99524 - GUEST LECTURES AND SEMINARS ON CLIMATE SCIENCE

Conoscenze e abilità da conseguire

This course made in the format of seminars and guest lecturers will expose the student to the frontier of knowdledge of climate and will apprenhend what are the topics available for the final thesis. The student will be able to grasp what are the emerging areas on climate science and be able to select the topic for future deepening of the knowldge.

Contenuti

IMPORTANT:

- The Guest Lecture course lasts 2 semesters. In order to get it pass it is needed to attend them both.

- The exact schedule is given in the list of lectures below. Students are asked to check the program frequently given that it is usually updated in the course of the year based on availability of speakers. Therefore DO NOT only look at the time schedule but check the actual program below.

- In some, sporadic, instances the lectures will be also given online and the students will be allowed to attend remotely.

 

Structure:

The course is structured with 1- or 2-hours long time slots and with three types of offers:

1) Seminars: >=1 hour on current research/technological challenges, delivered by specialist.

2) Lecture: >=2 hours on a more general topic of broader relevance and less technical details.

3) Short course: >=3 hours on an additional supplementary skill. Examples may include a focus on programming or on an area of transversal interest.

The program is published in the course of the semesters and then constantly updated based on the availability of external guest lectures and the coherence of the content.

PROGRAM

 

24 Oct 16.00-18.00

1st SEMINAR

Title: Modeling the Unpredictable: Why Nat Cat Models Matter for Insurance and Resilience

Speakers: Andrea Leoncini - Unipol Assicurazioni S.p.A., Bologna, Italy

Gianluca Mussetti - Unipol Assicurazioni S.p.A., Bologna, Italy

Abstract:

It is well known that the Law of Large Numbers breaks down in the presence of heavy-tailed probability distributions, where extreme events disproportionately drive aggregate outcomes. Catastrophe insurance is shaped by rare and severe natural events that challenge conventional actuarial models; as a result, pricing and risk management tools based on the Law of Large Numbers and the Central Limit Theorem often prove inadequate. Natural Catastrophe (Nat Cat) models have therefore become indispensable for insurers, reinsurers, and policymakers in assessing the financial risks of hazards such as storms, floods, and earthquakes. By integrating hazard science, exposure data, and vulnerability analysis, these models inform premium setting, capital allocation, and regulatory compliance. Professionals in Nat Cat modelling—spanning atmospheric scientists, engineers, and actuaries—play a vital role in bridging research and practice, improving model accuracy, and addressing uncertainties, particularly in the context of climate change.

2nd SEMINAR

Title: Unipol research projects: near-term changes in rainfall extremes and severe thunderstorms in Italy

Speakers:

Simone Persiano - Unipol Assicurazioni S.p.A., Bologna, Italy

Lorenzo Aiazzi - Unipol Assicurazioni S.p.A., Bologna, Italy

Abstract:

As reported by the European Environment Agency (EEA, 2024), economic losses due to weather- and climate-related extreme events have increased over time. As natural extreme events are expected to intensify further, there is a growing interest in the insurance industry in modeling hydrological hazards (floods) and meteorological hazards (storms) and understanding how they are expected to change in the future.

A considerable share of the economic losses in Europe is associated with floods caused by extreme rainfall events, that have been steadily increasing over recent decades, mainly driven by climate change and the increase in exposure to flooding due to land-use and land-cover modifications. Urban areas are particularly vulnerable to the impacts of these events and, with their high spatial concentration of assets, represent a key concern for the insurance industry. Thus, a thorough analysis of changes in the magnitude of daily and sub-daily rainfall extremes is crucial: gridded rainfall datasets from reanalysis products can help identify hotspots and inform targeted insurance strategies.

When it comes to severe thunderstorms, a change in their frequency and intensity has been observed in recent years, with a significant increase of hail across most of Europe. Northern Italy represents a hotspot for increasing hail hazard (Battaglioli et al., 2023), within an area characterized by high urbanization and thus of great interest from an insurance perspective. In this context, we aim to assess changes in thunderstorm-related hazard in the near-term (i.e., over the next 5–10 years) in Italy. To achieve this goal, we are planning to make use of seasonal and decadal predictions.

07 Nov 16.00-18.00

Title: Advancing Global Ocean Reanalysis and Forecast Systems: Convolutional Neural Network and the Assimilation of Diurnal Satellite Retrievals of Sea Surface Temperature

Speaker: Matteo Broccoli (CMCC, Italy)

Abstract: A variety of SST datasets have been produced in the satellite era. Each one of them has almost unique characteristics and differ from the SST variable of ocean general circulation models, so that assimilating such datasets in an optimal way requires a mapping to the first model level. However, this projection is non-trivial as it depends on the specific characteristics of the dataset, as well as on surface physical processes that impact the (sub)skin SST measured from satellite. Therefore, proper assimilation requires dataset-dependent operators.

In this talk we present a data-driven approach to construct the projection operator with machine learning (ML). The benefit is that it can be used over different SST dataset by only re-training the network. We compare conventional methodologies (climatology, linear regression) and ML networks with increasing complexity (random forest, U-Net, pix2pix) trained with L3 global diurnal subskin SST derived from AVHRR’s infrared channels on MetOp satellites to reproduce the ESA SST CCI product as target. Conventional approaches are able to correct the first momentum (bias) but fail in correcting the second momentum (RMSE) in a distribution sense. The pix2pix is the most effective operator and we test it over several one-year-long reanalysis-like experiments at 1/4° using different ML-assisted approaches for SST assimilation. The ML-based observation operator is beneficial mainly in the tropics but degrades in the areas of strong mesoscale activities that probably require higher resolution or longer training. The ML-unbiased approach instead improves globally the subsurface vertical temperature RMSE up to 20% against independent in situ data, when SST is the only variable assimilated. In a realistic set-up, assimilating also in situ observation and Sea Level Anomaly retrievals, the global improvement is about 10%.

Finally, we discuss work in progress to include such operator in the data assimilation system of CMCC’s Global Ocean Forecast System (GOFS16). GOFS16 is a short-term prediction system (6-day) based on a global eddy-resolving configuration at 1/16° resolution, whose purpose is to represent the full dynamics and life cycle of baroclinic eddies in most of the global ocean. Currently, the assimilation system ingests multiple data sources (in situ, altimetry and SST data) on the same model grid, and in particular the L3 subskin SST data are placed directly on the first model level. The application of the ML operator is meant to improve the skill of the system by reducing the error, especially within the model’s surface layers.

14 Nov 16.00-18.00

Title: TBC

Speaker: Chiara Marsigli (ARPAE, Italy)

17 Nov 13.00-14.00

Title: Advances in Seasonal Sea Ice Forecasting: Dynamical Models, Machine Learning, and Applications

Speaker: Yiguo Wang (NERSC, Norway)

Abstract:

Sea ice plays a critical role in the Earth’s climate system, influencing heat exchange, ocean circulation, and ecosystems in polar regions. Its rapid decline under global warming not only signals profound climate change but also impacts human activities such as shipping, fisheries, and offshore operations. Reliable seasonal sea ice prediction has therefore become an essential scientific and societal challenge. In this lecture, I will introduce the fundamentals of sea ice, including its key physical properties, variability, and the observed impacts of climate change on Arctic and Antarctic sea ice cover. We will then explore the importance of seasonal sea ice forecasting and its applications, e.g., supporting Arctic shipping routes. The lecture will present an overview of the main approaches to sea ice prediction, including physics-based dynamical models and emerging machine learning (ML) techniques. As a concrete example, I will present the Norwegian Climate Prediction Model (NorCPM) and its application to Arctic and Antarctic sea ice prediction. Also, we will highlight recent advances in ML-based post-processing, focusing on error-correction methods that enhance the skill of dynamical forecasts.

21 Nov 16.00-18.00

Title: Climate Prediction: advances and challenges

Speaker: Noel Keenlyside (U of Bergen, Norway)

Abstract:

 

24 Nov 13.00-14.00

Title: Expanding CMCC Seasonal Prediction System v3.5 applications to the local scale through statistical downscaling techniques

Speakers:

Leonardo Aragao (CMCC, Italy)

Abstract:

The Italian Peninsula’s climate is highly influenced by its complex topography and diverse regional weather systems, making high-resolution seasonal forecasting crucial for agriculture, water management, and energy sectors. Traditional seasonal prediction models, such as the CMCC Seasonal Prediction System v3.5 (SPS), provide valuable insights but lack the spatial resolution necessary to capture local-scale climatic details. Recent advances in statistical downscaling show promise for enhancing these coarse-resolution forecasts by generating more localised and accurate predictions. Thus, this study aims to provide a high-resolution seasonal forecast for the Italian Peninsula by enhancing the SPS model through statistical downscaling techniques tailored to the region’s demand for finer-scale climate information. The downscaling method involves a three-step process that utilises observational datasets and machine-learning techniques to refine SPS forecasts. The first regards the ground truth, composed of high-resolution observational data from ERA5 reanalysis for 2m temperature, sea surface temperature, and 10m wind components, and from CHIRPS for precipitation. Then, SPS daily predictions are spatially interpolated from 1° to 1/4° to match the observational data's grid. Finally, both datasets are combined through two different machine-learning methods based on the Empirical Quantile Mapping (EQM) and the k-Nearest Neighbours (kNN) technique, translating SPS into high-resolution fields by matching predicted conditions to observed patterns. Finally, both statistical downscaling methods were assessed over the Italian Peninsula domain through cross-validation along the 24-year hindcast period available for SPS (1993-2016). The results indicate that statistical downscaling significantly enhances seasonal predictions for the Italian Peninsula, achieving biases about 5-6 times smaller than the original SPS for all evaluated predictands. The main component of this improvement is the spatial accuracy promoted by downscaling, allowing the identification of domain characteristics unnoticed in SPS. Even though the statistical indices show appreciable values for the domain as a whole, when we evaluate smaller portions of this same domain, the original seasonal predictions are still far from the desired. As expected, bias increases with lead time also for EQM and kNN, with accuracy declining progressively from lead month 1 onward. For example, 2m temperature bias increased from -0.14/-0.31/-0.85°C in lead month 1 to -0.68/-0.71/-1.41°C in lead month 6 (kNN/EQM/SPS). This trend highlights the ongoing challenge of maintaining prediction skills over extended periods and the importance of adaptive correction strategies to extend lead-time reliability. Integrating statistical downscaling techniques with SPS outputs provides a promising solution for generating high-resolution seasonal predictions, offering valuable support for climate-sensitive applications by reducing forecast bias and enhancing spatial accuracy. This work demonstrates the potential of statistical downscaling as an effective tool for bridging the gap between coarse seasonal predictions and the localised weather information necessary for effective decision-making.

 

 

28 Nov 16.00-18.00

Title: AI-enhanced climate prediction of extremes

Speaker: Leone Cavicchia (CMCC, Italy)

Abstract

As climate records are being surpassed every year, there is a growing demand from different stakeholders for accessible and reliable short-term (seasonal to multi-annual) climate predictions of extremes. Despite major advances in the development of numerical Earth system models and the increase of available computer power, the prediction of extreme events at (sub)seasonal and longer time scales remains challenging. In recent developments, AI and machine learning techniques are bring increasingly used to improve the accuracy of predictions. In this talk, I will first review the state of the art of numerical climate predictions systems and their limitations. I will then show through selected examples how AI techniques can be exploited to increase the skill of climate predictions – in particular by exploiting large observational datasets to train algorithms to detect connections between the extreme events and their large-scale drivers.

05 Dec 16.00-18.00

Titolo: Operational numerical weather prediction with the ICON model at Agenzia Italia Meteo and Arpae Emilia-Romagna

Speaker: Thomas Gastaldo (Italia Meteo, Italy)

Abstract:
This seminar will present the operational numerical weather prediction system implemented by Agenzia Italia Meteo and Arpae Emilia-Romagna, centered on the ICON-2I model. The talk will outline the collaborative framework with European meteorological centers and describe the end-to-end operational workflow, including the KENDA data assimilation system based on the LETKF scheme, the Rapid Update Cycle (RUC), and the Ensemble Prediction System (EPS). Key aspects of forecast verification and product generation will be discussed, emphasizing their importance for forecast reliability and practical usability. Finally, the presentation will explore cascading applications and downstream modeling, illustrating how these tools contribute to improved forecast accuracy and operational decision-making.

15 Dec 16.00-18.00

Title: TBC

Speaker:

Federico Grazzini (ARPAE, Italy)

Modalità di verifica e valutazione dell'apprendimento

At the end of both semesters the students will edit a short report on 3 lectures at their choice and send it for evaluation.

A mandatory attendance of at least 70% is required.

Strumenti a supporto della didattica

Students with learning disabilities (LD) or temporary or permanent disabilities: please contact the relevant University office promptly (https://site.unibo.it/studenti-con-disabilita-e-dsa/it). They will be responsible for suggesting any adjustment to the students concerned. However, these adjustments must be submitted to the instructor for approval 15 days in advance, who will evaluate their suitability also in relation to the educational objectives of the course.

Orario di ricevimento

Consulta il sito web di Natale Alberto Carrassi