The carbon cycling in terrestrial ecosystems is complex. In our analyses, we found that both the global and the northern-high-latitude (NHL) ecosystems will continue to have positive net ecosystem production (NEP) in the next few decades under four global change scenarios but with large uncertainties. NHL ecosystems will experience faster climate warming but steadily contribute a small fraction of the global NEP. However, the relative uncertainty of NHL NEP is much larger than the global values.

The realistic representation of global teleconnections is an indispensable requirement for the reliable simulation of low-frequency climate variability and climate change. We present an application of the complex network framework to quantify and evaluate large-scale interactions within and between ocean and atmosphere in 22 historical CMIP6 climate projections with respect to two century-long reanalyses.

Permafrost thaws and releases carbon (C) as the Arctic warms. Most earth system models (ESMs) have poor estimates of C stored now, so their future C losses are much lower than using the permafrost C model with climate inputs from six ESMs. Bias-corrected soil temperatures and plant productivity plus geoengineering lowering global temperatures from a no-mitigation baseline scenario to a moderate emissions level keep C in the soil worth about USD 0–70 (mean 20) trillion in climate damages by 2100.

Solar geoengineering has been proposed as a way of counteracting the warming effects of increasing greenhouse gases by reflecting solar radiation. This work shows that solar geoengineering can slow down the northern-high-latitude permafrost degradation but cannot preserve the permafrost ecosystem as that under a climate of the same warming level without solar geoengineering.

Precipitation change is an important consequence of climate change, but it is hard to detect and quantify. Our intuitive method yields robust and interpretable detection of forced precipitation change in three observational datasets for global mean and extreme precipitation, but the different observational datasets show different magnitudes of forced change. Assessment and reduction of uncertainties surrounding forced precipitation change are important for future projections and adaptation.

For the first time, our decadal prediction system based on Max Planck Institute Earth System Model enables prognostic atmospheric CO₂ with an interactive carbon cycle. The evolution of CO₂ fluxes and atmospheric CO₂ growth is reconstructed well by assimilating data products; retrospective predictions show high confidence in predicting changes in the next year. The Earth system predictions provide valuable inputs for understanding the global carbon cycle and informing climate-relevant policy.

Climate projections and multi-model ensemble weighting are increasingly used for climate assessments. This study examines the sensitivity of projections to multi-model ensemble weighting strategies in the south-central United States. Model weighting and ensemble means are sensitive to the domain and variable used. There are numerous findings regarding the improvement in skill with model weighting and the sensitivity associated with various strategies.

In this study, we benchmark the forecast skill of the NASA’s Goddard Earth Observing System subseasonal-to-seasonal (GEOS-S2S version 2) hydrometeorological forecasts in the High Mountain Asia (HMA) region. Hydrometeorological forecast skill is dependent on the forecast lead time, the memory of the variable within the physical system, and the validation dataset used. Overall, these results benchmark the GEOS-S2S system’s ability to forecast HMA hydrometeorology on the seasonal timescale.

Multi-instrument records with varying signal-to-noise ratios are becoming increasingly common as legacy sensors are upgraded, and data sets are modernized. Induced changes in higher-order statistics such as the autocorrelation and variance are not always well captured by cross-calibration schemes. Here we investigate using synthetic examples how strong resulting biases can be and how they can be avoided in order to make reliable statements about changes in the resilience of a system.

In this study we investigate an ocean-based carbon dioxide removal method: macroalgae open-ocean mariculture and sinking (MOS), which aims to cultivate seaweed in the open-ocean surface and to sink matured biomass quickly to the deep seafloor. Our results suggest that MOS has considerable potential as an ocean-based CDR method. However, MOS has inherent side effects on marine ecosystems and biogeochemistry, which will require careful evaluation beyond this first idealized modeling study.

Climate change has been proven to be an indisputable fact and to be occurring at a faster rate in boreal forest areas. The results of this paper show that boreal forest coverage has shown an increasing trend in the past 3 decades, and the area of broad-leaved forests has increased more rapidly than that of coniferous forests. In addition, temperature rather than precipitation is the main climate factor that is driving change.

We provide evidence that biogeophysical effects of afforestation can counteract the favorable biogeochemical climate effect of reduced CO₂ concentrations. By changing the land surface characteristics, afforestation reduces vegetation surface temperatures, resulting in a reduced outgoing longwave radiation in summer, although CO₂ concentrations are reduced. Since forests additionally absorb a lot of solar radiation due to their dark surfaces, afforestation has a total warming effect.

Spatially compounding soy harvest failures can have important global impacts. Using causal networks, we show that soy yields are predominately driven by summer soil moisture conditions in North and South America. Summer soil moisture is affected by antecedent soil moisture and by remote extra-tropical SST patterns in both hemispheres. Both of these soil moisture drivers are again influenced by ENSO. Our results highlight physical pathways by which ENSO can drive spatially compounding impacts.

We present a simple system to forecast the Madden–Julian Oscillation (MJO). We use atmospheric circulation as input to our system. We found a good-skill forecast of the MJO amplitude within 40 d using this methodology. Comparing our results with ECMWF and machine learning forecasts confirmed the good skill of our system.