Linear Response Function Reveals the Most Effective Remote Forcing in Causing September Arctic Sea Ice Melting in CESM

Abstract

The Arctic sea ice reaches its minimum state during boreal summer and fall and is also the most susceptible to climate forcing during these seasons (e.g., Devasthale et al., 2013;Stroeve et al., 2012). Therefore, it is of both scientific and societal importance to understand the predictability of the Arctic sea ice during summer and fall seasons and its underlying physical mechanism. Several factors are found to affect the predictability skill of the September Arctic sea ice on interannual time scales including the initial states of sea ice, oceans and stratosphere in and around the Arctic in the pre-melting seasons (e.g., preceding winter and spring) (see a review paper by Guemas et al., 2014 and references therein) and atmospheric teleconnection in summer (Baxter et al., 2019;Ding et al., 2014Ding et al., , 2019. For example, Baxter et al. (2019) found a leading tropical-Arctic teleconnection during 1979-2017 summers in observations and showed that sea surface temperature (SST) cooling over the eastern central tropical Pacific ocean and resulting decreased convection likely drives a Rossby wave train toward the Arctic and results in an anomalous high pressure over northeastern Canada and Greenland that in turn plays a key role in melting the sea ice in the region. They also argued that this "Pacific-Arctic (PARC) teleconnection" pattern was responsible for the observed accelerated Arctic sea ice melting from 2007 to 2012. However, Baxter et al. (2019) also noted that a 1800-year long pre-industrial simulation of the Community Earth System Model version 1 (CESM1) can't replicate the observed PARC teleconnection and instead simulates a leading coupled pattern that links a warm western and central tropical Pacific SST to an anomalous high pressure over the Pacific side of the Arctic. The discrepancy between observations and models is not entirely clear and could possibly be due to either model limitations and/or

The Most Effective Remote Forcing in Causing U.S.‐Wide Heat Extremes as Revealed by CESM Green's Function Experiments

“…Thus $$\tilde{\mathbf{L}}$$ has M × N dimension, independent of the number of experiments. Then the singular value decomposition of $$\tilde{\mathbf{L}}$$ is calculated, and the most effective forcing and its corresponding response is provided by the pair of singular vectors associated with the smallest singular number and thus maximum response‐to‐forcing ratio (Barsugli & Sardeshmukh, 2002; Dong et al., 2019; Goodman & Marshall, 2002; Hassanzadeh & Kuang, 2016; Liu et al., 2018; Lu et al., 2020; Marshall & Molteni, 1993; Wu et al., 2021). The right singular vector, known as the neutral vector, reveals the most excitable mode of the response, and the left singular vector, which corresponds to the optimal forcing, is the most effective q ‐flux forcing in driving the response.…”

“…The CAM5 Green's function model output data used for the analysis in the study are available at Columbia University Academic Commons via (Wu, 2023). The ERA5 reanalysis data set is downloaded from the Copernicus Climate Change Service (C3S) Climate Date Store -https://cds.climate.copernicus.eu/#!/search?text=ER-A5&type=dataset.…”

Neutral Mode Dominates the Forced Global and Regional Surface Temperature Response in the Past and Future

“…Assuming quasi‐equilibrium, the slow‐evolving state of TS ($$T$$) can be thought of as being maintained by $$$\mathcal{L}{\delta}{\mathit{T}}_{\mathrm{l}}={\delta}f$$$ where $${T}_{l}$$ is the linear component of the TS response, $$\mathcal{L}$$ is the LRF that relates the linear response to the external radiative forcing, and $$\delta f$$ is the radiative forcing perturbation at TOA. There are multiple methods to estimate $$\mathcal{L}$$, among which the Green's function approach has proved to be more accurate and effective (Dong et al., 2019; Hassanzadeh et al., 2016; Lin et al., 2021; F. Liu, Lu, Garuba, Leung, et al., 2018; Y. Wu et al., 2021), despite the computational cost.…”

“…where 𝐴𝐴 𝐴𝐴𝑙𝑙 is the linear component of the TS response, 𝐴𝐴  is the LRF that relates the linear response to the external radiative forcing, and 𝐴𝐴 𝐴𝐴𝐴𝐴 is the radiative forcing perturbation at TOA. There are multiple methods to estimate 𝐴𝐴  , among which the Green's function approach has proved to be more accurate and effective (Dong et al, 2019;Hassanzadeh et al, 2016;Lin et al, 2021;Y. Wu et al, 2021), despite the computational cost.…”

Increased atmospheric river frequency slowed the seasonal recovery of Arctic sea ice in recent decades

“…To examine the role of the observed SST variability, we employ a "pacemaker" experiment (PAC2) same as LENS2, except that the tropical Pacific SSTa is nudged towards the observed variations (Methods). We focus on the tropical Pacific here as it is the region with the most prominent modes of natural climate variability, that can exert a stronger influence on the Arctic than any other ocean basins 37,38 . Figure 5(c) shows that PAC2 captures the significant AR increases around the ABK region with a spatial pattern similar to that of GOGA2.…”