Large-scale photovoltaic solar farms envisioned over the Sahara desert can meet the world's energy demand while increasing regional rainfall and vegetation cover. However, adverse remote effects resulting from atmospheric teleconnections could offset such regional benefits. We use state-of-the-art Earth-system model simulations to evaluate the global impacts of Sahara solar farms. Our results indicate a redistribution of precipitation causing Amazon droughts and forest degradation, and global surface temperature rise and sea-ice loss, particularly over the Arctic due to increased polarward heat transport, and northward expansion of deciduous forests in the Northern Hemisphere. We also identify reduced El Nino-Southern Oscillation and Atlantic Nino variability and enhanced tropical cyclone activity. Comparison to proxy inferences for a wetter and greener Sahara similar to 6,000 years ago appears to substantiate these results. Understanding these responses within the Earth system provides insights into the site selection concerning any massive deployment of solar energy in the world's deserts. Plain Language Summary Solar energy can contribute to the attainment of global climate mitigation goals by reducing reliance on fossil fuel energy. It is proposed that massive solar farms in the Sahara desert (e.g., 20% coverage) can produce energy enough for the world's consumption, and at the same time more rainfall and the recovery of vegetation in the desert. However, by employing an advanced Earth-system model (coupled atmosphere, ocean, sea-ice, terrestrial ecosystem), we show the unintended remote effects of Sahara solar farms on global climate and vegetation cover through shifted atmospheric circulation. These effects include global temperature rise, particularly over the Arctic; the redistribution of precipitation (most notably droughts and forest degradation in the Amazon) and northward shift of the Intertropical Convergence Zone; the northward expansion of deciduous forests in the Northern Hemisphere; and the weakened El Nino-Southern Oscillation and Atlantic Nino variability and enhanced tropical cyclone activity. All these remote effects are in line with the global impacts of the Sahara land-cover transition similar to 6,000 years ago when Sahara desert was wetter and greener. The improved understanding of the forcing mechanisms of massive Sahara solar farms can be helpful for the future site selection of large-scale desert solar energy facilities. Key Points . A set of state-of-the-art Earth-system model simulations are used to study the impacts of large-scale (20% coverage or more) Sahara solar farms These hypothetical solar farms increase local rainfall and vegetation cover through positive atmosphere-land(albedo)-vegetation feedbacks Conveyed by atmospheric teleconnections, the Sahara solar farms can induce remote responses in global climate and vegetation cover

The majority of northern peatlands were initiated during the Holocene. Owing to their mass imbalance, they have sequestered huge amounts of carbon in terrestrial ecosystems. Although recent syntheses have filled some knowledge gaps, the extent and remoteness of many peatlands pose challenges to developing reliable regional carbon accumulation estimates from observations. In this work, we employed an individual- and patch-based dynamic global vegetation model (LPJ-GUESS) with peatland and permafrost functionality to quantify long-term carbon accumulation rates in northern peatlands and to assess the effects of historical and projected future climate change on peatland carbon balance. We combined published datasets of peat basal age to form an up-to-date peat inception surface for the pan-Arctic region which we then used to constrain the model. We divided our analysis into two parts, with a focus both on the carbon accumulation changes detected within the observed peatland boundary and at pan-Arctic scale under two contrasting warming scenarios (representative concentration pathway-RCP8.5 and RCP2.6). We found that peatlands continue to act as carbon sinks under both warming scenarios, but their sink capacity will be substantially reduced under the high-warming (RCP8.5) scenario after 2050. Areas where peat production was initially hampered by permafrost and low productivity were found to accumulate more carbon because of the initial warming and moisture-rich environment due to permafrost thaw, higher precipitation and elevated CO2 levels. On the other hand, we project that areas which will experience reduced precipitation rates and those without permafrost will lose more carbon in the near future, particularly peatlands located in the European region and between 45 and 55 degrees N latitude. Overall, we found that rapid global warming could reduce the carbon sink capacity of the northern peatlands in the coming decades.

Carbon storage dynamics in vegetation and soil are determined by the balance of carbon influx and turnover. Estimates of these opposing fluxes differ markedly among different empirical datasets and models leading to uncertainty and divergent trends. To trace the origin of such discrepancies through time and across major biomes and climatic regions, we used a model-data fusion framework. The framework emulates carbon cycling and its component processes in a global dynamic ecosystem model, LPJ-GUESS, and preserves the model-simulated pools and fluxes in space and time. Thus, it allows us to replace simulated carbon influx and turnover with estimates derived from empirical data, bringing together the strength of the model in representing processes, with the richness of observational data informing the estimations. The resulting vegetation and soil carbon storage and global land carbon fluxes were compared to independent empirical datasets. Results show model-data agreement comparable to, or even better than, the agreement between independent empirical datasets. This suggests that only marginal improvement in land carbon cycle simulations can be gained from comparisons of models with current-generation datasets on vegetation and soil carbon. Consequently, we recommend that model skill should be assessed relative to reference data uncertainty in future model evaluation studies.