Model error in weather and climate forecasting

Author(s):  
Myles Allen ◽  
David Frame ◽  
Jamie Kettleborough ◽  
David Stainforth
Author(s):  
Julia Slingo ◽  
Tim Palmer

Following Lorenz's seminal work on chaos theory in the 1960s, probabilistic approaches to prediction have come to dominate the science of weather and climate forecasting. This paper gives a perspective on Lorenz's work and how it has influenced the ways in which we seek to represent uncertainty in forecasts on all lead times from hours to decades. It looks at how model uncertainty has been represented in probabilistic prediction systems and considers the challenges posed by a changing climate. Finally, the paper considers how the uncertainty in projections of climate change can be addressed to deliver more reliable and confident assessments that support decision-making on adaptation and mitigation.


2019 ◽  
Vol 147 (2) ◽  
pp. 645-655 ◽  
Author(s):  
Matthew Chantry ◽  
Tobias Thornes ◽  
Tim Palmer ◽  
Peter Düben

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.


2013 ◽  
Vol 20 (2) ◽  
pp. 199-206
Author(s):  
I. Trpevski ◽  
L. Basnarkov ◽  
D. Smilkov ◽  
L. Kocarev

Abstract. Contemporary tools for reducing model error in weather and climate forecasting models include empirical correction techniques. In this paper we explore the use of such techniques on low-order atmospheric models. We first present an iterative linear regression method for model correction that works efficiently when the reference truth is sampled at large time intervals, which is typical for real world applications. Furthermore we investigate two recently proposed empirical correction techniques on Lorenz models with constant forcing while the reference truth is given by a Lorenz system driven with chaotic forcing. Both methods indicate that the largest increase in predictability comes from correction terms that are close to the average value of the chaotic forcing.


2018 ◽  
Vol 99 (12) ◽  
pp. 2519-2527 ◽  
Author(s):  
Daphne S. LaDue ◽  
Ariel E. Cohen

AbstractProfessional meteorologists gain a great deal of knowledge through formal education, but two factors require ongoing learning throughout a career: professionals must apply their learning to the specific subdiscipline they practice, and the knowledge and technology they rely on becomes outdated over time. It is thus inherent in professional practice that much of the learning is more or less self-directed. While these principles apply to any aspect of meteorology, this paper applies concepts to weather and climate forecasting, for which a range of resources, from many to few, for learning exist. No matter what the subdiscipline, the responsibility for identifying and pursuing opportunities for professional, lifelong learning falls to the members of the subdiscipline. Thus, it is critical that meteorologists periodically assess their ongoing learning needs and develop the ability to reflectively practice. The construct of self-directed learning and how it has been implemented in similar professions provide visions for how individual meteorologists can pursue—and how the profession can facilitate—the ongoing, self-directed learning efforts of meteorologists.


2019 ◽  
Vol 12 (7) ◽  
pp. 2797-2809 ◽  
Author(s):  
Sebastian Scher ◽  
Gabriele Messori

Abstract. Recently, there has been growing interest in the possibility of using neural networks for both weather forecasting and the generation of climate datasets. We use a bottom–up approach for assessing whether it should, in principle, be possible to do this. We use the relatively simple general circulation models (GCMs) PUMA and PLASIM as a simplified reality on which we train deep neural networks, which we then use for predicting the model weather at lead times of a few days. We specifically assess how the complexity of the climate model affects the neural network's forecast skill and how dependent the skill is on the length of the provided training period. Additionally, we show that using the neural networks to reproduce the climate of general circulation models including a seasonal cycle remains challenging – in contrast to earlier promising results on a model without seasonal cycle.


2007 ◽  
Vol 88 (7) ◽  
pp. 1033-1044 ◽  
Author(s):  
KAREN PENNESI

One goal of weather and climate forecasting is to inform decision making. Effective communication of forecasts to various sectors of the public is essential for meeting that goal, yet studies repeatedly show that forecasts are not well understood by lay people. Using a case study from northeast Brazil, this article discusses some of the communication difficulties faced by forecasters and outlines an approach for adapting forecast language to users' needs and expectations. Analysis is based on data collected during 14 months of fieldwork, including interviews, a survey, and observations of meteorologists and local “rain prophets,” whose predictions are derived from empirical observations. The anthropological approach emphasizes the importance of language. For example, findings indicate that forecast communicators should look for multiple definitions of key terms that have common as well as technical meanings. Distinctions salient to meteorologists may be meaningless to the public, even when terms are clearly defined. In some cases, it maybe more helpful to work with lay concepts when communicating forecasts rather than dismissing such understandings as “incorrect.” Meteorologists should also recognize that scientific concepts are not accepted by everyone as the only correct way to think. This is especially relevant where scientific forecasts are competing with alternatives, such as those based on traditional knowledge. Finally, forecast communicators should develop the format and content of the forecast within each application. It is important to learn what people expect from forecasts and which communication styles are preferred.


2014 ◽  
Vol 1 (1) ◽  
pp. 131-153 ◽  
Author(s):  
O. Martínez-Alvarado

Abstract. Numerical climate models constitute the best available tools to tackle the problem of climate prediction. Two assumptions lie at the heart of their suitability: (1) a climate attractor exists, and (2) the numerical climate model's attractor lies on the actual climate attractor, or at least on the projection of the climate attractor on the model's phase space. In this contribution, the Lorenz '63 system is used both as a prototype system and as an imperfect model to investigate the implications of the second assumption. By comparing results drawn from the Lorenz '63 system and from numerical weather and climate models, the implications of using imperfect models for the prediction of weather and climate are discussed. It is shown that the imperfect model's orbit and the system's orbit are essentially different, purely due to model error and not to sensitivity to initial conditions. Furthermore, if a model is a perfect model, then the attractor, reconstructed by sampling a collection of initialised model orbits (forecast orbits), will be invariant to forecast lead time. This conclusion provides an alternative method for the assessment of climate models.


2015 ◽  
Vol 96 (5) ◽  
pp. 737-753 ◽  
Author(s):  
Wassila M. Thiaw ◽  
Vadlamani B. Kumar

Abstract Drought is one of the leading causes of death in Africa because of its impact on access to sanitary water and food. This challenge has mobilized the international community to develop famine early warning systems (FEWS) to bring safe food and water to populations in need. Over the past several decades, much attention has focused on advance risk planning in agriculture and water and, more recently, on health. These initiatives require updates of weather and climate outlooks. This paper describes the active role of NOAA’s African Desk in FEWS and in enhancing the capacity of African institutions to improve forecasts. The African Desk was established in 1994 to provide services to U.S. agencies and African institutions. Emphasis is on the operational products across all time scales from weather to climate forecasts in support of humanitarian relief programs. Tools to provide access to real-time weather and climate information to the public are described. These include the downscaling of the U.S. National Multimodel Ensemble (NMME) to improve seasonal forecasts. The subseasonal time scale has emerged as extremely important to many socioeconomic sectors. Drawing from advances in numerical models, operational subseasonal forecasts are included in the African Desk product suite. These capabilities along with forecast skill assessment, verifications, and regional hazards outlooks for food security are discussed. Finally, the African Desk residency training program, an effort aimed at enhancing the capacity of African institutions to improve forecasts, and supported by this seamless approach to operational forecasting, is described.


Climate ◽  
2019 ◽  
Vol 7 (1) ◽  
pp. 12 ◽  
Author(s):  
Oleg Onishchenko ◽  
Viktor Fedun ◽  
Wendell Horton ◽  
Oleg Pokhotelov ◽  
Gary Verth

According to modern concepts, the main natural sources of dust in the atmosphere are dust storms and associated dust devils—rotating columns of rising dust. The impact of dust and aerosols on climate change in the past, present and future is one of the poorly understood and, at the same time, one of the fundamental elements needed for weather and climate forecasting. The purpose of this review is to describe and summarise the results of the study of dust devils in the Earth’s atmosphere. Special attention is given to the description of the 3D structures, the external flows and atmospheric gradients of temperature that lead to the generation and maintenance of the dust devils.


Sign in / Sign up

Export Citation Format

Share Document