Research

Energy budget in initialized climate predictions

a) Earth's energy imbalance in observations and in hindcasts from CESM2 SMYLE. They capture the drop after the 1991 Pinatubo eruption and the swings associated with El Niño and La Niña. b) Prediction skill of internal variability in the energy imbalance, measured against satellite observations over 2000–2022 after removing the forced response. The skill lasts for more than a year.

Typical climate projections follow the forced response but have their own internal variability. In contrast, climate predictions on seasonal to decadal timescales are initialized from observations and thus synchronized with the real world, much like weather forecasts. We investigate whether this initialization can reconcile model–observation discrepancies in recent energy-budget trends, and how far ahead the energy imbalance can be predicted.

Earth’s energy imbalance over the last millennium

(Paper, Data, Code for proxy reconstruction) (CFMIP slides for simulations)

a) Global-mean surface temperature reconstruction over 850–2000, expressed as anomalies relative to 1961–1990. Our reconstruction, based on climate proxies like tree rings and corals, agrees well with instrumental products over the historical period. The thick black line is the ensemble mean, the thinner lines are individual ensemble members, all of which are consistent with the proxies. b) Our reconstruction of Earth's energy imbalance over the preindustrial period provides context for the current energy gain measured by satellites. The 21st-century trend in energy gain exceeds any value that can be explained by natural variability, indicating a human influence.

Earth’s energy imbalance, the net rate at which the planet gains heat, is the most fundamental metric of climate change, yet the observational record spans only a few decades. Using climate proxies such as tree rings, corals, and ice cores, we extend the energy-budget record 1000 years into the past, back to the year 850. Our reconstruction shows that the cooling trend over the last millennium was accompanied by persistent energy loss and sea ice growth, and it reaffirms that clusters of volcanic eruptions drove multidecadal cool periods as heat loss accumulated. The preindustrial record also provides context for how much the energy budget varies naturally, in the absence of anthropogenic forcing. Against that baseline, the current energy gain and its increasing trend are unprecedented.

a) Radiative response in an atmosphere-only simulation with preindustrial forcing and sea-surface temperatures from an ensemble member of our proxy-based reconstruction. Global-mean temperature alone explains 32% of the response, while 76% is explained when also accounting for temperature patterns. b) Empirical Green's function, derived from last-millennium atmosphere-only simulations by ridge regression and used to account for the pattern effect in a).

Our reconstruction also shows where the ocean warmed and cooled, not just by how much. We prescribe these patterns in atmosphere-only simulations to separate external forcing, like volcanic eruptions, from the climate response. We find that temperature patterns play a leading role in modulating the radiative response in the preindustrial period, in contrast to recent decades where global warming dominates. The last millennium thus lets us test the methods we use to understand today’s energy budget in a weakly forced climate.

High-accuracy radiation pressure modeling

(Paper, Code)

Ground track of a lunar orbit, colored by the irradiance due to lunar albedo. This irradiance is calculated from the incident solar radiation and a spherical harmonics expansion of the albedo distribution (DLAM-1). A spacecraft model then takes the direct solar and lunar albedo radiation to estimate the acceleration due to radiation pressure, resulting in small orbital perturbations.

Radiation pressure is a major perturbation in lunar orbits and must be accounted for in precision orbit determination. We compare models of varying complexity to weigh the accuracy benefits of high-fidelity radiation pressure modeling against its computational cost, implementing models for both the spacecraft and the radiation sources in the Tudat numerical astrodynamics framework. For the Lunar Reconnaissance Orbiter, we find that detailed spacecraft models are necessary to properly account for changing orientation and geometry, while complex models of lunar radiation add little benefit over simpler ones.