Content area
Detecting changes in the climate system and attributing them to human influence is one of the most fundamental tasks of climate science. Since the early studies in the 1990s, climate change detection has involved comparing observed climate change with estimates of naturally-arising internal climate variability, often provided by Earth System Models. Such comparisons form a two-way street that enhances the science possible when using either tool alone, and climate change detection is just one example of this synergy. Comparing models and observations to detect climate change, however, is hampered by the difficulty of creating fair comparisons between climate variables with fundamental differences in scale, sampling, uncertainty, and definition. This work addresses these challenges by introducing new methods for making such comparisons, developing tools that resolve definitional differences, and making use of new datasets that quantify observational uncertainty. In Chapter 2, I use two initial condition climate model large ensembles to investigate the emergence of trends in Arctic longwave radiation from internal climate variability. By decomposing time-of-emergence into contributions from forced change and internal variability, I show that seasonal patterns of warming and sea ice coverage explain large seasonal differences in time-of-emergence and that the absorption of shortwave energy during the summer melt season explains inter-model differences. In Chapter 3, I develop a satellite simulator tool to produce satellite-like radiance and brightness temperature fields within Earth System Models. Outputs from this tool (COSP-RTTOV) are consistent with instrument spectral response functions and orbit sampling, as well as the physics of the host model. I demonstrate the applications of COSP-RTTOV in model evaluation, climate change detection, and satellite mission planning. In Chapter 4, I demonstrate the detection of climate change using direct observations of spectral radiation from the NASA Atmospheric Infrared Sounder (AIRS) instrument. I optimally combine multiple spectral channels to speed detection and compare spatial patterns of variability and change. In Chapter 5, I use observational uncertainty ensembles and a set of Earth System Model experiments to investigate how observational and model uncertainty influence the time-of-emergence of regional surface temperature trends. While regional warming is robustly detected over 90% of the earth’s land surface as of 2020, observational and model uncertainty also impart delays greater than 10 years of 75% of the earth’s land surface. Furthermore, the concentration of large delays in the global south provides a striking example of how historical observational practices become embedded in critical IPCC science. Overall, this thesis demonstrates how the thoughtful use of methods, tools, and data enables better comparisons of models and observations to improve our studies of climate change detection.
