.png){kind=icon}
.png){kind=icon}
Return
Climate-change modelling at reduced floating-point precision with stochastic rounding
.png){kind=icon}
OA
AI
DOI：10.1002/qj.4435.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Reduced-precision floating-point arithmetic is now deployed routinely in numerical weather forecasting over short timescales. However, the applicability of these reduced-precision techniques to longer-timescale climate simulations-especially those that seek to describe a dynamical, changing climate-remains unclear. We investigate this question by deploying a global atmospheric, coarse-resolution model known as Simplified Parameterizations PrimitivE Equation DYnamics (SPEEDY) to simulate a changing climate system subject to increased CO2 concentrations, over a 100-year timescale. Whilst double precision is typically the operational standard for climate modelling, we find that reduced-precision solutions are sufficiently accurate. Rounding the floating-point numbers stochastically, rather than using the more common round-to-nearest technique, improves the performance of the reduced-precision solutions notably. Over 100 years, the mean bias error (MBE) in the global mean surface temperature (precipitation) relative to the double-precision solution is +1.8 x10(-2) K (-8 x 10(-4) mm.(6 hr)(-1)) when inte-grating numerically at half precision (10 significant bits) with stochastic round-ing. By examining the resultant climatic distributions that arise after 100 years, the difference in the expected value of the global surface temperature rela -tive to the double-precision solution is <= 5 x 10(-3) K and that for precipita-tion is 8 x 10(-4) mm.(6 hr)(-1). Whilst further research is necessary to extended these results to more complex and higher-resolution models, they indicate that reduced-precision techniques and stochastic rounding could be suitable for the next generation of climate models and motivate the use of low-precision hardware to this end.
Keywords:
climate models
numerical modelling
reduced-precision techniques
SPEEDY
Journal
IF:
2.9
Papers:
5.7K
Citations:
2.4W