Tuesday 26 September 15.30-17.15 CEST

Location: SISSA Main Auditorium

Session co-chairs: Fatima Driouech, Jason Evans and Douglas Maraun
Rapporteurs: Peace Olubukunmi Awoleye and AJ Komkoua Mbienda 

Empirical statistical downscaling (ESD) methods follow a wide range of approaches (PP, MOS and bias adjustment, weather generators) and techniques (analogs, regression, machine learning). . These are applied individually or in combination to produce regional/local information from global or regional model outputs. Hybrid RCM-ESD methods (such as emulators) are an emergent topic boosted by developments in deep machine learning. Some key questions:

Can we evaluate the suitability and added value of statistical downscaling methods for climate change applications? Can we produce comprehensive intercomparisons to understand their benefits and limitations? Can we build RCM emulators suitable for certain tasks (e.g. emulating CPRCM runs or producing large ensembles)? How do we solve distillation issues from ESD/RCM/GCM ensembles?

Session report

15.30-15.45 CEST Invited speaker

Advancing Convection-Permitting Climate Projections: Coordinated Ensemble Experiments and the Path Ahead José Manuel GUTIERREZ

15.45-17.15 CEST Oral presentations (10+2 min)

A storyline approach to select the CMIP6 model ensemble to be downscaled for the South America domain Andressa ANDRADE CARDOSO  
Convolutional neural networks for local climate downscaling: precipitation extremes in the FPS in Southeastern South America Maria Laura BETTOLLI
RCM-emulators: A study of applicability to large GCM ensembles Antoine DOURY 
Should we bias correct boundary conditions for regional climate models? Jason EVANS
Introducing eXplainable Artificial Intelligence to assess Deep Learning Models for Statistical Downscaling Jose GONZÁLEZ-ABAD
Can deep-learning models extrapolate to downscaling rainfall in future climates? Neelesh RAMPAL
Design of Experiments and Machine Learning (DoE & ML)-based approach to better capture uncertainty in future climate projections Carla VIVACQUA 

Due to unforeseen circumstances some speakers may be replaced or will not be able to participate.