seminari corsi meteorologia
In partenza il corso “Machine learning in weather and climate” dell’ECMWF

Sono aperte le registrazioni per partecipare al corso online Machine learning on weather and climate promosso da ECMWF (European Centre for Medium-Range Weather Forecasts) in collaborazione con IFAB (International Foundation on Big Data and Artificial Intelligence for Human Development) che partirà il 9 gennaio.

Il corso, completamente gratuito, è rivolto a chiunque sia interessato a conoscere i recenti sviluppi delle tecniche di machine learning e come queste possono essere applicate alle previsioni meteorologiche e climatiche. L’unico requisito richiesto è una conoscenza di base di meteorologia, statistica e programmazione.

Il corso si articola in tre moduli:

  • un’introduzione al machine learning in ambito meteorologico e climatico
  • un approfondimento sui concetti di machine learning
  • applicazioni pratiche

È possibile seguire uno o più moduli liberamente, al completamento di ognuno dei quali verrà rilasciato un certificato di partecipazione.

Le lezioni si svolgono nell’arco di 10 settimane, per un totale di 36 ore di formazione.

Ulteriori informazioni sul corso si trovano alla pagina https://lms.ecmwf.int/pages/index.html, da cui è anche possibile registrarsi.

opportunità di lavoro
ECMWF: Posizione di dottorato presso il centro Europeo

Posizione di dottorato presso il centro Europeo (ECMWF)

Segnaliamo una posizione PhD full time disponibile al centro Europeo ECMWF (European Centre for Medium-Range Weather Forecasts) nel dipartimento di Matematica e Statistica presso l’University of Reading dal titolo: “Mathematics: Randomized parallel algorithms for data assimilation in numerical weather prediction“, con scandenza il 20 Gennaio 2023.

Supervisors:  Jennifer Scott (UoR), Amos Lawless (UoR), Massimo Bonavita (ECMWF), Nicolas Bousserez (ECMWF).

Maggiori informazioni sulla domanda di ammissione sono disponibili al seguente link.

seminari corsi meteorologia
Seminario di Meteorologia Ambientale – Fatima Pillosu

Per la serie “Environmental Meteorology Seminar”, vi segnaliamo il seminario di Fatima Pillosu (European Centre for Medium-range Weather Forecasts and University of Reading, Reading, UK) dal titolo: “ecPoint, post-processing of global ensemble rainfall forecasts: current efforts and future challenges to improve the prediction of extreme localized rainfall events and flash floods at global scale”.

Il seminario si terrà Giovedì 20 Ottobre 2022 alle ore 14:30, online attraverso la piattaforma Zoom (https://unitn.zoom.us/j/85796377450 – Meeting ID: 857 9637 7450, Passcode: 182580) e in presenza sarà riprodotto nell’Aula 2Q del Dipartimento di Ingegneria Civile, Ambientale e Meccanica (DICAM) presso l’Università degli Studi di Trento, in Via Mesiano 77 (38123, Trento).

Vi lasciamo un abstract e la biografia di Fatima Pillosu in basso:

Abstract
ECMWF has always been at the forefront of numerical weather prediction
(NWP) model development, often ranking as one of the world’s leading
centres for weather prediction. ECMWF’s Integrated Forecasting System
(IFS) covers the global domain, different time horizons (e.g.,
medium-range, sub-seasonal and seasonal), and consists of several
components (e.g., the atmospheric general circulation model, the ocean
wave model, the land-surface model, and the perturbation models for data
assimilation and generation of forecast ensembles). Any NWP model
divides the Earth’s surface into grid boxes and predicts one value
(e.g., for rainfall or temperature) per grid box. The sizes of such
grid-boxes vary depending on the model, from 1 to 5 km for regional
high-resolution models to 10 to 50 km for global lower-resolution
models. At ECMWF, great efforts are made to consistently increase the
resolution of their NWP models to provide their users with better
forecasts for specific locations, especially for extreme events (e.g.,
wind gusts, convective storms, and flash floods). Currently, the global
ensemble forecasts (ENS) are provided at 18 km and are expected to go to
9 km next year. However, if weather varies markedly within a grid box or
the predictability of the atmosphere is low, forecasts for specific
sites will inevitably fail due to biases or representativeness errors in
the model. By post-processing the raw NWP forecasts, it is possible to
provide better predictions for specific locations. ecPoint has been one
of the first post-processing projects at ECMWF, mainly to anticipate
sub-grid variability and biases in rainfall forecasts. Currently,
ecPoint provides global point-scale rainfall forecasts up to day 10 to
ECMWF users worldwide. Furthermore, ongoing investigations are carried
out on how to use ecPoint products to predict one of the most
devastating natural hazards: flash floods. The seminar will provide an
overview of the ecPoint post-processing technique. It will also discuss
the research behind the definition of a global layer in the GloFAS
platform that will use ecPoint forecasts for the global prediction of
flash floods at medium ranges, which hopes to help mainly humanitarian
actions in developing countries.

Bio
Fatima Pillosu is a researcher at the European Centre for Medium-range
Weather Forecasts (ECMWF, Reading, UK) and a PhD student at the
University of Reading (Reading, UK). After obtaining a bachelor’s
degree in civil engineering at the University of Cagliari, she
specialised in hydrology through an MSc in Hydraulic Engineering at the
University of Cagliari with a thesis that analysed the change in
frequency of extreme rainfall events and flash floods in Sardinia. This
latter work encouraged her to gain experience in meteorological science,
which led her to a 1-year internship at ECMWF in 2016. The project
focused on developing an innovative post-processing technique (called
ecPoint) to determine the degree of sub-grid variability and biases in
short- to medium-range rainfall forecasts. In 2019, this research
project became operational at ECMWF, providing global point-scale
rainfall forecasts to ECMWF users worldwide. In 2017, she also started a
PhD at the University of Reading to investigate the utility of such new
post-processed rainfall forecasts for hydrological predictions, focusing
mainly on flash flood forecasting.

This series of seminars is primarily targeted to Students attending our
double-degree programme of MSc in Environmental Meteorology [1].
However all those who are interested are more than welcome to join!
Workshop e Congressi
Hackaton event, ECMWF

ECMWF organizza un Hackaton event dal titolo: Hackathon 2022: Visualising Meteorological Data

in Reading, UK, l’11 e 12 giugno 2022

Per maggiori info:

The aim of Hackathon 2022 is to explore how meteorological data, weather
and climate, can be visualised to be more useable, understandable and
impactful for users and the broader public.

This is not your usual Hackathon. We are looking for a wide range of
experience to help with the challenges, ideas and projects. If you are a
coder, designer, data wrangler, meteorologist, storyteller,
journalist…or just have an interest in meteorological data and
visualisation this is the event for you.

We are in the process of establishing challenges which will shape the
ideas and projects for Hackathon 2022. There will also be an Open
Challenge for those who want to work on something different. A GitHub
page [1] has been set up for the event to facilitate pre-event idea
discussions and team building.

A variety of data and ECMWF experts will be provided throughout the
weekend to help with your ideas and develop your projects.

The event is the weekend following the Using ECMWF Forecasts (UEF2022)
[2] event on the same theme and those attending this event are
encouraged to join in with Hackathon 2022.

This event will be held in-person at ECMWF in Reading (UK) (free to
attend but registration is needed).

All details and Registration at https://events.ecmwf.int/event/305/ [3]

We hope you can join us!

ECMWF Support

#VisMetHack @ECMWF
seminari corsi meteorologia
Weather and Climate Seminar: Francesca Di Giuseppe

Martedì 29 marzo alle ore 15.30 si terrà via Zoom il terzo seminario della serie “Weather and Climate: From Fundamentals to Applications”. Maggiori informazioni sull’iniziativa, il modulo di pre-registrazione Zoom, ed il programma completo a questo link.

Il seminario di giovedì prossimo, dal titolo How Good are We at Predicting Fires?, sarà ospitato dall’Università dell’Aquila e tenuto da Francesca di Giuseppe (ECMWF).

Abstract: The prediction of fire danger conditions allows fire management agencies to implement fire prevention, detection and pre-suppression action plans before fire damages occur. However, in many countries fire danger rating relies on observed weather data, which only allows for daily environmental monitoring of fire conditions. Even when this estimation is enhanced with the combined use of satellite data, such as hot spots for early fire detection and land cover and fuel conditions, it normally only provides 4 to 6 h warnings. By using forecast conditions from advanced numerical weather models, early warning could be extended by up to 1–2 weeks, allowing for greater coordination of resource-sharing and mobilization within and across countries. 
Using 1 year of pre-operational service in 2017 and the Fire Weather Index (FWI), in this talk we assess the capability of the system to predict fire danger globally and analyse in detail three major events in Chile, Portugal and California. The analysis shows that the skill provided by an ensemble forecast system extends to more than 10 days when compared to the use of mean climate, making a case for extending the forecast range to the sub-seasonal to seasonal timescale. However, accurate FWI prediction does not translate into accuracy in the forecast of fire activity globally. Indeed, when all fires detected in 2017 are considered, including agricultural- and human-induced burning, high FWI values only occur in 50 % of the cases and are limited to the Boreal regions. Nevertheless, for very large events which were driven by weather conditions, FWI forecasts provide advance warning that could be instrumental in setting up management and containment strategies.