Temperature extremes of 2022 reduced carbon uptake by forests in Europe

Europe experienced record-breaking heat in summer and autumn 2022, associated with persistent atmospheric blocking patterns and elevated Mediterranean Sea surface temperatures. Approximately 30% of Europe underwent severe summer drought (SPEI < −1.2), centered over central and southeastern regions, differing from the northern-focused 2018 drought. Such extremes, linked to positive NAO phases and human-induced warming, now have return times of roughly 10–15 years. The study aims to quantify the impact of the 2022 summer drought on carbon exchange between European forests and the atmosphere using multiple observation platforms, and to assess whether the 2022 impacts were exceptional or indicative of a new normal for future drought impacts on forests, with implications for Europe’s net-zero strategies relying on forest carbon uptake.

Prior European drought and heatwave events (2003, 2015, 2018) were associated with hemispheric wave-7 blocking and NAO phases, with several studies attributing increased frequency and severity to anthropogenic climate change. The 2003 drought caused Europe-wide reductions in primary productivity. The 2018 event exhibited spring-enhanced carbon uptake followed by summer reductions, with legacy and compound effects persisting beyond the event. Forest vulnerability studies indicate sensitivity of beech and Scots pine to drought, including increased mortality and reduced resilience. Projections suggest more frequent hot extremes and droughts across Europe under continued warming, with particular risks in southern, western, and eastern regions. Monitoring gaps, especially in eastern Europe, hinder comprehensive assessment of drought impacts and independent verification of forest carbon sequestration.

- Region selection: Drought-affected regions were delineated using 3-month SPEI for July 2018 and 2022 (regridded to 0.2° × 0.1°), selecting continuous areas >1.5 million km² with SPEI < −1.2. Four regions were defined: Centre (overlap of 2018 and 2022), North (2018-only), South (2022-only under high-pressure anomaly), and East (2022-only away from high-pressure anomaly; Hungary, Romania, Bulgaria, Ukraine, Albania). - Atmospheric data: CO2 mole fractions from 26 European stations (ICOS 2023 Obspack pre-release and Obspack 2022) were used. Representative data per station were selected (night-time/well-mixed). Anomalies were computed relative to 2019–2021 after detrending by station-specific linear trends. Background CO2 advection influence was quantified and subtracted to isolate surface flux effects; an atmospheric inverse estimate of NEE (excluding large fire emissions) was produced from these data. - Eddy-covariance (EC) data: Half-hourly flux and meteorological data from 14 forest sites (ICOS datasets) were processed: quality control and cleaning, u* threshold filtering, gap-filling with MDS, and partitioning NEE into GPP and TER via night-time temperature response (REddyProc). Anomalies were referenced to 2016–2021 (excluding 2018). - Remote sensing: MODIS BRDF-corrected surface reflectances were used to compute NIRv at 0.5° resolution. NIRv was detrended (2001–2021) to account for forest growth. NIRv anomalies were converted to GPP using biome-specific linear relationships derived from EC observations. - Biosphere modeling: The SiB4 biosphere model, driven by ERA5 meteorology with modified rooting zone depths, was downscaled to 0.1° × 0.2° by land-use type (CTE-HR framework). SiB4 provided GPP, TER, and NEE (excluding fires). Anomalies were computed against appropriate baselines (2019–2021 for atmospheric, 2008–2021 for NIRv, 2016–2021 for EC comparisons). Relative importance of stress factors (excess leaf temperature, high VPD, root-zone soil moisture deficit) was quantified to diagnose GPP reductions. - Fire emissions: GFAS (2003–2022) provided European fire carbon emissions and temporal evolution; VIIRS active fire detections assessed anomalies and spatial patterns. A high-resolution assessment of biomass loss in France used Sentinel-2 burned area and GEDI/NFI biomass maps; combustion completeness assumptions were used to infer emissions from biomass loss. Integrated additional fire-related carbon losses over the drought regions were quantified. - Climate drivers: ERA5 and ERA5-Land provided VPD, temperature, soil moisture, and geopotential height anomalies to characterize atmospheric and soil drought. SMAP L-band observations supported soil moisture assessments. - Uncertainty and validation: Cross-comparisons among NIRv-derived GPP, SiB4 outputs, EC data, and atmospheric inversions were performed. Spatial correlations and slope comparisons (AGPP/ASPEI) evaluated model performance. Limitations due to data scarcity (especially in the East) and model driver discrepancies were noted.

- Summer 2022 net carbon uptake reduction: JJA reduction of 56–62 TgC over drought-affected areas, similar to JJA 2018 (50–66 TgC). In southern France, summertime atmospheric CO2 showed >2.5 ppm excess relative to 2019–2021; about 75% of the regional CO2 anomaly in July–August was attributed to reduced biospheric uptake. - Spatial extent and drivers: Approximately 3.0 million km² under severe drought (SPEI < −1.2) in July 2022, centered in central and southeastern Europe; overlapping area with 2018 drought was 0.8 million km². JJA-2022 had the highest VPD of the last 20 years in 45%/51%/15% of the Centre/South/East areas, respectively. VPD dominated GPP reductions in South and Centre; soil moisture deficits dominated in the East. - Regional NEE anomalies: Centre area showed smaller JJA NEE anomaly in 2022 (7.8 TgC) than in 2018 (19.7 TgC). Per-unit-area response was smaller in South 2022 (2.0 gC m−2 month−1) than North 2018 (3.0 gC m−2 month−1). The East region contributed 76% of the 2022 NEE anomaly in the drought-influenced area (SiB4-based due to sparse observations). - GPP reductions: NIRv-converted GPP anomalies averaged −44.1 ± 17.4 (Centre), −50.7 ± 18.5 (South), and −47.3 ± 18.2 TgC/month (East) in JJA, totaling 142.1 ± 31.3 TgC/month across regions; forests contributed 29.3 ± 6.7 TgC/month of this GPP anomaly. SiB4 GPP anomalies integrated to −12/−18/−25 TgC/month for Centre/South/East. EC sites with July 2022 SPEI < −1 showed a 21% GPP reduction. - Autumn compensation: Warm, wetter-than-normal autumn delayed senescence, yielding 66 ± 12 TgC/month higher GPP in October–November (~30% above normal). SON atmospheric CO2 in the affected area was 2.6 ppm lower than normal (95% CI [2.4–2.9]), with ~0.6 ppm in October attributable to background advection. The enhanced autumn uptake compensated up to 32% of the summer deficit in Centre and South (contrast to ~75% spring compensation in 2018). - Fires: Despite notable regional fires (e.g., Les Landes, France), European-scale 2022 fire emissions were near normal (14.6–16.0 TgC yr−1) and ranked 5th (2003–2022). Peak GFAS emissions reached 4.5 TgC month−1 in July (highest July on record). Country peaks: Spain 1.8 TgC month−1 (July, 40% of Europe), Portugal 0.8 TgC month−1 (August), France 0.6 TgC month−1 (July). Integrated additional loss over drought regions was 5.2 TgC yr−1. French biomass loss estimated at ~0.5 TgC yr−1. JJA fire-related carbon loss was similar to 2003 (8.8–9.4 TgC) and higher than 2018 (3.9 TgC). - Model-observation consistency: SiB4 captured spatial patterns of drought impact with strong correlation to NIRv (R = 0.78; N = 41; p = 1e−5) and reasonable agreement with EC-derived GPP (R = 0.54). Atmospheric inversion results were consistent with SiB4 estimates for net summer anomalies, supporting near-real-time CTE-HR capability.

The study demonstrates that the 2022 summer drought substantially reduced European forest CO2 uptake, with magnitudes comparable to 2018 but without strong off-season compensation. This suggests that drought-induced reductions in forest carbon uptake are no longer exceptional and should be integrated into climate mitigation planning. Mechanistically, high VPD predominated in reducing GPP over southern and central Europe in 2022, while soil moisture deficits were the principal driver in eastern Europe, consistent with regional climate and land-use differences (e.g., beech forests). Although GPP and TER often co-vary, atmospheric CO2 constraints indicate that reduced TER also contributed to the observed NEE anomalies, particularly where soil moisture limitations were significant; however, large-scale TER quantification remains challenging. The warm autumn provided only partial compensation (up to 32%), contrasting with the strong spring compensation in 2018, reflecting seasonal asymmetries in phenological controls and light/soil moisture conditions. Fire emissions, while regionally impactful, contributed a relatively small fraction of the overall carbon balance anomaly at the continental scale in 2022. The findings highlight the heightened vulnerability of European forests, especially in eastern regions, under increasing drought and heat extremes, and underscore the need for enhanced monitoring to support reliable carbon budget assessments and verification of forest carbon sequestration within EU net-zero strategies.

Using multi-platform observations and modeling, the study quantifies a 56–62 TgC reduction in net summer (JJA) carbon uptake across drought-affected European regions in 2022, similar to 2018 but with limited autumn compensation (up to 32%). High VPD was the dominant stressor in the South and Centre, while soil moisture deficits drove impacts in the East. Fire emissions were regionally significant but near-normal at the European scale. Results indicate that frequent drought-driven reductions in forest carbon uptake are becoming characteristic of the current climate and must be factored into European net-zero plans that rely on forest sinks. Future work should expand observational networks in eastern Europe, improve mechanistic representation of phenology (particularly autumn processes) and TER in models, and investigate legacy and compound effects of successive warm/dry seasons on forest carbon dynamics.

- Observational gaps: Sparse atmospheric CO2 and ecosystem flux measurements in eastern Europe limited direct constraints; soil moisture observations with sufficient quality/continuity were lacking in many locations. - Model uncertainties: SiB4 underestimates GPP reductions at some EC sites due to discrepancies in environmental drivers (e.g., gridded SPEI vs site conditions); autumn phenology and TER responses are difficult to simulate mechanistically, leading to poorer model performance in SON. - TER quantification: Large-scale, observation-based constraints on TER are limited; inferences rely on co-variation with GPP and model behavior. - Atmospheric inversion: Inverse estimates used a limited station set and exclude large fire emissions (quantified separately); inversions are time-consuming and not fully real-time. - Advection influences: Although background CO2 advection was estimated and removed for anomaly interpretation, residual transport uncertainties can affect regional attribution. - Generalizability: Results for the East region rely heavily on model outputs due to limited observations, affecting confidence in regional attribution and magnitude of anomalies.