Milan Kloewer (he|him)

Abstract:

Most Earth-system simulations run on conventional central processing units in 64-bit double precision floating-point numbers Float64, although the need for high-precision calculations in the presence of large uncertainties has been questioned. Fugaku, currently the world's fastest supercomputer, is based on A64FX microprocessors, which also support the 16-bit low-precision format Float16. We investigate the Float16 performance on A64FX with ShallowWaters.jl, the first fluid circulation model that runs entirely with 16-bit arithmetic. The model implements techniques that address precision and dynamic range issues in 16 bits. The precision-critical time integration is augmented to include compensated summation to minimize rounding errors. Such a compensated time integration is as precise but faster than mixed precision with 16 and 32-bit floats. As subnormals are inefficiently supported on A64FX the very limited range available in Float16 is 6 × 10−5 to 65,504. We develop the analysis-number format Sherlogs.jl to log the arithmetic results during the simulation. The equations in ShallowWaters.jl are then systematically rescaled to fit into Float16, using 97% of the available representable numbers. Consequently, we benchmark speedups of up to 3.8x on A64FX with Float16. Adding a compensated time integration, speedups reach up to 3.6x. Although ShallowWaters.jl is simplified compared to large Earth-system models, it shares essential algorithms and therefore shows that 16-bit calculations are indeed a competitive way to accelerate Earth-system simulations on available hardware.

Abstract:

International outdoor athletics championships are typically hosted during the summer season, frequently in hot and humid climatic conditions. Therefore, we analyzed the association between apparent temperature and heat-related illnesses occurrence during international outdoor athletics championships and compared its incidence rates between athletics disciplines. Heat-related illnesses were selected from illness data prospectively collected at seven international outdoor athletics championships between 2009 and 2018 using a standardized methodology. The Universal Thermal Climate Index (UTCI) was calculated as a measure of the apparent temperature based on weather data for each day of the championships. Heat-related illness numbers and (daily) incidence rates were calculated and analyzed in relation to the daily maximum UTCI temperature and between disciplines. During 50 championships days with UTCI temperatures between 15℃ and 37℃, 132 heat-related illnesses were recorded. Average incidence rate of heat-related illnesses was 11.7 (95%CI 9.7 to 13.7) per 1000 registered athletes. The expected daily incidence rate of heat-related illnesses increased significantly with UTCI temperature (0.12 more illnesses per 1000 registered athletes/°C; 95%CI 0.08-0.16) and was found to double from 25 to 35°C UTCI. Race walkers (RR = 45.5, 95%CI 21.6-96.0) and marathon runners (RR = 47.7, 95%CI 23.0-98.8) had higher heat-related illness rates than athletes competing in short-duration disciplines. Higher UTCI temperatures were associated with more heat-related illnesses, with marathon and race walking athletes having higher risk than athletes competing in short-duration disciplines. Heat-related illness prevention strategies should predominantly focus on marathon and race walking events of outdoor athletics championships when high temperatures are forecast.

Abstract:

Hundreds of petabytes are produced annually at weather and climate forecast centers worldwide. Compression is essential to reduce storage and to facilitate data sharing. Current techniques do not distinguish the real from the false information in data, leaving the level of meaningful precision unassessed. Here we define the bitwise real information content from information theory for the Copernicus Atmospheric Monitoring Service (CAMS). Most variables contain fewer than 7 bits of real information per value and are highly compressible due to spatio-temporal correlation. Rounding bits without real information to zero facilitates lossless compression algorithms and encodes the uncertainty within the data itself. All CAMS data are 17× compressed relative to 64-bit floats, while preserving 99% of real information. Combined with four-dimensional compression, factors beyond 60× are achieved. A data compression Turing test is proposed to optimize compressibility while minimizing information loss for the end use of weather and climate forecast data.

Abstract:

Abstract Growth in aviation contributes more to global warming than is generally appreciated because of the mix of climate pollutants it generates. Here, we model the CO2 and non-CO2 effects like nitrogen oxide emissions and contrail formation to analyse aviation’s total warming footprint. Aviation contributed approximately 4% to observed human-induced global warming to date, despite being responsible for only 2.4% of global annual emissions of CO2. Aviation is projected to cause a total of about 0.1 °C of warming by 2050, half of it to date and the other half over the next three decades, should aviation’s pre-COVID growth resume. The industry would then contribute a 6%–17% share to the remaining 0.3 °C–0.8 °C to not exceed 1.5 °C–2 °C of global warming. Under this scenario, the reduction due to COVID-19 to date is small and is projected to only delay aviation’s warming contribution by about five years. But the leveraging impact of growth also represents an opportunity: aviation’s contribution to further warming would be immediately halted by either a sustained annual 2.5% decrease in air traffic under the existing fuel mix, or a transition to a 90% carbon-neutral fuel mix by 2050.