Climate change drives degradation of future observations with ground-based telescopes

Exoplanet observations with ground-based instruments are subject to climate conditions on Earth. Therefore, one important aspect in site selection for ground-based telescopes is the study of current climate conditions to optimise observing time. Since anthropogenic climate change is leading to a significant increase in global mean surface temperature, consequences for ground-based telescopes are likely [1], yet remain mostly unknown. The timescale needed to select the site and build a large telescope until its first light can easily take up more than a decade. In the case of the European Extremely Large Telescope, this process takes approximately 20 years. Together with a typical lifetime of 30 years for large telescopes, climate change  potentially degrades site conditions assessed during the site selection process noticeably until end of lifetime.
We present a study of eight sites around the world where ground-based telescopes are already in operation. The selected sites are namely Mauna Kea on the island of Hawaii (USA), San Pedro Mártir in Baja California in Mexico, the three Chilean sites Cerro Paranal, Cerro Tololo and La Silla, La Palma on the Canary Islands (Spain), Sutherland in South Africa and Siding Spring in Australia. From the observatories hosting these telescopes, we collect in situ measurements of temperature, specific and relative humidity, precipitable water vapour, cloud cover and astronomical seeing. We compare these in situ measurements to the fifth generation atmospheric reanalysis (ERA5) of the European Centre for Medium-Range Weather Forecasts and score the agreement. A reanalysis is a global and continuous assimilation of observations combined with weather and climate modelling and provides a connecting link between measurements and global climate models (GCMs). 
For a more holistic comparison and to study future trends, we use an ensemble of six of the highest resolution GCMs available with a horizontal grid spacing of 25-50 km. These GCMs are provided by the High-Resolution Model Intercomparison Project and developed as part of the EU Horizon 2020 PRIMAVERA project. We compare ERA5 climate output against historical GCM simulations and score their agreement. With this evaluation, we gain insights into the trustworthiness of future GCM simulations that were run up to 2050. Finally, we perform a Bayesian analysis of future trends. 
We find that ERA5 provides a good representation since it agrees well with in situ measurements over most sites. The comparison between ERA5 and PRIMAVERA shows a good agreement for temperature, specific humidity and precipitable water vapour, for which we find increasing future trends leading to a deteriorating quality of astronomical observations. For relative humidity, cloud cover and astronomical seeing, the confidence in future trends projected by the GCMs is low, due to an inadequate representation of climate conditions in comparison to ERA5. Also, the trends found for these variables are not significant.
With this study, we show that climate change should be considered an important aspect of instrumentation design for ground-based telescopes, especially for high-contrast imaging observations.

References:

[1] Cantalloube, F., Milli, J., Böhm, C. et al. The impact of climate change on astronomical observations. Nature Astronomy 4, 826–829 (2020).