Search Results
You are looking at 1 - 4 of 4 items for
- Author or Editor: Matthew Chantry x
- Refine by Access: All Content x
Abstract
Attempts to include the vast range of length scales and physical processes at play in Earth’s atmosphere push weather and climate forecasters to build and more efficiently utilize some of the most powerful computers in the world. One possible avenue for increased efficiency is in using less precise numerical representations of numbers. If computing resources saved can be reinvested in other ways (e.g., increased resolution or ensemble size) a reduction in precision can lead to an increase in forecast accuracy. Here we examine reduced numerical precision in the context of ECMWF’s Open Integrated Forecast System (OpenIFS) model. We posit that less numerical precision is required when solving the dynamical equations for shorter length scales while retaining accuracy of the simulation. Transformations into spectral space, as found in spectral models such as OpenIFS, enact a length scale decomposition of the prognostic fields. Utilizing this, we introduce a reduced-precision emulator into the spectral space calculations and optimize the precision necessary to achieve forecasts comparable with double and single precision. On weather forecasting time scales, larger length scales require higher numerical precision than smaller length scales. On decadal time scales, half precision is still sufficient precision for everything except the global mean quantities.
Abstract
Attempts to include the vast range of length scales and physical processes at play in Earth’s atmosphere push weather and climate forecasters to build and more efficiently utilize some of the most powerful computers in the world. One possible avenue for increased efficiency is in using less precise numerical representations of numbers. If computing resources saved can be reinvested in other ways (e.g., increased resolution or ensemble size) a reduction in precision can lead to an increase in forecast accuracy. Here we examine reduced numerical precision in the context of ECMWF’s Open Integrated Forecast System (OpenIFS) model. We posit that less numerical precision is required when solving the dynamical equations for shorter length scales while retaining accuracy of the simulation. Transformations into spectral space, as found in spectral models such as OpenIFS, enact a length scale decomposition of the prognostic fields. Utilizing this, we introduce a reduced-precision emulator into the spectral space calculations and optimize the precision necessary to achieve forecasts comparable with double and single precision. On weather forecasting time scales, larger length scales require higher numerical precision than smaller length scales. On decadal time scales, half precision is still sufficient precision for everything except the global mean quantities.
Abstract
Motivated by recent advances in operational weather forecasting, we study the efficacy of low-precision arithmetic for climate simulations. We develop a framework to measure rounding error in a climate model, which provides a stress test for a low-precision version of the model, and we apply our method to a variety of models including the Lorenz system, a shallow water approximation for flow over a ridge, and a coarse-resolution spectral global atmospheric model with simplified parameterizations (SPEEDY). Although double precision [52 significant bits (sbits)] is standard across operational climate models, in our experiments we find that single precision (23 sbits) is more than enough and that as low as half precision (10 sbits) is often sufficient. For example, SPEEDY can be run with 12 sbits across the code with negligible rounding error, and with 10 sbits if minor errors are accepted, amounting to less than 0.1 mm (6 h)−1 for average gridpoint precipitation, for example. Our test is based on the Wasserstein metric and this provides stringent nonparametric bounds on rounding error accounting for annual means as well as extreme weather events. In addition, by testing models using both round-to-nearest (RN) and stochastic rounding (SR) we find that SR can mitigate rounding error across a range of applications, and thus our results also provide some evidence that SR could be relevant to next-generation climate models. Further research is needed to test if our results can be generalized to higher resolutions and alternative numerical schemes. However, the results open a promising avenue toward the use of low-precision hardware for improved climate modeling.
Significance Statement
Weather and climate models provide vital information for decision-making, and will become ever more important in the future with a changed climate and more extreme weather. A central limitation to improved models are computational resources, which is why some weather forecasters have recently shifted from conventional 64-bit to more efficient 32-bit computations, which can provide equally accurate forecasts. Climate models, however, still compute in 64 bits, and adapting to lower precision requires a detailed analysis of rounding errors. We develop methods to quantify rounding error in a climate model, and find similar precision acceptable across weather and climate models, with even 16 bits often sufficient for an accurate climate. This opens a promising avenue for computational efficiency gains in climate modeling.
Abstract
Motivated by recent advances in operational weather forecasting, we study the efficacy of low-precision arithmetic for climate simulations. We develop a framework to measure rounding error in a climate model, which provides a stress test for a low-precision version of the model, and we apply our method to a variety of models including the Lorenz system, a shallow water approximation for flow over a ridge, and a coarse-resolution spectral global atmospheric model with simplified parameterizations (SPEEDY). Although double precision [52 significant bits (sbits)] is standard across operational climate models, in our experiments we find that single precision (23 sbits) is more than enough and that as low as half precision (10 sbits) is often sufficient. For example, SPEEDY can be run with 12 sbits across the code with negligible rounding error, and with 10 sbits if minor errors are accepted, amounting to less than 0.1 mm (6 h)−1 for average gridpoint precipitation, for example. Our test is based on the Wasserstein metric and this provides stringent nonparametric bounds on rounding error accounting for annual means as well as extreme weather events. In addition, by testing models using both round-to-nearest (RN) and stochastic rounding (SR) we find that SR can mitigate rounding error across a range of applications, and thus our results also provide some evidence that SR could be relevant to next-generation climate models. Further research is needed to test if our results can be generalized to higher resolutions and alternative numerical schemes. However, the results open a promising avenue toward the use of low-precision hardware for improved climate modeling.
Significance Statement
Weather and climate models provide vital information for decision-making, and will become ever more important in the future with a changed climate and more extreme weather. A central limitation to improved models are computational resources, which is why some weather forecasters have recently shifted from conventional 64-bit to more efficient 32-bit computations, which can provide equally accurate forecasts. Climate models, however, still compute in 64 bits, and adapting to lower precision requires a detailed analysis of rounding errors. We develop methods to quantify rounding error in a climate model, and find similar precision acceptable across weather and climate models, with even 16 bits often sufficient for an accurate climate. This opens a promising avenue for computational efficiency gains in climate modeling.
Abstract
Based on the principle “learn from past errors to correct current forecasts,” statistical postprocessing consists of optimizing forecasts generated by numerical weather prediction (NWP) models. In this context, machine learning (ML) offers state-of-the-art tools for training statistical models and making predictions based on large datasets. In our study, ML-based solutions are developed to reduce forecast errors of 2-m temperature and 10-m wind speed of the ECMWF’s operational medium-range, high-resolution forecasts produced with the Integrated Forecasting System (IFS). IFS forecasts and other spatiotemporal indicators are used as predictors after careful selection with the help of ML interpretability tools. Different ML approaches are tested: linear regression, random forest decision trees, and neural networks. Statistical models of systematic and random errors are derived sequentially where the random error is defined as the residual error after bias correction. In terms of output, bias correction and forecast uncertainty prediction are made available at any point from locations around the world. All three ML methods show a similar ability to capture situation-dependent biases leading to noteworthy performance improvements (between 10% and 15% improvement in terms of root-mean-square error for all lead times and variables), and a similar ability to provide reliable uncertainty predictions.
Abstract
Based on the principle “learn from past errors to correct current forecasts,” statistical postprocessing consists of optimizing forecasts generated by numerical weather prediction (NWP) models. In this context, machine learning (ML) offers state-of-the-art tools for training statistical models and making predictions based on large datasets. In our study, ML-based solutions are developed to reduce forecast errors of 2-m temperature and 10-m wind speed of the ECMWF’s operational medium-range, high-resolution forecasts produced with the Integrated Forecasting System (IFS). IFS forecasts and other spatiotemporal indicators are used as predictors after careful selection with the help of ML interpretability tools. Different ML approaches are tested: linear regression, random forest decision trees, and neural networks. Statistical models of systematic and random errors are derived sequentially where the random error is defined as the residual error after bias correction. In terms of output, bias correction and forecast uncertainty prediction are made available at any point from locations around the world. All three ML methods show a similar ability to capture situation-dependent biases leading to noteworthy performance improvements (between 10% and 15% improvement in terms of root-mean-square error for all lead times and variables), and a similar ability to provide reliable uncertainty predictions.
Abstract
Benchmark datasets and benchmark problems have been a key aspect for the success of modern machine learning applications in many scientific domains. Consequently, an active discussion about benchmarks for applications of machine learning has also started in the atmospheric sciences. Such benchmarks allow for the comparison of machine learning tools and approaches in a quantitative way and enable a separation of concerns for domain and machine learning scientists. However, a clear definition of benchmark datasets for weather and climate applications is missing with the result that many domain scientists are confused. In this paper, we equip the domain of atmospheric sciences with a recipe for how to build proper benchmark datasets, a (nonexclusive) list of domain-specific challenges for machine learning is presented, and it is elaborated where and what benchmark datasets will be needed to tackle these challenges. We hope that the creation of benchmark datasets will help the machine learning efforts in atmospheric sciences to be more coherent, and, at the same time, target the efforts of machine learning scientists and experts of high-performance computing to the most imminent challenges in atmospheric sciences. We focus on benchmarks for atmospheric sciences (weather, climate, and air-quality applications). However, many aspects of this paper will also hold for other aspects of the Earth system sciences or are at least transferable.
Significance Statement
Machine learning is the study of computer algorithms that learn automatically from data. Atmospheric sciences have started to explore sophisticated machine learning techniques and the community is making rapid progress on the uptake of new methods for a large number of application areas. This paper provides a clear definition of so-called benchmark datasets for weather and climate applications that help to share data and machine learning solutions between research groups to reduce time spent in data processing, to generate synergies between groups, and to make tool developments more targeted and comparable. Furthermore, a list of benchmark datasets that will be needed to tackle important challenges for the use of machine learning in atmospheric sciences is provided.
Abstract
Benchmark datasets and benchmark problems have been a key aspect for the success of modern machine learning applications in many scientific domains. Consequently, an active discussion about benchmarks for applications of machine learning has also started in the atmospheric sciences. Such benchmarks allow for the comparison of machine learning tools and approaches in a quantitative way and enable a separation of concerns for domain and machine learning scientists. However, a clear definition of benchmark datasets for weather and climate applications is missing with the result that many domain scientists are confused. In this paper, we equip the domain of atmospheric sciences with a recipe for how to build proper benchmark datasets, a (nonexclusive) list of domain-specific challenges for machine learning is presented, and it is elaborated where and what benchmark datasets will be needed to tackle these challenges. We hope that the creation of benchmark datasets will help the machine learning efforts in atmospheric sciences to be more coherent, and, at the same time, target the efforts of machine learning scientists and experts of high-performance computing to the most imminent challenges in atmospheric sciences. We focus on benchmarks for atmospheric sciences (weather, climate, and air-quality applications). However, many aspects of this paper will also hold for other aspects of the Earth system sciences or are at least transferable.
Significance Statement
Machine learning is the study of computer algorithms that learn automatically from data. Atmospheric sciences have started to explore sophisticated machine learning techniques and the community is making rapid progress on the uptake of new methods for a large number of application areas. This paper provides a clear definition of so-called benchmark datasets for weather and climate applications that help to share data and machine learning solutions between research groups to reduce time spent in data processing, to generate synergies between groups, and to make tool developments more targeted and comparable. Furthermore, a list of benchmark datasets that will be needed to tackle important challenges for the use of machine learning in atmospheric sciences is provided.
