To limit global warming below 2°C, a drastic overall reduction from current CO2 emissions is needed. We argue that scientists should also participate in this effort in their professional activity and especially Earth scientists, on the grounds of maintaining credibility and leading by example. The strategies and measures to reach a low-carbon scientific activity require detailed estimates of the current footprint of laboratories. Here, we present the footprint of six laboratories in Earth, environmental and space sciences, representative of the AGU community, with a comprehensive scope also including international research infrastructures. We propose a novel method to attribute the footprint of any research infrastructure to any given research laboratory. Our results highlight that most laboratories have annual footprints reaching 10-20 tonnes CO2 equivalent per person (tCO2e.p-1), dominated by infrastructures and specifically satellites in three cases (with footprints up to 11 tCO2e.p-1 or 60%), while air-travels and purchases remain within the top three sources in all cases (2-4 tCO2e p-1 or 10-30% each). Consequently, footprints related to commuting and laboratory functioning, about 2 tCO2e.p-1 (20%) or less, are relatively modest compared to infrastructures, purchases and air-travels. Thus, reduction measures ignoring infrastructures may not be able to achieve reductions larger than 20 to 35% even with flight quotas and a substantial reduction of purchases. Finally, we also discuss how a deeper transformation of scientific practices, away from a fast science ideal, could make Earth and environmental sciences more sustainable and at the forefront of a rapid and drastic social bifurcation.

Convective rolls contribute largely to the exchange of momentum, sensible heat and moisture in the boundary layer. They have been shown to reinforce air-sea interaction under strong wind conditions. This raises the question of how surface turbulent fluxes can, in turn, affect the rolls. Representing the air-sea exchanges during extreme wind conditions is a major challenge in weather prediction and can lead to large uncertainties in surface wind speed. The sensitivity of rolls to different representations of surface fluxes is investigated using Large Eddy Simulations. The study focuses on the Mediterranean windstorm Adrian, where convective rolls resulting from thermal and dynamical instabilities are responsible for the transport of strong winds to the surface. Considering sea spray in the parameterization of surface fluxes significantly influences roll morphology. Sea spray increases heat fluxes and favors convection. With this more pronounced thermal instability, the rolls are 30\% narrower and extend over a greater height, and the downward transport of momentum is intensified by 40\%, resulting in higher wind speeds at the surface. Convective rolls vanish within a few minutes in the absence of momentum fluxes, which maintain the wind shear necessary for their organization. They also quickly weaken without sensible heat fluxes, which feed the thermal instability required for their development, while latent heat fluxes play minor role. These findings emphasize the necessity of precisely representing the processes occurring at the air-sea interface, as they not only affect the thermodynamic surface conditions but also the vertical transport of momentum within the windstorm.