Climbing down Charney's ladder: Machine Learning and the post-Dennard era of computational climate science

  • 2020-07-30 16:19:20
  • V. Balaji
  • 0


The advent of digital computing in the 1950s sparked a revolution in thescience of weather and climate. Meteorology, long based on extrapolatingpatterns in space and time, gave way to computational methods in a decade ofadvances in numerical weather forecasting. Those same methods also gave rise tocomputational climate science, studying the behaviour of those same numericalequations over intervals much longer than weather events, and changes inexternal boundary conditions. Several subsequent decades of exponential growthin computational power have brought us to the present day, where models evergrow in resolution and complexity, capable of mastery of many small-scalephenomena with global repercussions, and ever more intricate feedbacks in theEarth system. The current juncture in computing, seven decades later, heralds an end towhat is called Dennard scaling, the physics behind ever smaller computationalunits and ever faster arithmetic. This is prompting a fundamental change in ourapproach to the simulation of weather and climate, potentially as revolutionaryas that wrought by John von Neumann in the 1950s. One approach could return usto an earlier era of pattern recognition and extrapolation, this time aided bycomputational power. Another approach could lead us to insights that continueto be expressed in mathematical equations. In either approach, or any synthesisof those, it is clearly no longer the steady march of the last few decades,continuing to add detail to ever more elaborate models. In this prospectus, weattempt to show the outlines of how this may unfold in the coming decades, anew harnessing of physical knowledge, computation, and data.


Quick Read (beta)

loading the full paper ...