Ghil has played an important role in the foundations of modern theoretical
climate dynamics. During the late 1970s, he worked in the application of dynamical systems theory to problems of the climate sciences. Starting from the work of
Budyko and
Sellers, Ghil proposed a 1D Energy Balance Model able to provide a succinct but essentially correct description of the climate system. Ghil's analysis complemented the ones by Budyko and Sellers and played a key role for understanding the multistability of the Earth system, which features competing
snowball and warm states. Paleoclimatological evidence that the
Earth had indeed experienced snowball episodes in the
Pre-Cambrian emerged in the 1990s. Energy balance models like Ghil's, once supplemented with stochastic forcings (along the direction of
Hasselmann's programme) led to the discovery of phenomena like
stochastic resonance. Throughout the 1980s and 1990s Ghil contributed to the development of
data assimilation techniques in
meteorology and
oceanography, and to the theory of low-frequency variability of the atmosphere (with a special emphasis on the study of blocking), as well as to the understanding of large-scale
ocean dynamics. He introduced the use of advanced
spectral methods for the analysis of
chaotic geophysical time series, and most prominently the singular-spectrum analysis technique (SSA). In the 2000s, he extended his studies of the
El Niño-Southern Oscillation phenomenon (ENSO) using
Boolean delay and
delay differential equations, and worked on the statistics and dynamics of
extreme events. Recently, Ghil proposed the
pullback attractor as a mathematical framework able to encompass the
random and time-dependent nature of the
climate system. Another area of research has been the development of data-driven methods for reconstructing the surrogate dynamics of partially observed systems. Additionally, he has contributed to data analysis and modeling in macroeconomics and population dynamics, as well as to coupled climate-economy-biosphere modeling. == Honors and awards ==