Anna Trevisan (1946 — 2016).

Anna was born on January 25th 1946 in Bologna where she resided with brief interruptions for her entire life. Anna was a very energetic person, loved several sports and, although science always played a central role in her life, maintained strong ties with her parents and dedicated time and attention to her family, in particular to her son Luca.

Anna got the Italian degree (Laurea) in physics at the University of Bologna with a thesis on the application of time series analysis techniques to a historical record of tidal waves in the Adriatic Sea. Although very interested in physics, her strong natural inclination towards mathematics emerged very early, as she took courses in algebra, mathematical methods, etc. She was also attracted by numerical methods as soon as digital computers begun to be available to her.

Since her thesis work Anna developed and maintained international connections. In the early seventies she spent long periods of time at the Institute of Geophysiscs and Planetary Physics of University of California in San Diego, where she studied, under the guidance of Walter Munk, time series analysis applications to tidal waves. The basic problem consisted in inferring, through spectral correlation analysis, the relationship between astronomic and atmospheric (pressure, wind) forcing agents and observed tides. Already at the time of her thesis, Anna had joined the Italian National Research Council (CNR), for which she continued working till the end of her life.

During the seventies Anna studied meteorology at MIT under the guidance of Jule Charney. She further developed her skills in the discipline at the National Center for Atmospheric Research (NCAR) in Boulder, Colo., where she interacted mostly with Chester Newton; at the University of Oslo, collaborating with Arnt Eliassen;and at the University of Miami, interacting with Reiner Bleck. During these years Anna developed the first primitive equation model to use isentropic coordinates. This model was applied to numerical studies of orographic cyclogenesis: her 1976 paper (Trevisan, A., 1976: Numerical Experiments on the Influence of Orography on Cyclone Formation with an Isentropic Primitive Equation Model. J. Atmos. Sci., 33, 768–780) is one of the first contributions to the development of numerical models for orographic cyclogenesis.

In the early 1980’s Anna became interested in upper-level frontogenesis, a challenging problem in dynamic meteorology (Buzzi, Trevisan and Salustri, 1980; Newton and Trevisan, 1984, a and b). As for other problems she tackled, Anna’s approach relied upon a combination of semi-analytical and numerical methods. This combination proved very successful, again and again, in the understanding of complex phenomena and processes characterizing atmospheric dynamics.

The Alpine Experiment (ALPEX) was part of the Global Atmospheric Research Programme (GARP), a fifteen-year program jointly organized by the World Meteorological Organization (WMO) and the International Council of Scientific Unions (ICSU) that led to dramatic progress in meteorology as a whole. ALPEX enabled further investigation of all the aspects of Alpine cyclogenesis, and Anna was actively involved in the preparation phase, the special ALPEX Observing Period (1 March-30 April 1982), as well as the follow-up studies extending to 1985. Anna’s main contribution consisted in developing analytical and numerical representation of the basic mechanisms involved in secondary Alpine cyclogenesis. During those years and subsequent ones, Anna made fundamental contributions to the Bologna research group’s work. This work evolved into a complete theory of the onset of orographic cyclones, a theory that analyzed in depth and helped explain the modification of baroclinic instability induced by the presence of orography (Tosi, Fantini and Trevisan, 1983; Buzzi, Trevisan and Speranza, 1984; Speranza, Buzzi, Trevisan and Malguzzi, 1985; Malguzzi, Trevisan and Speranza, 1987; Trevisan, Ferranti and Malguzzi, 1988; Buzzi; Malguzzi and Trevisan, 1990; Trevisan and Giostra, 1990).

Since the time of her studies in time series analysis, Anna was captivated by weakly nonlinear interaction mechanisms, such as triadic resonance, in spectral analysis per se, as well as in the dynamics of planetary waves connected with the blocking phenomenon (Trevisan and Buzzi, 1980). In the mid-eighties that she became, furthermore, involved in studying the local properties in phase space of nonlinear systems (Malguzzi P., A. Trevisan, and A. Speranza, 1990).

Anna’s interest in atmospheric flow regimes and their predictability led her to compare paradigmatic system behavior (Trevisan and Legnani, 1995) with observed properties of the real atmosphere (Trevisan, 1995) and of detailed numerical models (Trevisan, Pancotti and Molteni, 2001). These studies evolved in subsequent years into devising ways to estimate instabilities in atmospheric models, by introducing the methodology of breeding on the data assimilation system (BDAS: e.g. Carrassi, Trevisan and Uboldi, 2007; Uboldi and Trevisan, 2015) along with a computational method to estimate the covariant Lyapunov vectors (Trevisan and Pancotti, 1998). The latter study anticipated by almost 20 years the development of other efficient methods for the computation of Lyapunov vectors, in the late 2000s. The use of covariant Lyapunov vectors, as a key tool to analyze the atmospheric, oceanic or climate dynamics, has grown dramatically in recent years.

In the early 2000’s, Anna became interested in the problem of combining observational data with dynamical models, a problem referred to as data assimilation. Those were the years when variational methods were being developed and applied worldwide, along with ensemble-based approaches to sequential estimation (a.k.a. “Kalman filtering”). But, rather than being attracted by the operational aspect of the problem, her point of view was based, once again, on the dynamical systems approach. She pioneered the class of methods known as assimilation in the unstable subspace (AUS: Trevisan and Uboldi, 2004), to assimilate observational data and select targeted observations using the information on the dynamics (Carrassi, Trevisan, Descamps, Talagrand and Uboldi, 2008).

In the last years of her activity, Anna was deeply involved in the theoretical and applied-oriented development of the AUS method (e.g. Carrassi, Ghil, Trevisan and Uboldi, 2008; Trevisan, D’Isidoro and Talagrand, 2010; Trevisan and Palatella, 2011; Palatella and Trevisan, 2015). The topic is still matter of intensive research by several groups worldwide, as testified by the number of related publications.

In all of her working life, Anna proved her ability to work both independently and on a team. Although never involved in formal teaching, she also devoted much time and energy to the formation of young scientists, many of whom play an important role in current research.