General knowledge

The role of land feedbacks depends on drought type. E.g., stomatal response to VPD: more available water but larger ecosystem stress.
The role land feedbacks may differ for drought event vs. aridity trends. But understanding event-scale can improve model projections.
The role of precipitation in drought trends is indisputable. But large uncertainties on the demand side, which is projected to vary more.
Drivers may be strongly region-dependent. Difficulty to generalise results.

General recommendations

Certain indices are better for certain drought types and applications. Indices using E for water availability, E_p for ecosystem stress.
Need to consider VPD and [CO₂] effects on plant response in trend assessments. Also when computing drought index projections.
CMIP archives should provide aerodynamic and surface resistances. That would favour benchmarking and help attribution.
Need to consider drought impacts (streamflow, forest mortality, fires). These impacts also depend on (and affect) land feedbacks.

Requirements: process-understanding

Main focus should be on understanding aerodynamic and surface resistances. And factors affecting those (e.g., ground water access, root depth, xylem conductivity).
Importance of disentangling the role of VPD and soil moisture in transpiration. Particularly given the contrasting projections in both.
New observations of fluxes and states are needed. But also the innovative use of existing ones (SIF, VOD, etc.).
Value of observation-based coupling metrics that display physical processes. For process-understanding and model benchmarking.
Differential land role in the onset versus the demise of events. Considering also a seasonal perspective on soil dry-out and biosphere growth.
Focus should expand from coupling at local scales. And embrace downwind effects and meso-scale and synoptic circulation impacts.

Requirements: model benchmarking and representation

Need for correct physics and chemistry (right results for right reasons). Important for projections in non-stationary climate conditions.
Need to evaluate model inter-variable sensitivities and links. But be able to go back to specific parameters and formulations.
Need to incorporate the effects of management and adaptation. Ecosystem adaptation, but also agricultural practices, land management, etc.

TOOLS