MESMER-M: an Earth System Model emulator for spatially resolved monthly temperatures

Abstract

Abstract. The degree of trust placed in climate model projections is commensurate to how well their uncertainty can be quantified, particularly at timescales relevant to climate policy makers. On interannual to decadal timescales, model uncertainty due to internal variability dominates and is imperative to understanding near-term and seasonal climate events, but hard to quantify owing to the computational constraints on producing large ensembles. To this extent, emulators are valuable tools for approximating climate model runs, allowing for exploration of the model uncertainty space surrounding select climate variables at a significantly reduced computational cost. Most emulators, however, operate at annual to seasonal timescales, leaving out monthly information that may be essential to assessing climate impacts. This study extends the framework of an existing spatially resolved, annual-scale Earth System Model (ESM) emulator (MESMER, Beusch et al. 2020) by a monthly downscaling module (MESMER-M), thus providing local monthly temperatures from local yearly temperatures. We first linearly represent the mean response of the monthly temperature cycle to yearly temperatures using a simple harmonic model, thus maintaining month to month correlations and capturing changes in intra-annual variability. We then construct a month-specific local variability module which generates spatio-temporally correlated residuals with month and yearly temperature dependent skewness incorporated within. The performance of the resulting emulator is demonstrated on 38 different ESMs from the 6th phase of the Coupled Model Intercomparison Project (CMIP6). The emulator is furthermore benchmarked using a simple Gradient Boosting Regressor based, physical model trained on biophysical information. We find that while regional-scale, biophysical feedbacks may induce non-uniformities in the yearly to monthly temperature downscaling relationship, statistical emulation of regional effects shows comparable skill to approaches with physical representation. Thus, MESMER-M is able to generate ESM-like, large initial-condition ensembles of spatially explicit monthly temperature fields, thereby providing monthly temperature probability distributions which are of critical value to impact assessments.

TIMBER v0.1: a conceptual framework for emulating temperature responses to tree cover change

“…In order to jointly explore future tree cover and GHG scenarios, coupling TIMBER v0.1 with other temperature emulators such as MESMER-M or -X (Beusch et al, 2020;Nath et al, 2022b;Quilcaille et al, 2022) also proves worthwhile. In doing so, care would have to be taken to not "double-count" the tree cover change signal as MESMER-M and -X are trained on SSP runs, which contain both GHG and tree cover change signals.…”

“…By statistically representing select climate variables, emulators are able to reduce the dimensionality of climate model outputs, allowing for agile exploration of the uncertainty phase space surrounding climate projections. Climate model emulators designed to reproduce regional/grid point level, annual to monthly temperature projections usually operate as ESM-specific and start by deterministically representing the regional/grid point level mean response of temperatures to a certain forcing, after which the residual variability -treated as the uncertainty due to natural climate variability -is sampled or stochastically generated (Alexeeff et al, 2018;McKinnon and Deser, 2018;Link et al, 2019;Castruccio et al, 2019;Beusch et al, 2020;Nath et al, 2022b). Outputs of such emulators act as approximations of multi-model initial-condition ensembles, providing distributions of temperature responses to the forcing of choice for impact assessments.…”

STITCHES: creating new scenarios of climate model output by stitching together pieces of existing simulations

“…Thus far, the need for additional scenarios not available in ESM output archives has been addressed -if at all -by simple emulators of ESM output, usually producing multidecadal averages of temperature and -separately -precipitation change fields. Most popular has been simple pattern scaling, starting from its initial conception (Santer et al, 1990), popularized by the software MAGICC-SCENGEN (http://www.magicc.org/ (last access: 2 November 2022); Meinshausen et al, 2011), and made more sophisticated by the possibility of producing higher-frequency fields, thus representing internal variability, for example by Link et al (2019) and Nath et al (2022). More complex emulators have also been proposed, departing from pattern scaling (Castruccio et al, 2014) or extensions of pattern scaling that use zonal averages to drive the emulation (Schlosser et al, 2013) or that emulate other metrics besides average temperature and precipitation (Huntingford and Cox, 2000), even extremes (Tebaldi et al, 2020;Quilcaille et al, 2022).…”

Showcasing MESMER-X: Spatially resolved emulation of annual maximum temperatures of Earth System Models

“…For the representation of natural variability in spatially resolved emulators, there is no single most established method. Some emulators resample actual ESM fields (Alexeeff et al, 2018;McKinnon et al, 2017), some resample principle components with perturbed phases (Link et al, 2019), and others rely on autoregressive processes with spatially correlated innovations (Beusch et al, 2020;Nath et al, 2021). Almost all spatially resolved emulation approaches have been developed to emulate mean quantities.…”