Periodic Reporting for period 1 - ForExD (Forest vulnerability to compound extremes and disturbances in a changing climate)
Reporting period: 2022-09-01 to 2025-02-28
Forests play a crucial role not only in providing breathable oxygen, supporting biodiversity and regulating water and energy exchanges, but also in mitigating climate change. Currently, forests absorb 25% of human CO2 emissions, but their sequestration potential may be negated by more frequent and intense weather extremes and disturbances in the future. Mega-droughts, massive wildfires and widespread tree mortality reported in the last decade might be early warnings of the upcoming threats to forests under climate change. Yet, process-understanding about the interactions between climate, forests and key disturbances is limited. Consequently, impacts of compound extremes and disturbances on forest CO2 sequestration potential are not yet realistically simulated in future climate projections by Earth System models.
To overcome this knowledge gap, ForExD proposes a cross-disciplinary conceptual framework placing disturbances as intrinsic components of coupled ecoclimatic variability. We will apply novel high spatiotemporal resolution remote sensing data to reconstruct disturbance histories and explore emerging analytical tools to gain understanding about the interactions between climate, forests and disturbances. We will then develop mechanistic models of forest-disturbance interactions, which will allow simulating, for the first time, compound extremes and disturbances (fires, droughts, insects, storms) and impacts in a state-of-the-art land-surface model.
The project will build on the PI's expertise on ecological monitoring and modelling and strong cross-disciplinary background on ecoclimatology and biogeochemistry to develop a systemic understanding of interactions between climate, forests and disturbances. ForExD will provide fundamental knowledge about forest dynamics and feedbacks between climate and the carbon-cycle. In doing so, ForExD will make a vital contribution to resolve long-standing uncertainties about forests' climate change mitigation potential.
PI Bastos further co-led the Chapter “Forest disturbances and carbon sinks” in the European Climate Risk Assessment (EUCRA), developed by European Energy Agency and the European Topic Centre on Climate Adaptation, in cooperation with the European Commission. The chapter provides a review of the current issues in policy-making associated with climate risks for forests and their carbon uptake potential. The chapter thus provides a crucial contribution to inform decision makers on land-based climate change mitigation strategies and the risks imposed by climate change and extreme events.
To overcome this knowledge gap, ForExD proposes a cross-disciplinary conceptual framework placing disturbances as intrinsic components of coupled ecoclimatic variability. We will apply novel high spatiotemporal resolution remote sensing data to reconstruct disturbance histories and explore emerging analytical tools to gain understanding about the interactions between climate, forests and disturbances. We will then develop mechanistic models of forest-disturbance interactions, which will allow simulating, for the first time, compound extremes and disturbances (fires, droughts, insects, storms) and impacts in a state-of-the-art land-surface model.
The project will build on the PI's expertise on ecological monitoring and modelling and strong cross-disciplinary background on ecoclimatology and biogeochemistry to develop a systemic understanding of interactions between climate, forests and disturbances. ForExD will provide fundamental knowledge about forest dynamics and feedbacks between climate and the carbon-cycle. In doing so, ForExD will make a vital contribution to resolve long-standing uncertainties about forests' climate change mitigation potential.
PI Bastos further co-led the Chapter “Forest disturbances and carbon sinks” in the European Climate Risk Assessment (EUCRA), developed by European Energy Agency and the European Topic Centre on Climate Adaptation, in cooperation with the European Commission. The chapter provides a review of the current issues in policy-making associated with climate risks for forests and their carbon uptake potential. The chapter thus provides a crucial contribution to inform decision makers on land-based climate change mitigation strategies and the risks imposed by climate change and extreme events.
PI Bastos contributed to the new DEFID2 dataset on forest disturbances in Europe (Forzieri et al., 2023).
In WP1 we focused on USA, where systematic forest inventory data and disturbance surveys are conducted on a regular basis across the whole domain. We first conducted a careful comparison of agreement between inventory-based and remote-sensing datasets, which allowed to derive uncertainty estimates of timing and location of disturbance events. Given limitations in the quality of labels to train our models identified in the preliminary analysis, we first worked on developing a new disturbance reference dataset based on the integration of the ground-based inventories, the uncertainty information from the analysis and a new radar-based disturbance detection approach.
Given the poor consistency of datasets on other disturbances beyond fire and drought, most of the work in WP2 has, so far, focused on drought and fire disturbances, and more generally on the links between climate variability and extremes and vegetation/carbon-cycle impacts. Therefore, the order of the tasks departs from the originally planned, but the overall progress is consistent with the objectives. We have first focused on quantifying fire regimes based on long-term satellite record and of drought regimes in the Iberian Peninsula (return time, intensity) and their link to vegetation activity.
Following the causal framework proposed in ForExD, we have applied multivariate logistic regression models to evaluate preconditioning and temporally compounding effects of winter climate variables on summer leaf-area index (a proxy for vegetation growth) extremes based on remote-sensing data, published in Anand et al. (2024). Using satellite vegetation optical depth data, we have analysed the large-scale patterns of vegetation drought responses to identify the relative contributions of climatic vs. land-cover and land-management factors (Xiao et al., 2023). C.Xiao was funded by HI MPG as part of the IMPRS graduate school funding.
In Na et al. (2024), we tested the hypothesis that the doubling sensitivity of global CO2 growth rate to tropical temperature was associated with climate-change driven changes in tropical drought regimes, as reported in previous studies. We used single model perturbed initial condition large ensembles (SMILES) to evaluate the role of internal climate variability versus anthropogenic climate change in the reported changes in CO2 sensitivity to temperature.
We developed a new impacts module in QUINCY for two primary insect types: bark beetles and defoliators. We further implemented the SPITFIRE model in QUINCY and extended the model to consider fire impacts on nutrient cycling, which carried scientific and technical challenges.
The perspective in Bastos et al. (2023) established the theoretical basis for developing storylines to separate different anthropogenic contributions (climate change, elevated CO2, management) to extreme event impacts, including an example illustrating their applicability.
In WP1 we focused on USA, where systematic forest inventory data and disturbance surveys are conducted on a regular basis across the whole domain. We first conducted a careful comparison of agreement between inventory-based and remote-sensing datasets, which allowed to derive uncertainty estimates of timing and location of disturbance events. Given limitations in the quality of labels to train our models identified in the preliminary analysis, we first worked on developing a new disturbance reference dataset based on the integration of the ground-based inventories, the uncertainty information from the analysis and a new radar-based disturbance detection approach.
Given the poor consistency of datasets on other disturbances beyond fire and drought, most of the work in WP2 has, so far, focused on drought and fire disturbances, and more generally on the links between climate variability and extremes and vegetation/carbon-cycle impacts. Therefore, the order of the tasks departs from the originally planned, but the overall progress is consistent with the objectives. We have first focused on quantifying fire regimes based on long-term satellite record and of drought regimes in the Iberian Peninsula (return time, intensity) and their link to vegetation activity.
Following the causal framework proposed in ForExD, we have applied multivariate logistic regression models to evaluate preconditioning and temporally compounding effects of winter climate variables on summer leaf-area index (a proxy for vegetation growth) extremes based on remote-sensing data, published in Anand et al. (2024). Using satellite vegetation optical depth data, we have analysed the large-scale patterns of vegetation drought responses to identify the relative contributions of climatic vs. land-cover and land-management factors (Xiao et al., 2023). C.Xiao was funded by HI MPG as part of the IMPRS graduate school funding.
In Na et al. (2024), we tested the hypothesis that the doubling sensitivity of global CO2 growth rate to tropical temperature was associated with climate-change driven changes in tropical drought regimes, as reported in previous studies. We used single model perturbed initial condition large ensembles (SMILES) to evaluate the role of internal climate variability versus anthropogenic climate change in the reported changes in CO2 sensitivity to temperature.
We developed a new impacts module in QUINCY for two primary insect types: bark beetles and defoliators. We further implemented the SPITFIRE model in QUINCY and extended the model to consider fire impacts on nutrient cycling, which carried scientific and technical challenges.
The perspective in Bastos et al. (2023) established the theoretical basis for developing storylines to separate different anthropogenic contributions (climate change, elevated CO2, management) to extreme event impacts, including an example illustrating their applicability.
The perspective published in Bastos et al. (2023) laid out the theoretical framework upon which ForExD is based. The perspective aimed to bridge two disparate communities/research fields: climate risk assessment and disturbance ecology, proposing a joint framework for analysis of climate risks and causal links between climatic and human drivers, ecological factors and impacts. The framework has inspired a broadening of climate risk assessments to other types of impacts across physical, ecological and human systems (Jezequel et al., 2024).
The semi-automatic labelling approach in Task 1.2 was motivated by scientific challenges regarding the quality of the reference datasets. Our approach addresses a major bottleneck in remote-sensing classification, the production of high-quality labels to train machine/deep-learning models, that often resorts to ethically questionable approaches, for example outsourcing human labour to the Global South through private providers. Our approach allows to produce a high number of labels automatically, which then can be evaluated based on expert knowledge and/or physical constraints in a more efficient and sustainable manner.
The study by Xiao et al. (2023) was the first to provide a global quantitative assessment of land management effects on drought vulnerability of vegetation.
The work developed in WP3 goes beyond the current state of the art in insect and fire disturbance modelling in that it considers fully the feedback between carbon, nitrogen and phosphorus dynamics, including dynamic stoichiometry, changes in allocation, nutrient acquisition strategies and soil microbial dynamics. This allows to simulate emerging dynamics such as non-structural carbon reserves depletion for compensatory leaf growth or boosted growth due to increased nutrient availability from insect litter.
The semi-automatic labelling approach in Task 1.2 was motivated by scientific challenges regarding the quality of the reference datasets. Our approach addresses a major bottleneck in remote-sensing classification, the production of high-quality labels to train machine/deep-learning models, that often resorts to ethically questionable approaches, for example outsourcing human labour to the Global South through private providers. Our approach allows to produce a high number of labels automatically, which then can be evaluated based on expert knowledge and/or physical constraints in a more efficient and sustainable manner.
The study by Xiao et al. (2023) was the first to provide a global quantitative assessment of land management effects on drought vulnerability of vegetation.
The work developed in WP3 goes beyond the current state of the art in insect and fire disturbance modelling in that it considers fully the feedback between carbon, nitrogen and phosphorus dynamics, including dynamic stoichiometry, changes in allocation, nutrient acquisition strategies and soil microbial dynamics. This allows to simulate emerging dynamics such as non-structural carbon reserves depletion for compensatory leaf growth or boosted growth due to increased nutrient availability from insect litter.