Automatic input variable selection for analog methods using genetic algorithms

2021 ◽  
Author(s):  
Pascal Horton ◽  
Olivia Martius
Keyword(s):  
Genetic Algorithms ◽  
Climate Models ◽  
Prediction Models ◽  
Weather Prediction ◽  
Precipitation Prediction ◽  
Great Consistency ◽  
Operational Forecasting ◽  
Target Date ◽  
New Variables ◽  
Selection Of

<p>Analog methods (AMs) are statistical downscaling methods often used for precipitation prediction in different contexts, such as operational forecasting, past climate reconstruction of climate change impact studies. It usually relies on predictors describing the atmospheric circulation and the moisture content of the atmosphere to sample similar meteorological situations in the past and establish a probabilistic forecast for a target date. AMs can be based on outputs from numerical weather prediction models in the context of operational forecasting or outputs from climate models in climatic applications.</p><p>AMs can be constituted of multiple predictors organized in different subsequent levels of analogy that refines the selection of similar situations. The development of such methods is usually a manual process where some predictors are assessed in different structures. As most AMs use multiple predictors, a comprehensive assessment of all combinations becomes quickly impossible. The selection of predictors in the application of the AM often builds on previous work and does not evolve much. However, the climate models providing the predictors evolve continuously and new variables might become relevant to be considered in AMs. Moreover, the best predictors might change from one region to another or for another predictand of interest. There is a need for a method to automatically explore potential variables for AMs and to extract the ones that are relevant for a predictand of interest.</p><p>We propose using genetic algorithms (GAs) to proceed to an automatic selection of the predictor variables along with all other parameters of the AM. We even let the GAs automatically pick the best analogy criteria, i.e. the metric that quantifies the analogy between two situations. The first test consisted of letting the GAs select the single best variable to predict daily precipitation for each of 25 selected catchments in Switzerland. The results showed great consistency in terms of spatial patterns and the underlying meteorological processes. Then, different structures were assessed by varying the number of levels of analogy and the number of variables per level. Finally, multiple optimizations were conducted on the 25 catchments to identify the 12 variables that provide the best prediction when considered together.</p>

Frontal Scale Air–Sea Interaction in High-Resolution Coupled Climate Models

2010 ◽  
Vol 23 (23) ◽  
pp. 6277-6291 ◽  
Author(s):  
Frank O. Bryan ◽  
Robert Tomas ◽  
John M. Dennis ◽  
Dudley B. Chelton ◽  
Norman G. Loeb ◽  
...  
Keyword(s):  
Boundary Layer ◽  
High Resolution ◽  
Climate Models ◽  
Prediction Models ◽  
Weather Prediction ◽  
Surface Wind ◽  
Coupled Model ◽  
Climate System Model ◽  
Surface Wind Stress ◽  
Coupled Climate Models

Abstract The emerging picture of frontal scale air–sea interaction derived from high-resolution satellite observations of surface winds and sea surface temperature (SST) provides a unique opportunity to test the fidelity of high-resolution coupled climate simulations. Initial analysis of the output of a suite of Community Climate System Model (CCSM) experiments indicates that characteristics of frontal scale ocean–atmosphere interaction, such as the positive correlation between SST and surface wind stress, are realistically captured only when the ocean component is eddy resolving. The strength of the coupling between SST and surface stress is weaker than observed, however, as has been found previously for numerical weather prediction models and other coupled climate models. The results are similar when the atmospheric component model grid resolution is doubled from 0.5° to 0.25°, an indication that shortcomings in the representation of subgrid scale atmospheric planetary boundary layer processes, rather than resolved scale processes, are responsible for the weakness of the coupling. In the coupled model solutions the response to mesoscale SST features is strongest in the atmospheric boundary layer, but there is a deeper reaching response of the atmospheric circulation apparent in free tropospheric clouds. This simulated response is shown to be consistent with satellite estimates of the relationship between mesoscale SST and all-sky albedo.

scholarly journals Southern Ocean Cloud and Aerosol data: a compilation of measurements from the 2018 Southern Ocean Ross Sea Marine Ecosystems and Environment voyage

2020 ◽  
Author(s):  
Stefanie Kremser ◽  
Mike Harvey ◽  
Peter Kuma ◽  
Sean Hartery ◽  
Alexia Saint-Macary ◽  
...  
Keyword(s):  
Southern Ocean ◽  
Climate Models ◽  
Dimethyl Sulfide ◽  
Prediction Models ◽  
Climate Model ◽  
Ross Sea ◽  
Evaluation Studies ◽  
Weather Prediction ◽  
Data Sets

Abstract. Due to its remote location and extreme weather conditions, atmospheric in situ measurements are rare in the Southern Ocean. As a result, aerosol-cloud interactions in this region are poorly understood and remain a major source of uncertainty in climate models. This, in turn, contributes substantially to persistent biases in climate model simulations, numerical weather prediction models and reanalyses. It has been shown in previous studies that in situ and ground-based remote sensing measurements across the Southern Ocean are critical for complementing satellite data sets due to the importance of boundary layer and low-level cloud processes. These processes are poorly sampled by satellite-based measurements which are typically obscured by near-continuous overlying cloud cover observed in this region. In this work we present a comprehensive set of ship-based aerosol and meteorological observations collected on the TAN1802 voyage of R/V Tangaroa across the Southern Ocean, from Wellington, New Zealand, to the Ross Sea, Antarctica. The voyage was carried out from 8 February to 21 March, 2018. Many distinct, but contemporaneous, data sets were collected throughout the voyage. The compiled data sets include measurements from a range of instruments, such as (i) meteorological conditions at the sea surface and profile measurements; (ii) the size and concentration of particles; (iii) trace gases dissolved in the ocean surface such as dimethyl sulfide and carbonyl sulfide; (iv) and remotely sensed observations of low clouds. Here, we describe the voyage, the instruments, data processing, and provide a brief overview of some of the data products available. We encourage the scientific community to use these measurements for further analysis and model evaluation studies, in particular, for studies of Southern Ocean clouds, aerosol and their interaction. The data sets presented in this study are publicly available at https://doi.org/10.5281/zenodo.4060237 (Kremser et al. 2020).

scholarly journals Weather Forecasting Using GPU-Based Large-Eddy Simulations

2015 ◽  
Vol 96 (5) ◽  
pp. 715-723 ◽  
Author(s):  
Jerôme Schalkwijk ◽  
Harmen J. J. Jonker ◽  
A. Pier Siebesma ◽  
Erik Van Meijgaard
Keyword(s):  
Climate Models ◽  
Prediction Models ◽  
Weather Forecasting ◽  
Multicore Processors ◽  
Weather Prediction ◽  
Model Quality ◽  
Numerical Weather ◽  
Eddy Simulation ◽  
Large Eddy ◽  
Computational Resources

Abstract Since the advent of computers midway through the twentieth century, computational resources have increased exponentially. It is likely they will continue to do so, especially when accounting for recent trends in multicore processors. History has shown that such an increase tends to directly lead to weather and climate models that readily exploit the extra resources, improving model quality and resolution. We show that Large-Eddy Simulation (LES) models that utilize modern, accelerated (e.g., by GPU or coprocessor), parallel hardware systems can now provide turbulence-resolving numerical weather forecasts over a region the size of the Netherlands at 100-m resolution. This approach has the potential to speed the development of turbulence-resolving numerical weather prediction models.

scholarly journals Validation of subgrid scale ensemble precipitation forecasts based on the ECMWF’s ecPoint Rainfall project

Időjárás ◽  
2021 ◽  
Vol 125 (3) ◽  
pp. 397-418
Author(s):  
Boglárka Tóth ◽  
István Ihász
Keyword(s):  
Extreme Precipitation ◽  
Prediction Models ◽  
Weather Prediction ◽  
Precipitation Forecast ◽  
Time Range ◽  
Weather Forecasts ◽  
Extreme Precipitation Events ◽  
Operational Forecasting ◽  
Processed Products ◽  
Short Time

Nowadays, state-of-the-art numerical weather prediction models successfully predict the general weather characteristics several days ahead, but forecasting extreme precipitation is quite challenging even in the short time range. In the framework of the ecPoint Project, the European Centre for Medium-Range Weather Forecasts (ECMWF) developed a new innovative probabilistic post-processing tool which produces 4-day precipitation forecast as accurate as the raw ensemble forecast at day 1. In the framework of the scientific co-operation between ECMWF and the Hungarian Meteorological Service (OMSZ), we were invited to participate in the validation of the experimental products. Quasi operational post-processed products have been available since July 1, 2018. During our work, besides using different verification technics, a new ensemble meteogram was also developed which can support operational forecasters during extreme precipitation events. As a result of our work, products of the ecPoint Project have been included in the operational forecasting activity.

scholarly journals Machine Learning for Climate Precipitation Prediction Modeling over South America

2021 ◽  
Vol 13 (13) ◽  
pp. 2468
Author(s):  
Juliana Aparecida Anochi ◽  
Vinícius Albuquerque de Almeida ◽  
Haroldo Fraga de Campos Velho
Keyword(s):  
Machine Learning ◽  
South America ◽  
Prediction Models ◽  
Weather Prediction ◽  
Data Availability ◽  
Learning Models ◽  
Precipitation Patterns ◽  
Precipitation Prediction ◽  
Computational Performance ◽  
Machine Learning Models

Many natural disasters in South America are linked to meteorological phenomena. Therefore, forecasting and monitoring climatic events are fundamental issues for society and various sectors of the economy. In the last decades, machine learning models have been developed to tackle different issues in society, but there is still a gap in applications to applied physics. Here, different machine learning models are evaluated for precipitation prediction over South America. Currently, numerical weather prediction models are unable to precisely reproduce the precipitation patterns in South America due to many factors such as the lack of region-specific parametrizations and data availability. The results are compared to the general circulation atmospheric model currently used operationally in the National Institute for Space Research (INPE: Instituto Nacional de Pesquisas Espaciais), Brazil. Machine learning models are able to produce predictions with errors under 2 mm in most of the continent in comparison to satellite-observed precipitation patterns for different climate seasons, and also outperform INPE’s model for some regions (e.g., reduction of errors from 8 to 2 mm in central South America in winter). Another advantage is the computational performance from machine learning models, running faster with much lower computer resources than models based on differential equations currently used in operational centers. Therefore, it is important to consider machine learning models for precipitation forecasts in operational centers as a way to improve forecast quality and to reduce computation costs.

scholarly journals Empirical values and assumptions in the convection of numerical models

2021 ◽  
Author(s):  
Anahí Villalba-Pradas ◽  
Francisco J. Tapiador
Keyword(s):  
Climate Models ◽  
Prediction Models ◽  
Numerical Models ◽  
Global Climate ◽  
Weather Prediction ◽  
Global Climate Models ◽  
Spatial And Temporal Scales ◽  
Weather Events ◽  
Wide Range ◽  
Convection Schemes

Abstract. Convection influences climate and weather events over a wide range of spatial and temporal scales. Therefore, accurate predictions of the time and location of convection and its development into severe weather are of great importance. Convection has to be parameterized in Numerical Weather Prediction models, Global Climate Models, and Earth System Models (NWPs, GCMs, and ESMs) as the key physical processes occur at scales much lower than the model grid size. The convection schemes described in the literature represent the physics by simplified models that require assumptions about the processes and the use of a number of parameters based on empirical values. The present paper examines these choices and their impacts on model outputs and emphasizes the importance of observations to improve our current understanding of the physics of convection.

scholarly journals Hurricane Juan (2003). Part II: Forecasting and Numerical Simulation

2006 ◽  
Vol 134 (7) ◽  
pp. 1748-1771 ◽  
Author(s):  
Ron McTaggart-Cowan ◽  
Lance F. Bosart ◽  
John R. Gyakum ◽  
Eyad H. Atallah
Keyword(s):  
Prediction Models ◽  
Weather Prediction ◽  
The United States ◽  
Mean Flow ◽  
Initial State ◽  
Near Surface ◽  
Steering Flow ◽  
Operational Forecasting ◽  
Model Set ◽  
Regional Forecasts

Abstract The landfall of Hurricane Juan (September 2003) in the Canadian Maritimes represents an ideal case in which to study the performance of operational forecasting of an intense, predominantly tropical feature entering the midlatitudes. A hybrid cyclone during its genesis phase, Juan underwent a tropical transition as it drifted slowly northward 1500 km from the east coast of the United States. Shortly after reaching its peak intensity as a category-2 hurricane, the storm accelerated rapidly northward and made landfall near Halifax, Nova Scotia, Canada, with maximum sustained winds of 44 m s−1. Although the forecasts and warnings produced by the U.S. National Hurricane Center and the Canadian Hurricane Centre were of high quality throughout Hurricane Juan’s life cycle, guidance from numerical weather prediction models became unreliable as the storm accelerated toward the coast. The short-range, near-surface forecasts from eight operational models during the crucial prelandfall portion of Juan’s track are investigated in this study. Despite continued improvements to operational numerical forecasting systems, it is shown that those systems not employing advanced tropical vortex initialization techniques were unable to provide forecasters with credible near-surface guidance in this case. A pair of regional forecasts, one successful and one from the failed model set, are compared in detail. Spurious asymmetries in the initial vortex of the deficient model are shown to hamper structural predictions and to cause nonnegligible track perturbations from the trajectory implied by the well-described deep-layer mean flow. The Canadian Mesoscale Compressible Community model is rerun with an improved representation of the hurricane’s vortex in the initial state. The hindcast produced following the tropical cyclone initialization contains reduced track, structure, and intensity errors compared with those generated by the model in real time. The enhanced initial intensity produces a direct improvement in the forecast storm strength throughout the period, and the symmetrization of the vortex eliminates the interactions that plague the operational system. The southeastward relocation of the implanted vortex to Juan’s observed location eliminates a significant northwestward track bias under the influence of a broad area of southerly steering flow. The study concludes that the initialization of Hurricane Juan’s structure and position adds value to numerical guidance even as the storm accelerates poleward at a latitude where the implantation of a quasi-symmetric vortex may not be generally valid.

Forecasting of Tornadoes and Downbursts

2020 ◽  
Author(s):  
Djordje Romanic
Keyword(s):  
Chaos Theory ◽  
Prediction Models ◽  
Weather Prediction ◽  
Small Scale ◽  
Weather Forecasts ◽  
Wind Speeds ◽  
The Past ◽  
Extreme Wind ◽  
Extreme Wind Speeds ◽  
Numerical Weather Prediction Models

Tornadoes and downbursts cause extreme wind speeds that often present a threat to human safety, structures, and the environment. While the accuracy of weather forecasts has increased manifold over the past several decades, the current numerical weather prediction models are still not capable of explicitly resolving tornadoes and small-scale downbursts in their operational applications. This chapter describes some of the physical (e.g., tornadogenesis and downburst formation), mathematical (e.g., chaos theory), and computational (e.g., grid resolution) challenges that meteorologists currently face in tornado and downburst forecasting.

Sign in / Sign up

Export Citation Format

Share Document