Winter-proofing land surface models - quantifying the critical role of cold season processes in vegetation-permafrost feedbacks

Climate change has an amplified impact on the Arctic. The region is warming three times faster than the rest of the world. This warming is even more intense in winter, when extreme winter events such as rain-on-snow and frost droughts have been documented to cause wide-spread damage to vegetation. The ongoing warming of the cold season also raises soil temperatures, which increases the loss of permafrost carbon. Vegetation damage and permafrost thaw may both lead to an increase in the release of the greenhouse gases CO2 and methane. When global warming sets in motion natural processes that cause even more warming, like an increased release of greenhouse gases, we call this a positive climate feedback. These feedbacks need to be understood and quantified in order to project how our climate will change in the future. However, much remains uncertain. The models we use to predict the response of the arctic to climate change are not built to accurately simulate the winter. In warmer parts of the globe, where winters are short and mild, this is not a problem, but winter conditions can last for nine months in the Arctic. When models focus only on the remaining three months, they are not able to predict how climate feedbacks from the Arctic will develop in the future. This project employed two PhD students, who both defended in the winter of 2022-2023. One worked with the dynamic vegetation model LPJ-GUESS, the other with the land surface model CLM-FATES. The work with LPJ-GUESS identified that model predictions of permafrost extent and carbon loss are highly sensitive to the implemented snow scheme. Moreover, climate models predict divergent trends in snow depth into the future, with colder areas seeing a strong increase in snow depth. Thicker snow acts much like a blanket, which amplifies the warming of soils and enhances permafrost carbon loss in the coldest regions. New methods to show how greenhouse gases are pushed out of the soil during autumn freeze-in have also been developed. The research with CLM-FATES showed that vegetation models that simulate the transport of water through a plant need to account for cold adaptation, which reduces water loss in winter. Moreover, the simulation of frost damage was improved to vary with the amount of cold adaptation, leading to a more realistic timing of vegetation damage. These improvements were tested by modeling the impact of a frost drought that struck Norway in the winter of 2013-2014. In addition, this project has helped to create a pan-Arctic database of flux measurements, which revealed that carbon release during the winter may offset the uptake during summer (published in Nature Climate Change). This same work has led to a revised estimate of the annual CO2 exchange of the Arctic-Boreal Zone. The PI also contributed to a perspective in Nature Climate Change that highlighted the challenges of monitoring the dynamics of vegetation growth across the Arctic. This project has improved model projections on how changing arctic-boreal winters may affect carbon-climate feedbacks. These results are important for policy makers to inform them more accurately on the consequences of climate change for northern vegetation and the release of greenhouse gases from permafrost soils.

WINTERPROOF has shown that climatic changes in winter act as a strong control on arctic carbon feedbacks. Not only because of the demanding conditions for vegetation survival, but also by being highly important for permafrost carbon loss. This project hired two PhD students, one at the University of Oslo in Norway and one at project partner Lund University in Sweden, who successfully defended their thesis in the winter of 2022/23. The focus of their PhDs has been to make two leading ecosystem models, CLM-FATES and LPJ-GUESS, more capable in simulating cold season processes. These developments include 1) a cold hardening scheme that emulates a number of physiological changes at the molecular level that arctic and boreal perennial plants carry out in autumn, to increase their tolerance to freezing, 2) a frost mortality scheme that is dependent on the cold hardening of vegetation, varying with plant functional type 3) an advanced multi-layer snow scheme to improve simulations of soil temperature and permafrost thaw, and 4) a dynamic soil gas reservoir that redefines the way in which greenhouse gas release from permafrost soils is simulated during winter. The cold hardening and soil gas reservoir schemes are both highly novel and not present in any other kind of process model, while rooted in strong observational evidence of their high relevance for the arctic carbon budget. Cold hardening is essential to accurately model plant hydraulics and survival in winter, and this was first identified by this project. This went undetected until now because most vegetation models have a highly simplified representation of plant hydraulics, or are developed for the tropics, where frost, snow and permafrost are absent. Another major outcome of this project is that mid-winter snow depth will strongly increase across most of the permafrost region with climate warming, despite an overall shortening of the length of the snow season. Thicker snow increases the insulation of soils, which stay warmer, accelerating permafrost thaw. This project is the first to identify a divergent pattern in pan-Arctic snow depth as a major control on permafrost carbon loss. The model developments by WINTERPROOF will be included in the operational versions of CLM-FATES and LPJ-GUESS through a close collaboration with main developers. This means that project outcomes will be used by dozens of research groups, across Europe, the USA, China and elsewhere, since it strongly improves the ability of these models to accurately simulate snow dynamics, plant growth and permafrost extent. WINTERPROOF has made a strong leap forward in projecting arctic-boreal carbon feedbacks by fixing some major oversights in how winter processes are simulated. The project outcomes are highly important to better inform policy makers on the consequences of climate change for northern ecosystems and the release of greenhouse gases from permafrost soils, with potentially global consequences.

Wintertime processes have emerged as a critical influence on the arctic carbon cycle: arctic browning has been linked to plant-damaging frost events, and permafrost soils emit up to half of the annual amount of greenhouse gases during the cold season. However, land surface models lack the ability to simulate these processes, and their response to the rapid warming of the Arctic, since they tend to prematurely switch off at the onset of winter. WINTERPROOF will, therefore, assess how warming arctic winters contribute to arctic browning, permafrost thaw, and associated climate feedbacks, using the regional Earth system models WRF-CLM and RCA-GUESS. The land surface components of these models, CLM and LPJ-GUESS, will gain the unique capability to simulate frost damage to vegetation and the wintertime release of CO2 and CH4. With these tools, WINTERPROOF aims to resolve the following overarching research question: How do rapidly warming arctic winters affect vegetation productivity and greenhouse gas release from permafrost, and do interactions between these processes enhance biogeochemical and biogeophysical climate feedbacks? This project will quantify the impact of climate change on the arctic system, and show how midwinter warming caused by sea ice loss may lead to extreme winter events that damage vegetation. These events contribute to arctic browning, whose future development will be assessed. Furthermore, WINTERPROOF will investigate whether successive winter damage to vegetation can become so excessive that ecosystem composition changes - altering the surface balance, enhancing permafrost thaw, and ultimately leading to an enhanced loss of permafrost carbon. WINTERPROOF will uniquely identify connections to sea ice loss and snow cover changes, assess vegetation-permafrost interactions, and be the first to simulate arctic browning events in land surface models, strongly advancing our understanding of the arctic carbon cycle and associated climate feedbacks.