In the last decade, ground-based high-resolution spectroscopy (HRS) has emerged as a powerful method to probe exoplanet atmospheres both in transit and thermal emission. With tens of instruments worldwide, HRS is now producing numerous observations on many targets, revealing these planet's thermal, compositional, and dynamical structure in three dimensions. As HRS science continues to mature, novel strategies and interpretation tools will be key to extracting the maximal scientific output from these rich datasets. For this article we investigated the potential of exploiting eclipses in order to retrieve spatial constraints on exoplanet dayside atmospheres with HRS, an approach that has been successfully applied at lower spectral resolutions with space-based facilities. To attempt this, we obtained an observation programme covering eight eclipses (ingress and egress) of the ultra-hot Jupiter WASP-33b using the (R̊m∼70,000) SPIRou spectropolarimeter on the 3.6m CFHT. We analysed these data with the publicly available ATMOSPHERIX pipeline, which we combined with the Python package to fit the eclipse mapping signal. Additionally, we performed injection-recovery tests on archival SPIRou data to evaluate the detection limits that could be reached with further observations. From eight ingresses and egresses of WASP-33b, we obtain a tentative detection of CO consistent with the literature values and archival SPIRou data, validating that eclipses may be stacked coherently to boost detection limits. In combination with longer phase-coverage dayside data, our eclipses marginally improve constraints on the planetary rotation velocity. Through injection recovery tests, we show that our results are scaling according to expectations for a synchronously rotating WASP-33b, implying that a further 15 eclipses with SPIRou (̊m∼20 h) would be required to measure the planet's rotation using this method. While our study reveals the potential of HRS eclipses observations, it also highlights the importance of solving the remaining challenges in data processing for short time series and/or slowly accelerating planets ahead of the next generation of telescopes.
Yariv et al. (2026) studied this question.