CYCLOPS box model

CYCLOPS box modelThe CYCLOPS box model was originally developed by Robin Keir in the mid 1980’s — using 16 boxes to represent the surface, intermediate depth and abyss of the Atlantic, Indian, south Pacific, north Pacific, boreal north Atlantic and Southern Ocean. Keir’s CYCLOPS and PANDORA by Wally Broecker and Tsung-Hung Peng are the pinnacle of box modeling and have been used widely in the scientific literature — and CYCLOPS continues to be used today.

To represent the distinct component regions of the Southern Ocean surface CYCLOPS’ basic geometry was mildly modified by splitting the original Southern Ocean surface box into a Subantarctic zone (SAZ) box, open Antarctic Zone box and a Polar Antarctic Zone (PAZ) box (Robinson et al., 2005; Sigman et al., 2009a/b). Hence, modern CYCLOPS now has 18 boxes. Also, CYCLOPS is the only global model representing the nitrate dual-isotope system (Sigman et al., 2009a/b), which I incorporated into the original Pascal-language sources code that was passed down from Robin Keir to Danny Sigman to me.

Pascal is essentially a dead programing language at this point, the only compiler to be found today is not maintained and lacks the implementation of new CPU features such as threading and vectorization. To maintain CYCLOPS as a tool for science I re-coded CYCLOPS in the C++ programing language -- calling this new implementation CYCLOPS++ or CY++ for short. Paying close attention to computational efficiency I was able to make CY++ 200x faster than legacy CYCLOPS when run on a single CPU core making it the fastest model of its type. Moreover, I can now use all 12 cores in my workstation to yield an effective 2000x speed-up. CYCLOPS++ is the only comprehensive global ocean biogeochemical model that does not choke when tasked to run multi billion year model experiments.

Recent publications based on CYCLOPS++

My publications based on legacy CYCLOPS

back to Research page