A Comprehensive Approach to Account for Weather Uncertainties in Ship Route Optimization

This work aims at defining in a probabilistic manner objectives and constraints typically considered in route optimization systems. Information about weather-related uncertainties is introduced by adopting ensemble forecast results. Classical reliability methods commonly used in structural analysis are adopted, allowing to achieve a simple yet effective evaluation of the probability of failure and the variability associated with the predicted fuel consumption and time of arrival. A quantitative example of application is provided, taking into consideration one of the main North Atlantic routes.

On the ROC Area of Ensemble Forecasts for Rare Events

The relative operating characteristic (ROC) curve is a popular diagnostic tool in forecast verification, with the area under the ROC curve (AUC) used as a verification metric measuring the discrimination ability of a forecast. Along with calibration, discrimination is deemed as a fundamental probabilistic forecast attribute. In particular, in ensemble forecast verification, AUC provides a basis for the comparison of potential predictive skill of competing forecasts. While this approach is straightforward when dealing with forecasts of common events (e.g. probability of precipitation), the AUC interpretation can turn out to be oversimplistic or misleading when focusing on rare events (e.g. precipitation exceeding some warning criterion). How should we interpret AUC of ensemble forecasts when focusing on rare events? How can changes in the way probability forecasts are derived from the ensemble forecast affect AUC results? How can we detect a genuine improvement in terms of predictive skill? Based on verification experiments, a critical eye is cast on the AUC interpretation to answer these questions. As well as the traditional trapezoidal approximation and the well-known bi-normal fitting model, we discuss a new approach which embraces the concept of imprecise probabilities and relies on the subdivision of the lowest ensemble probability category.

Development and Evaluation of the Aerosol Forecast Member in NCEP’s Global Ensemble Forecast System (GEFS-Aerosols v1)

NOAA’s National Weather Service (NWS) is on its way to deploy various operational prediction applications using the Unified Forecast System (https://ufscommunity.org/), a community-based coupled, comprehensive Earth modeling system. An aerosol model component developed in a collaboration between the Global Systems Laboratory, Chemical Science Laboratory, the Air Resources Laboratory, and Environmental Modeling Center (GSL, CSL, ARL, EMC) was coupled online with the FV3 Global Forecast System (FV3GFS) using the National Unified Operational Prediction Capability (NUOPC)-based NOAA Environmental Modeling System (NEMS) software framework. This aerosol prediction system replaced the NEMS GFS Aerosol Component (NGAC) system in the National Center for Environment Prediction (NCEP) production suite in September 2020 as one of the ensemble members of the Global Ensemble Forecast System (GEFS), dubbed GEFS-Aerosols v1. The aerosol component of atmospheric composition in GEFS is based on the Weather Research and Forecasting model (WRF-Chem) that was previously included into FIM-Chem (Zhang et al, 2021). GEFS-Aerosols includes bulk modules from the Goddard Chemistry Aerosol Radiation and Transport model (GOCART). Additionally, the biomass burning plume rise module from High-Resolution Rapid Refresh (HRRR)-Smoke was implemented; the GOCART dust scheme was replaced by the FENGSHA dust scheme (developed by ARL); the Blended Global Biomass Burning Emissions Product (GBBEPx V3) provides biomass burning emission and Fire Radiative Power (FRP) data; and the global anthropogenic emission inventories are derived from the Community Emissions Data System (CEDS). All sub-grid scale transport and deposition is handled inside the atmospheric physics routines, which required consistent implementation of positive definite tracer transport and wet scavenging in the physics parameterizations used by NCEP’s operational Global Forecast System based on FV3 (FV3GFS). This paper describes the details of GEFS-Aerosols model development and evaluation of real-time and retrospective runs using different observations from in situ measurement, satellite and aircraft data. GEFS-Aerosols predictions demonstrate substantial improvements for both composition and variability of aerosol distributions over those from the former operational NGAC system.

The Reanalysis for the Global Ensemble Forecast System, version 12.

NOAA has created a global reanalysis data set, intended primarily for initialization of reforecasts for its Global Ensemble Forecast System, version 12 (GEFSv12), which provides ensemble forecasts out to +35 days lead time. The reanalysis covers the period 2000-2019. It assimilates most of the observations that were assimilated into the operational data assimilation system used for initializing global predictions. These include a variety of conventional data, infrared and microwave radiances, Global Positioning System radio occultations, and more. The reanalysis quality is generally superior to that from NOAA’s previous-generation Climate Forecast System Reanalysis (CFSR), demonstrated in the fit of short-term forecasts to the observations and in the skill of 5-day deterministic forecasts initialized from CFSR vs. GEFSv12. Skills of reforecasts initialized from the new reanalyses are similar but slightly lower than skills initialized from a pre-operational version of the real-time data assimilation system conducted at the higher, operational resolution. Control member reanalysis data on vertical pressure levels are made publicly available.

A precipitation forecasting model using machine learning on big data in clouds environment

Numerical weather prediction (NWP) has long been a difficult task for meteorologists. Atmospheric dynamics is extremely complicated to model, and chaos theory teaches us that the mathematical equations used to predict the weather are sensitive to initial conditions; that is, slightly perturbed initial conditions could yield very different forecasts. Over the years, meteorologists have developed a number of different mathematical models for atmospheric dynamics, each making slightly different assumptions and simplifications, and hence each yielding different forecasts. It has been noted that each model has its strengths and weaknesses forecasting in different situations, and hence to improve performance, scientists now use an ensemble forecast consisting of different models and running those models with different initial conditions. This ensemble method uses statistical post-processing; usually linear regression. Recently, machine learning techniques have started to be applied to NWP. Studies of neural networks, logistic regression, and genetic algorithms have shown improvements over standard linear regression for precipitation prediction. Gagne et al proposed using multiple machine learning techniques to improve precipitation forecasting. They used Breiman’s random forest technique, which had previously been applied to other areas of meteorology. Performance was verified using Next Generation Weather Radar (NEXRAD) data. Instead of using an ensemble forecast, it discusses the usage of techniques pertaining to machine learning to improve the precipitation forecast. This paper is to present an approach for mapping of precipitation data. The project attempts to arrive at a machine learning method which is optimal and data driven for predicting precipitation levels that aids farmers thereby aiming to provide benefits to the agricultural domain.

Seasonal Predictions of Holopelagic Sargassum Across the Tropical Atlantic Accounting for Uncertainty in Drivers and Processes: The SARTRAC Ensemble Forecast System

The holopelagic macroalgae sargassum has proliferated across the tropical Atlantic since 2011, of consequence for coastal populations from West Africa to the Caribbean with limited early warning of major beaching events. As part of an interdisciplinary project, ‘Teleconnected SARgassum risks across the Atlantic: building capacity for TRansformational Adaptation in the Caribbean and West Africa’ (SARTRAC), an ensemble forecast system, SARTRAC-EFS, is providing seasonal predictions of sargassum drift. An eddy-resolving ocean model hindcast provides the winds and currents necessary to generate ensemble members. Ensemble forecasts are then obtained for different combinations of ‘windage’, the fractional influence of winds on sargassum mats, and in situ rates of growth, mortality, and sinking. Forecasts for north and south of Jamaica are evaluated with satellite-observed distributions, associated with beaching events in specific years of heavy inundation, 2015 and 2018-20. These seasonal forecasts are evaluated, on lead times of up to 180 days. Forecasts are subject to leading modes of tropical climate variability, in particular the Atlantic Meridional Mode (AMM). More accurate forecasts for a given year are obtained with ensemble members from hindcast years with a similar spring AMM-index. This is most clearly evident during negative AMM phases in spring of 2015 and 2018, when positive sea surface temperature anomalies and anomalously weak trade winds were established across the northern tropics. On this evidence, SARTRAC-EFS is potentially useful in providing early warning of high sargassum prevalence. Extended to sargassum drift off West Africa, extensive cloud cover limits availability of the satellite data needed for full application and evaluation of SARTRAC-EFS in this region, although experimental forecasts off the coast of Ghana are found highly sensitive to the windage that is associated with strong onshore winds during boreal summer. Alongside other forecast systems, SARTRAC-EFS is providing useful early warnings of sargassum inundation at seasonal timescale.