The unprecedented Pacific Northwest heat-wave of June 2021

An unprecedented heatwave struck the Pacific Northwest (PNW) from 25 June to 2 July 2021 across British Columbia and Alberta in Canada and Washington and Oregon in the United States. Near-surface air temperature anomalies reached 16–20 °C above normal, and many locations broke all-time maximum temperature records by more than 5 °C. The Canadian national temperature record was broken three days in a row, peaking at 49.6 °C in Lytton, BC, on 29 June—reportedly the hottest temperature recorded north of 45°N and hotter than any recorded in Europe or South America. The exceedance of prior records surpassed notorious events such as the 2003 European and 2010 Russian heatwaves, though the PNW event was shorter in duration. The purpose of this paper is to comprehensively document the meteorology, prediction, climate context, and multi-sector impacts (human health, marine ecosystems, agriculture, cryosphere, hydrology, wildfires, and geomorphic hazards) of the heatwave, and to discuss lessons for preparedness in a warming climate.

The paper situates the PNW heatwave within the context of prior extreme heat events, notably the 2003 European and 2010 Russian heatwaves, which caused tens of thousands of deaths. Prior studies of the June 2021 event highlight the role of atmospheric blocking and diabatic heating in its development, and identify contributions from sensible heating, mixing, subsidence, and soil moisture feedbacks. The authors review subseasonal forecast performance literature indicating skill for heatwaves at lead times of about 10–15 days. For climate change attribution and dynamics, the paper discusses ongoing debates about Arctic amplification’s influence on midlatitude circulation and wave amplification, with mixed evidence in the literature. It also reviews proposed mechanisms by which climate change could affect heatwave dynamics, including soil moisture feedbacks, tropical convection influences, Rossby wave propagation, and latent heating impacts on blocking. Comparisons to other studies suggest increased atmospheric water vapor in a warmer climate could strengthen blocking highs and associated surface temperature anomalies. The authors note rapid attribution studies that estimate large anthropogenic contributions to the severity and altered return period of the event, albeit with high uncertainty for such rare extremes.

The study employs multiple methodological components: - Temperature analyses: Daily maximum 2 m temperatures (TX) from ERA5 reanalysis (1° grid) and weather stations (including Lytton, BC). TX anomalies computed relative to 1981–2020 daily climatology; maximum 3-day running mean anomalies identified for 22 June–3 July 2021. Record exceedances calculated as TXx during the event relative to maximum TXx over 1950–prior to event dates; Canadian station data from the Long-Term Climate Extremes (LTCE) database included. - Synoptic and trajectory analysis: ERA5 fields (geopotential height, winds, precipitable water, RH, temperature, sea-level pressure) analyzed to characterize evolution of the blocking high. Four-day backward trajectories (terminating at 500 m AGL, 01:00 UTC 28 June) computed using NOAA HYSPLIT with GFS 0.25° forecast data for 63 parcels. Diabatic vs adiabatic temperature changes quantified via changes in potential temperature and pressure along trajectories using thermodynamic relations; mean diabatic and adiabatic contributions estimated. - Forecast evaluation: Operational medium-range forecasts from the North American Ensemble Forecast System (NAEFS; bias-corrected and probabilistically calibrated), deterministic GDPS and GFS v16 examined for signals from 18–25 June. Subseasonal forecasts from ECMWF IFS (S2S database) assessed probabilistically for 2 m temperature anomalies and blocking frequency exceeding the 95th percentile for 25 June–1 July, using reforecast-based climatologies and a modified blocking detection algorithm (persistent positive z500 anomalies and meridional gradient reversal; thresholds normalized by latitude; persistence ≥5 days). - Climate change context: Regional temperature trends computed using Berkeley Earth anomalies for 45–52°N, 119–123°W, 1875–2020 (annual, JJA mean, and JJA TX). Atmospheric waviness metrics—anti-cyclonic Local Finite-Amplitude Wave Activity (a-LWA) from z500 and a recurrent Rossby-wave index (R) from v250—applied to PAMIP experiments (IPSL-CM6A-LR, CanESM5) to test impacts of sea-ice loss and SST changes on North American waviness/blocking. - Impacts data and analyses: • Human health: Excess and heat-attributable deaths from provincial/state coroner and health departments; non-fatal emergency department visits from US HHS Region 10; contextual factors (e.g., COVID-19 pandemic). • Marine life: Intertidal thermal imaging (FLIR E40) on 28 June; mortality surveys for mussels and barnacles using randomized 100 cm² quadrats, scaling to shoreline extents at representative sites; qualitative surveys across the Salish Sea. • Wildfires: Fire Weather Index (FWI) derived from WRFv4.1.2 mesoscale NWP runs (nested 36/12/4 km) with NAM initialization; BlueSky smoke modeling for PM2.5; satellite hotspots as fire proxies; CIFFC situational reports for fire counts/area burned; analysis of pyrocumulonimbus (CbFg) lightning. • Agriculture: Statistics Canada crop yields (field, fruit, vegetable) used to estimate linear yield trends (to 2020) and predict 2021 yields; anomalies computed relative to trends and expressed in units of detrended interannual standard deviation. NDVI (CCAP) at weekly resolution used to assess within-heatwave vegetation greenness changes by agricultural region. • Glacier and snow melt: Streamflow from Environment and Climate Change Canada hydrometric stations compared to 1979–2020 medians and variability; ERA5-Land snow water equivalent; Randolph Glacier Inventory for basin glacier coverage; identification of daily and all-time streamflow records. • Landslides: Post-wildfire debris-flow mapping from field observations and Sentinel imagery; burn severity via dNBR and BARC classification; field validation of soil burn severity and hydrophobicity; documentation of infrastructure impacts.

- Severity and records: Near-surface air temperature anomalies reached 16–20 °C above normal over a broad area. Many locations exceeded prior all-time maxima by >5 °C. Canada’s national temperature record was broken three consecutive days, peaking at 49.6 °C at Lytton, BC (4.6 °C above the previous national record), reportedly the highest temperature recorded north of 45°N and higher than any recorded in Europe or South America. Compared to the 2003 Europe and 2010 Russia heatwaves, record exceedances were larger though duration was shorter. - Meteorological drivers: A strong blocking high developed downstream of Pacific frontal zones. Back-trajectory analysis showed low-level air mass warming was dominated by diabatic processes (~78% or ~14 K) versus adiabatic subsidence (~22% or ~4 K) over four days. Upstream condensational heating within frontal clouds and orographic clouds (Alaska Panhandle) and diabatic heating over land under clear skies near the summer solstice were key contributors. Local factors (arid canyon setting at Lytton, limited nocturnal cooling, coastal outflow bringing hot interior air and adiabatic warming) amplified extremes. - Forecast performance: Medium-range ensemble forecasts (NAEFS) first indicated a heatwave on 18 June (8 days before onset; 12 days before peak). By 21 June, ensemble signals indicated an extreme, with high confidence of exceeding the 95th percentile for several days. By 25 June, forecasts captured magnitude and longevity well, though peaks were underpredicted by ~1–3 °C. Subseasonal ECMWF forecasts initialized from 10–17 June showed increased probabilities of extremes, with up to ~70% of members predicting extreme temperatures by 14 June somewhere in the affected region; extreme blocking probabilities increased as early as 7 June, though with location errors shifting the block too far west. - Climate change context: The regional summer background climate was ~1.0 °C warmer than late 19th century conditions. While dynamic attribution remains uncertain, thermodynamic warming clearly increased event severity. Two PAMIP models analyzed showed no robust, North America–wide changes in waviness/blocking attributable to sea-ice loss or SST changes in this framework; this negative result is not conclusive and further comprehensive analyses are required. - Human health impacts: Preliminary attribution totals indicate at least 868 heat-related deaths across the PNW: 619 in BC (93% between 25 June–1 July), ~66 in Alberta, 100 in Washington, and 83 in Oregon. Most deaths occurred in private residences; higher risks in neighborhoods with higher material/social deprivation and lower green space, among ages 65–84, and females; mental illness and substance use disorder were significant risk factors. In US HHS Region 10, heat-related ED visits from 25–30 June were 69 times higher than in equivalent days of 2019, with disproportionate impacts on males and those ≥75 years. COVID-19–related social isolation likely increased vulnerability and reduced use of cooling centers. - Marine ecosystem impacts: Intertidal surface temperatures exceeded 50 °C during low tide on 28 June. High mortality was observed across sessile and mobile taxa. At one mussel-dominated site, bay mussel mortality exceeded 70%, with ~1.2 million mussels estimated dead along a 100 m shoreline. At a barnacle-dominated site, mortality exceeded 70%, implying ~10 million barnacles killed along a 100 m stretch. Overall, mortalities of intertidal invertebrates across the Salish Sea were almost certainly in the billions. - Wildfires: Extreme heat/dryness drastically increased fire danger. BC went from 6 active wildfires (123.5 ha burned) on 20 June to 175 wildfires (78,939 ha) by 3 July. Lytton’s observed FWI peaked at 132 (typical 0–30). Pyrocumulonimbus (CbFg) formed over major fires (Sparks Lake, McKay Creek) on 29–30 June; approximately 120,800 cloud-to-ground lightning strikes occurred on 30 June, with at least 127 new lightning-ignited wildfires 30 June–2 July. National preparedness level reached 5 by 11 July. - Agriculture: Of 26 BC field/fruit/vegetable crops analyzed, 24 had 2021 yields below trend-predicted values. In BC, estimated declines relative to predicted yields: spring wheat −31%, barley −30%, canola −21%, oats −5% (smaller decline consistent with cooler Peace River region). In Alberta, field crop yields were the lowest since 2002 drought: canola −32%, spring wheat −31%, barley −32%, oats −29%, with anomalies >3σ. Fruit declines >20% for sweet cherries, grapes, plums, raspberries; >10% for apples, nectarines, peaches, pears; cranberries −2% vs predicted (and +10% vs 2011–2020 average due to strong historical trend). Several vegetables dropped >20% (pumpkins, tomatoes, radishes) and others >10% (Brussels sprouts, lettuce, green peas, squash/zucchini). Weekly cropland NDVI decreased during the heatwave in 6 of 8 BC agricultural regions, strongest in Cariboo, Kootenay, and Thompson-Okanagan. The livestock sector experienced at least 651,000 farm animal mortalities (24–30 June). - Hydrology and cryosphere: Rapid snow and glacier melt increased streamflows in snow/ice-bearing basins, with many daily records and some all-time records, prompting flood warnings and evacuations (e.g., Pemberton Valley). In some basins, flows dropped during the event due to snowpack depletion despite persistent heat. Extensive snowmelt exposed darker glacier ice, and sustained warm July conditions led to large seasonal glacier mass loss. Glaciated basins maintained near-average late-summer flows, in contrast to non-glaciated basins that fell below normal, implying significant glacier mass loss. Capacity for such buffering will diminish with continued glacier retreat. - Landslides: Hundreds of post-wildfire debris flows occurred in summer/fall 2021 in southwestern BC. The Lytton Creek Fire (~84,000 ha) was followed by debris flows that damaged highways and railways. On 16 August, debris flows impacted vehicles on the TransCanada Highway (no fatalities). A subsequent atmospheric river on 14–15 November remobilized debris, undermining bridge approaches and causing a bridge collapse. Overall, widespread infrastructure damage likely contributed to one of the most expensive natural disasters in Canada’s history.

The study demonstrates that diabatic processes—especially upstream condensational heating within frontal systems and local land-surface heating—played a dominant role in creating the unprecedented low-level warmth, with adiabatic subsidence also contributing. The evolution and persistence of a strong blocking high were central to the event’s severity, consistent with theory and related studies of atmospheric blocking. Forecast systems provided actionable early warning: ensemble forecasts indicated an impending extreme as early as 8–12 days before peak, enabling meteorologists to communicate risks of daily and potentially all-time records. Subseasonal forecasts captured elevated probabilities of extreme temperature and blocking with 10–20 days lead, albeit with spatial biases that reduced utility at longer lead times. The findings contextualize the event within a warming climate: while a precise attribution of dynamic drivers remains uncertain for such a rare, record-shattering event, the thermodynamic background warming (~1.0 °C regionally in summer relative to the late 19th century) increased the likelihood and severity of extremes. Initial PAMIP analyses showed no robust changes in North American waviness from sea-ice loss or SST changes in two models, underscoring the need for broader, more nuanced assessments. The catastrophic impacts across health, ecosystems, agriculture, hydrology, and infrastructure illustrate how compound and cascading hazards arise during and after extreme heat, and highlight social inequities in vulnerability (e.g., First Nations communities, elderly, and materially/socially deprived neighborhoods). The study argues for improved early warning, risk communication, and preparedness, including new alert categories (e.g., BC’s “Extreme Heat Emergency”) and targeted protections for vulnerable populations.

This paper provides a comprehensive, cross-sectoral assessment of the June 2021 PNW heatwave, documenting its unprecedented temperature anomalies and record exceedances, the dominant role of diabatic heating and atmospheric blocking, the strong but imperfect forecast performance, and widespread catastrophic impacts—from hundreds of deaths and massive intertidal die-offs to severe agricultural losses, wildfire outbreaks, flood and debris-flow hazards, and substantial glacier mass loss. The work underscores that extreme heat events will become more frequent and intense with ongoing anthropogenic warming and that preparedness and early warning must be enhanced. Future research should refine attribution methods for rare extremes; better quantify the roles of moisture, diabatic heating, and land-atmosphere feedbacks in blocking development; expand evaluation of Arctic amplification impacts using diverse metrics, models, and configurations; improve subseasonal prediction of block location and intensity; and develop sector-specific impact models and intervention strategies to reduce both hazard and vulnerability, with a focus on climate justice and protection of at-risk communities.

- Trajectory analysis relied on forecast (GFS) data rather than reanalysis or observations; while forecasts were skillful, they underestimated peak temperatures by ~1–3 °C and may influence quantified heating contributions. - Attribution of dynamic drivers is limited: PAMIP analysis used only two models with atmosphere-only configurations, fixed boundary conditions, and daily data availability; sea-ice loss alone may not capture the full Arctic amplification response. Results should not be taken as conclusive regarding AA’s role. - Estimating return periods for such rare events is highly uncertain given limited observational records and non-stationarity under climate change. - Agricultural yield attribution is difficult because annual yields integrate multiple drivers; while NDVI timing supports within-heatwave damage, isolating the heatwave’s exact contribution remains uncertain, and irrigation practices vary regionally. - Station ‘virtual’ records for extremes have not been fully quality-controlled for non-climatic artifacts; ERA5 spatial averaging may dampen extremes relative to point observations. - Subseasonal forecast skill for location was limited at longer leads, with blocks often placed too far west. - Impact assessments (e.g., marine mortality totals, landslide inventories) include site-based scaling and qualitative reports; comprehensive region-wide quantification is uncertain.