Rapid summer Russian Arctic sea-ice loss enhances the risk of recent Eastern Siberian wildfires

The study investigates why summer wildfires have increased over eastern Siberia in recent decades. Boreal wildfires affect climate, ecosystems, and human systems, and their activity depends on meteorological and surface conditions including temperature, humidity, precipitation, soil moisture, and lightning, which are influenced by atmospheric circulation. Eastern Siberia has become a hotspot of summer boreal fires and CO2 emissions since 2000, with notable increases in burned area and emissions, particularly in 2021. Summer Arctic warming on long (≥decadal) timescales is primarily linked to declining summer sea ice due to rising CO2 and ocean warming. Because sea-ice loss is strongest along the Russian Arctic, the authors hypothesize that background Arctic warming tied to Russian Arctic sea-ice decline is a major driver of increased eastern Siberian wildfire risk, potentially exceeding the influence of internal atmospheric variability such as Siberian blocking. The study focuses on vapor pressure deficit (VPD) as a key meteorological variable controlling fire risk and aims to quantify the relative contributions of background Arctic warming and internal variability (via Siberian blocking) to recent wildfire trends over 2004–2021.

Prior work shows boreal wildfire activity is sensitive to air temperature increases and humidity/precipitation decreases, with eastern Siberia exhibiting stronger recent increases in fire activity than western Siberia or North America. Internal climate variability, including the Arctic Oscillation and atmospheric blocking, has been linked to Russian fires. Long-term Arctic warming trends are mainly associated with summer sea-ice decline driven by increased CO2, Atlantic and Pacific decadal variability (AMO, PDO), and enhanced ocean heat transport into the Russian Arctic. VPD has been demonstrated to explain more variance in fire activity than individual meteorological drivers. Models simulate climatological Arctic warming due to sea-ice loss but struggle to reproduce blocking statistics, motivating observational approaches to partition influences on wildfire risk.

- Study regions and periods: Eastern Siberia defined as 90°–150°E, 60°–75°N. Russian Arctic sea-ice concentration (SIC) region 30°–130°E, 65°–85°N. Summer is June–August (JJA). Analyses split into 1979–2003 and 2004–2021. - Data: ERA5 reanalysis (SAT, dew point, Z500, winds), FWI from EFFIS/Global ECMWF Fire Forecast model, GFAS, MODIS, GFED fire datasets, SIC datasets, and ocean reanalyses for ocean heat transport (OHT). De-seasonalized daily and monthly anomalies computed relative to 1979–2021 climatology. - VPD: Computed as VPD = es(T) − ea(Td) with es = 6.109×exp[17.27T/(T+237.3)] and ea = 6.109×exp[17.27Td/(Td+237.3)], using daily-mean 2 m air temperature and dew point; also sensitivity using daily maximum temperature (SATmax). - Fire risk proxy: Use JJA-mean VPD slope over 2004–2021 as a proxy for wildfire trend, justified by strong correlation between VPD and FWI. - Siberian blocking (SB) identification: Tibaldi–Molteni (TM) one-dimensional blocking index applied over 90°–120°E. A blocking event requires GHGS > 0 and GHGN < −10 gpm (deg lat)−1 persisting ≥3 days over ≥15° longitude. - Partitioning approach: Compute JJA-mean eastern Siberian SAT and VPD time series with all days, then recompute after removing days from lag −10 to +10 around each SB event (lag 0 = block peak). The trend with SB days removed represents background Arctic warming (BAW). The difference between full and SB-removed trends represents the SB-related contribution. Contributions expressed as ratios of slopes (R_BAW and R_SB). - Composites under sea-ice states: Define low and high SIC summers using detrended SIC anomalies thresholds (≤ −0.5 and ≥ +0.5 standard deviations). Composite daily SAT, Z500, VPD, and precipitation over SB life cycles (lag −10 to +10) for low vs high SIC states; analyze persistence, zonal scale, movement, and decay. - Ocean heat transport: Compute upper (surface–150 m) zonal OHT at 30°E, 65°–85°N near Barents Sea Opening (BSO) and correlate with Russian Arctic SIC. - PVy mechanism: Calculate nondimensional meridional background potential vorticity gradient (PVy) trends (with SB days removed) to assess how enhanced BAW (from SIC loss) reduces PVy, favoring more persistent and larger-scale blocking. - Modeling support: CESM1 experiments with 1%/yr CO2 increase with interactive vs fixed sea ice to isolate sea-ice loss effect on summer Arctic warming; PAMIP CMIP6 ensembles with prescribed preindustrial vs future SIC to corroborate sea-ice-forced summer warming. Statistical significance assessed via Student’s t-test and Mann–Kendall tests.

- Strengthened linkage between eastern Siberian fire risk and Russian Arctic sea ice: JJA FWI over eastern Siberia correlates with Russian Arctic SIC at −0.36 (p<0.05) during 1979–2021; −0.56 (p<0.01) over 2004–2021; −0.03 over 1979–2003. Detrended correlations: −0.27 (1979–2021), −0.42 (2004–2021), and −0.20 (1979–2003). - Trends (normalized): Over 2004–2021, FWI increases at 1.18 STDs/decade (p<0.01), SIC declines at −1.04 STDs/decade (p<0.01). - SAT trends over eastern Siberia (2004–2021): 1.39 STDs/decade with SB included; 1.17 STDs/decade with SB days removed, implying ~84% of SAT trend from BAW and ~16% from SB changes. Using SATmax yields ~85% (BAW) and ~15% (SB). - VPD trends (proxy for wildfire risk) over eastern Siberia (2004–2021): 1.70 STDs/decade with SB; 1.35 STDs/decade without SB, indicating ~79% of VPD trend due to BAW and ~21% due to SB changes. Using SATmax gives ~84% (BAW) and ~16% (SB). VPD correlates strongly with FWI (0.89 over 2004–2021) and with SAT (0.89 over 2004–2021). VPD vs SIC correlation is −0.69 (p<0.01) over 2004–2021 and insignificant over 1979–2003. - Precipitation decreases: Summer precipitation trend over eastern Siberia is −1.09 STDs/decade (2004–2021), anti-correlated with BAW (r=−0.44, p<0.05), consistent with anticyclonic trends, subsidence, reduced cloudiness, and enhanced shortwave heating. - Blocking frequency: 76 SB events (1979–2021): 47 in 1979–2003 (1.88 per summer), 29 in 2004–2021 (1.61 per summer). Low SIC summers (≤ −0.5 STD) have 22 SB events in 12 summers (1.83 per summer), high SIC summers (≥ +0.5 STD) have 27 SB events in 12 summers (2.25 per summer). Thus, SIC decline does not increase SB frequency. - SB characteristics under low SIC: SBs occur at higher latitudes, have larger zonal scales, longer lifetimes (~9 vs ~6 days), weaker eastward movement, and slower decay; they produce stronger, more widespread heatwaves, higher VPD, and larger precipitation deficits, leading to more intense and extensive wildfires, especially during mature and decaying phases. - Mechanism via PVy: Enhanced BAW from Russian Arctic SIC loss reduces meridional PV gradient (PVy) south of Arctic warming, favoring persistent, large-scale, slow-moving blocks. Observed trends (2004–2021) show stronger negative U500 and PVy trends over eastern Siberia (with SB days removed), consistent with BAW influence. - Ocean forcing: Upper (0–150 m) BSO OHT increases (1979–2018) and correlates with Russian Arctic SIC at −0.51 (p<0.01), implicating Atlantic inflow; positive AMO and negative PDO strengthen OHT and contribute to SIC decline. - Modeling support: CESM1 and PAMIP ensembles show summer Arctic warming is primarily caused by sea-ice loss, corroborating that BAW is driven by SIC decline under external forcing.

The study addresses whether recent increases in eastern Siberian wildfires are driven primarily by Arctic sea-ice–related background warming or by internal atmospheric variability. Results indicate that BAW associated with rapid Russian Arctic summer SIC loss dominantly drives increases in VPD and fire risk (~79%), while decadal changes in Siberian blocking contribute a secondary but amplifying role (~21%). Mechanistically, sea-ice loss enhances high-latitude warming and a stationary anticyclonic anomaly, reducing PVy and favoring more persistent, larger-scale, and slower-decaying SBs. These SB characteristics intensify and extend heatwaves, dryness, and high VPD, increasing the severity and spatial extent of wildfires, especially at higher latitudes. The strengthened correlations between FWI/VPD and SAT in 2004–2021 highlight that rapid sea-ice loss enhances the sensitivity of wildfires to regional warming. Ocean heat transport via the Barents Sea Opening linked to AMO/PDO phases further explains SIC decline, tying ocean variability and external forcing to wildfire risk through BAW.

Rapid summer sea-ice loss in the Russian Arctic has substantially increased background Arctic warming over eastern Siberia, which in turn has driven most of the recent rise in wildfire risk, as quantified by VPD. Approximately 79% of the VPD trend (and ~84% of SAT trend) over 2004–2021 is attributed to BAW from SIC decline, with the remaining ~21% due to decadal changes in Siberian blocking that amplify and spatially broaden wildfire impacts by increasing block persistence, scale, and longevity through reduced PVy. The work provides an observationally grounded partitioning framework linking sea-ice loss, atmospheric dynamics, and wildfire risk, supported by climate model experiments that isolate the role of sea-ice forcing on summer Arctic warming. Future research should integrate improved satellite observations, models capable of realistically simulating high-latitude blocking and wildfire processes, and machine learning approaches to better quantify multi-factor contributions, interannual variability (e.g., ENSO), and ecosystem-specific responses (e.g., tree species, peatlands).

- Attribution granularity: The method partitions trends between BAW and SB changes via removal of SB days, but cannot disentangle individual meteorological contributors (temperature, humidity, wind, precipitation) because these variables are interdependent; VPD is used as an integrated proxy. - Model limitations: Current climate models inadequately simulate blocking frequency/structure, limiting direct model-based estimates of internal variability contributions. - Scope: Focus is on 2004–2021 trends, not detailed interannual variability or ENSO influences; SB frequency changes are assessed, but not all modes of variability are exhaustively treated. - Omitted processes: Potential influences from wildfire feedbacks, snowmelt timing, vegetation cover dynamics, peatland burning, frozen soil/hydrology, and land–atmosphere coupling are not explicitly modeled; species-specific ecosystem responses may modulate regional fire behavior. - Data and methodological assumptions: Defining low/high SIC summers via standardized thresholds and removing ±10 days around SB peaks may affect precise partitioning; PVy calculation and compositing rely on reanalysis products with inherent uncertainties.