IAP-25-071

Unravelling the Co-evolution of Climate Extremes and Vegetation Dynamics across Global Ecosystems

Vegetation both responds to and regulates climate variability under anthropogenic warming [1]. The frequency and intensity of climate extremes—such as droughts, heatwaves, floods, and compound events—are projected to increase further in the coming decades [2]. At the individual plant level, vegetation may adapt through shifts in phenology, thermal acclimation, and water-use efficiency (WUE), or alternatively experience damage, mortality, and subsequent regrowth. At the ecosystem level, changes in species composition and distribution, together with land-use and land-cover change, play a critical role in modifying the biophysical properties of the land surface [3]. These cross-scale ecosystem responses influence evapotranspiration, albedo, and soil moisture, thereby modulating local carbon, energy, and water exchanges and ultimately feeding back to affect climate variability, the global carbon cycle, and teleconnections influencing atmospheric and sea-ice dynamics [4–7].

Current Earth System Models (ESMs) still struggle to capture the coupled evolution of climate extremes and vegetation feedbacks. Phenological shifts and physiological acclimation are often treated as static or weakly dynamic processes, while vegetation adaptation, mortality, and post-disturbance recovery under extreme events remain poorly constrained by observations and experimental data [8–10]. Consequently, ESMs may misrepresent the extent to which vegetation amplifies or dampens temperature and precipitation anomalies, particularly during prolonged or compound stress events. Urgent model developments are therefore needed to incorporate dynamic vegetation responses—such as acclimation, resilience, and structural reorganisation—into ESM frameworks, supported by data assimilation from multi-scale Earth observations to better constrain feedback processes and reduce uncertainties in projections of climate extremes.

This PhD project will advance understanding of the causal relationships between vegetation feedbacks and climate extremes by integrating global remote sensing data assimilation, improved land-surface model representations, and causal statistical inference. The research will quantify how vegetation responses alter the intensity, duration, and spatial variability of extreme events across contrasting biomes—from tropical forests to boreal and semi-arid systems.

To achieve this, the project will pursue the following key objectives:

(1) Constrain vegetation phenology and water fluxes through the assimilation of global satellite and flux observations into land-surface models.

(2) Develop improved vegetation modules in land-surface models to represent phenological flexibility, thermal acclimation, and dynamic WUE.

(3) Assess vegetation feedback impacts on the frequency, duration, and magnitude of climate extremes using model–data experiments.

(4) Apply causal inference methods to attribute spatial and temporal variations in extremes to vegetation feedback mechanisms.

This project is a collaboration between Dr Wenxin Zhang (University of Glasgow), Dr Armando Marino (University of Stirling), and Prof Alistair Jump (University of Stirling), bringing together complementary expertise in ecosystem modelling, remote sensing, forest ecology, and climate extremes.

Task 1 — Data assimilation of vegetation and water fluxes (Lead: Zhang & Jump)
• Compile multi-sensor datasets: MODIS/Sentinel phenology, SMAP/ESA-CCI soil moisture, GRACE water storage, and water fluxes (FLUXNET/ICOS, SAPFLUXNET)
• Implement a Bayesian or ensemble Kalman filter framework to assimilate phenology, water fluxes, and soil moisture into a land-surface or dynamic vegetation model (e.g. JULES-ES or LPJ-GUESS).
• Generate global maps of constrained vegetation states and fluxes, identifying regions of strong vegetation–hydrology coupling.
• Evaluate model performance against the compiled observation datasets.

Task 2 — Model development and feedback experiments (Lead: Zhang & Marino)

• Conduct paired experiments: (a) baseline model with static vegetation parameters and land cover fractions; (b) enhanced model including dynamic responses.
• Evaluate changes in key extreme metrics (e.g. heatwave duration, maximum consecutive dry days, soil moisture anomalies).
• Perform sensitivity experiments across biomes to separate the contributions of phenology, acclimation, and WUE feedbacks.

Task 3 — Causal attribution and synthesis (Lead: Zhang & Jump)
• Apply causal inference frameworks (structural equation modelling, Granger causality, transfer entropy) to observed and simulated datasets to infer directional linkages between vegetation change, surface fluxes, and climate extremes.
• Produce spatial maps of feedback strength (“Vegetation–Extreme Feedback Index,” VEFI) and identify amplification or damping regions.
• Evaluate model–observation consistency and derive emergent relationships to guide future model development.

Data and Resources
• Global satellite datasets (MODIS, Sentinel-1/2, SMAP, GRACE).
• Carbon/Water flux measurements (FLUXNET, ICOS, and SAPFLUX)
• Climate forcing: ERA5 and CMIP6.
• Land-surface model frameworks (JULES-ES, LPJ-GUESS) available through partners’ HPC resources at Glasgow, UKCEH, and Stirling.

Year 1

Activities: literature review; data collection and processing; develop assimilation framework; pilot assimilation of phenology and flux data; EGU/AGU

Milestones: Global phenology–flux constraint maps; Presentation at IAPETUS training event

Year 2

Activities: Implement improved vegetation response modules; perform baseline and feedback simulations; EGU/AGU

Milestones: Climate extremes characteristics; Feedback experiments; first manuscript draft

Year 3

Activities: Causal inference and attribution analyses; EGU/AGU

Milestones: Vegetation feedback and Climate Extreme causality maps; second manuscript draft

Year 3.5

Activities: Synthesis and integration; thesis writing and publications; conference dissemination

Milestones: PhD thesis; peer-reviewed papers; open-source code and datasets

Training in data assimilation and land-surface modelling; statistical methods for causal analysis; advanced programming skills in Python, C++, and Fortran; scientific communication and peer-reviewed journal writing; and teaching.