16 Apr (CLIM) Shukla, A Personal Retrospective

Origins of Reanalysis, Seasonal Prediction, and Inclusion of Land Effects: A Personal Retrospective 

Wed, 16 Apr, 1:30pm 

Innovation 134 and via Zoom (for link, email bklinger@gmu.edu) 

This seminar has two parts. The first part is to thank my students, professors and research colleagues at MIT, GSFC/NASA, UMD, COLA, GMU, and other research centers in the world with whom I had the privilege of working during the past five decades.  

The second part of the seminar was inspired by a question one of the students asked me some time ago, “How did you get research ideas?” The same question was asked by my collaborator who helped me write my memoir, “A Billion Butterflies: A Life in Climate and Chaos Theory”. The seminar will give a brief personal retrospective of the origins of ideas for modern reanalysis, dynamical seasonal prediction (DSP), and the importance of land surface processes for modelling and prediction of weather and climate.   

During the mid-20th Century, the butterfly effect and the limits of weather predictability were the dominant paradigms, indicating that dynamical seasonal prediction would not be possible. Yet the demonstration of significant impacts of slowly varying boundary conditions of sea surface temperature and soil wetness using fledgling climate models of early 1980s provided a scientific basis for research on dynamical seasonal prediction (DSP). This led to the establishment of Centre for Ocean – Land – Atmosphere Studies to demonstrate the feasibility of societally beneficial dynamical seasonal prediction. At the same time, several weather and climate prediction centers began to produce seasonal forecast with prescribed and persistent boundary conditions of sea surface temperature. Within a decade, global coupled Ocean-Atmosphere models succeeded in simulation and prediction of sea surface temperature for 1-2 seasons, and DSP became operational like NWP.  

Reanalysis products have now become an indispensable data source for weather and climate research and have recently been important for training artificial intelligence (AI)/machine learning (ML) models. There are several national and international research programs and field experiments for measurements and parametrizations of land surface processes demonstrating the importance of land surface processes in enhancing predictability of sub-seasonal and seasonal variations of weather and climate.