-pattern effects is not possible in this design, and model biases and internal variability still influence realized extremes and phase positioning of waves. Further work is needed to clarify causal pathways between high-latitude land warming and wave dynamics and to attribute persistence versus amplitude contributions. Implications include elevated compound risk across distant breadbasket and vulnerable regions under background states with La Niña-like SSTs, high-latitude warming, and soil drying, which may become more frequent in a warming climate.

Using a very large global–regional ensemble, the study reproduces the 2010 atmospheric wave train linking the Russian heatwave and Pakistan floods and quantifies three key drivers that increased their likelihood: (1) 2010 SST anomalies and radiative forcings elevated heat and rainfall extreme probabilities (≈2× for heat, 3–4× for rainfall, and ≈4× for concurrence); (2) early-summer soil moisture deficits over western Russia further increased both local heat extremes and remote Pakistan rainfall extremes by strengthening the wave response; and (3) high-latitude land warming further favored wave-train occurrence, boosting both extremes and concurrent events (to ≈12%). These findings highlight complex, synergistic land–ocean–atmosphere interactions that modulate circumglobal wave trains and compound extremes. Future research should (i) disentangle the causal roles of high-latitude land warming in wave amplification versus persistence; (ii) better separate GHG-driven mean-state changes from SST-forced variability; (iii) assess how projected increases in high-latitude warming, soil drying, and potential changes in ENSO/Indian Ocean states may alter the frequency, intensity, and concurrence of such wave-train-linked extremes; and (iv) improve regional process representation (e.g., lower-tropospheric dynamics, convection, and orography) to refine risk estimates for compound events.

• Forcings not separable: The experimental design forces the 2010 ensembles with observed SSTs and radiative agents together, preventing a clean separation of direct GHG versus SST-pattern effects.
• Model biases: HadAM3P exhibits a warm bias in mid-latitudes; though standardized indices mitigate biases, the ensemble underestimates the extremity of the observed WRussia SAT 2010 threshold (return time >500 years in the model for the 2-week window).
• Ensemble nesting mismatch: RegClim is driven by different global simulations than those used for GlobClim, limiting direct joint-event calculations between global SAT and regional rainfall in the climatological baseline.
• Soil moisture temporal resolution: Soil moisture is monthly, constraining lead–lag and process diagnostics; causality between soil drying, blocking, and downstream rainfall remains partly inferential.
• Process diagnostics: Lower-tropospheric wind fields critical for monsoon–extratropical interactions were not explicitly analyzed here; orographic convection processes are better represented in the regional model but still uncertain.
• Phase uncertainty: Ensemble means smooth internal variability; phase positioning and amplitude of wave patterns vary across realizations, and some differences from observations remain (e.g., location and strength of SAT anomalies).