Spatial Controls of Lower Tropospheric Stability

TL;DR

This work investigates the spatial controls of lower tropospheric stability (EIS) and its role in marine low cloud radiative feedbacks. Using regularized linear regression (ridge) on four GCMs and ERA5 data, it maps how regional EIS responds to near-surface temperature patterns, revealing that EIS rises with warming in tropical ascent regions and falls with warming in descent regions, with substantial nonlocal (remote) influence. The study further shows that Rossby-wave teleconnections modulate subtropical EIS, challenging the idea that the West Pacific Warm Pool alone governs low-cloud stability, and demonstrates that observed Southeast Pacific EIS trends since 1980 are dominated by remote warming, a finding that helps explain discrepancies with GCMs that misrepresent SST patterns. Nonlinearities are found to be weak in the historical period, supporting the use of linear sensitivity maps to interpret the pattern effect and providing a pathway to constrain low-cloud radiative feedbacks under future warming scenarios by conditioning on observed surface temperature patterns.

Abstract

Marine low clouds play a crucial role in Earth's radiation budget. These clouds efficiently reflect sunlight and drive the magnitude and sign of the global cloud feedback. Despite their relevance, the evolution of shallow cloud decks over the last decades is not well understood. One of the dominant controls of this low cloud cover is the lower tropospheric stability, quantified by the estimated inversion strength (EIS). Here, we quantify how regional EIS depends on local and remote surface temperature, revealing the dynamics controlling the characteristics of shallow clouds. We find that global EIS increases with warming in tropical regions of ascent and decreases with warming in regions of descent, as expected. In addition to the West Pacific Warm Pool, the Atlantic convection regions and the central Pacific are important predictors. Focusing on subtropical ocean upwelling regions in different ocean basins, where the low cloud decks reside, EIS increases with a fairly complex pattern of remote warming and decreases with local warming. The spatial relationship between surface temperature and EIS is robust across different climate models and reanalyses, allowing us to constrain the large spread in estimates of historical EIS trends. In the Southeast Pacific, where historical temperature trends are not well understood, we attribute the observed increased EIS since 1980 entirely to remote warming, indicating that local cooling did not increase stability in this region. Our results put into question the dominance of the West Pacific Warm Pool in controlling low cloud feedbacks in the eastern Pacific and give insights into mechanisms underlying the spatial dependence of radiative feedbacks on surface temperature patterns.

Figures (14)

• Figure 1: Controls of regional estimated inversion strength. Top row shows the sensitivity to surface temperature of $\langle {\rm EIS} \rangle$ averaged over regions indicated by red boxes: (A) Southeast Pacific, (B) Northeast Pacific, (C) Southeast Atlantic, based on ridge regression on data from four climate models. Hatching indicates regions where sensitivity maps obtained from the four models separately do not agree on the sign. Bottom rom (D-F) shows the sensitivity maps in the same regions, but using ERA5 data. $R^2$ values are calculated for held-back testing members and displayed above each map.
• Figure 2: Free tropospheric temperature response to surface warming. (A) Sensitivity of $\langle T_{700}\rangle_{\rm SEP}$ to surface temperature based on ridge regression on data from four climate models. Hatching indicates regions where sensitivity maps obtained from the four models separately do not agree on the sign. (B and C) Average response of $T_{700}$ to localized surface warming in the West Pacific Warm Pool and central Pacific (indicated by green ovals). Red boxes show the SEP, NEP, and SEA regions. Red arrows illustrate tropical temperature transport via gravity waves, blue arrows illustrate Rossby wave propagation.
• Figure 3: Attribution of the estimated inversion strength trend in 1980-2024 in the Southeast Pacific from ERA5. (A) Surface temperature trends in 1980-2024 in ERA5. (B) Sensitivity of $\langle{\rm EIS}\rangle_{\rm SEP}$ (EIS averaged over the red box in the Southeast Pacific) to surface temperature (as Fig. 2D but including the trend during training) (C) Attribution map of $\langle{\rm EIS}\rangle_{\rm SEP}$ calculated by multiplying the observed surface temperature trend (A) with the sensitivity map (B). (D) $\langle{\rm EIS}\rangle_{\rm SEP}$ in 1980-2024 in ERA5 (black), compared to the prediction from ridge regression using the global surface temperature map (red) versus the surface temperature in the SEP box only (blue; see text for details). All values are anomalies with respect to the 1980-2024 average. (E) Mean $\langle{\rm EIS}\rangle_{\rm SEP}$ trend (1980-2024) in reanalyses and GCMs. Green shows the spread among different reanalyses (crosses indicate individual reanalyses, from left to right: MERRA2, JRA-3Q, and ERA5). Purple shows the spread among different ensemble members in four different GCMs. Orange shows the spread among predicted trends by applying our regression models (trained either on ERA5 or GCMs) to different observed surface temperature datasets. See methods and SI Fig. \ref{['SIfig8']} for details.
• Figure 4: Trends in estimated inversion strength averaged over the Southeast Pacific (SEP), Northeast Pacific (NEP), and Southeast Atlantic (SEA) in 1980-2024 and the contribution from local temperature changes. All values are in $K/decade$ with $5%$-$95%$ confidence bounds, bold indicates trends that are significantly different from zero.
• Figure S1: Climatology of estimated inversion strength (EIS) and vertical velocity at $700hPa$ ($\omega_{700}$) in 1991-2014. Top row shows the climatology averaged over three coupled climate models (CanESM5, MIROC6, and MPI-ESM1.2-LR). Bottom row shows the climatology in ERA5.