Statistical up/downscaling

Welcome to the SAMA group's Up/Downscaling page.
This theme is coordinated by P.A. Michelangeli (P.-A. Michelangeli email).

Atmospheric or oceanic general circulation models, not even speaking of climate simulations, have spatial resolution that is generally too low for a good local representation of the atmosphere or the ocean. Therefore, downscaling methods are needed.

Dynamical downscaling is an option, with tools like Limited Area Models (MM5, WRF, Météo-France's Méso-nh, ...) that are forced at their lateral boundaries by low resolution fields (outputs from GCM's, analysis). Using LAMs is made difficult by (i) the need of accurate vegetation/land-use and topography data, (ii) the understanding and tuning of subgrid parameterizations schemes and (iii) the availability of numerical resources (particularly for climate downscaling).

An other way to downscale is by using statistical models built on relations between large scale and local climate. Shifting from one spatial resolution to another is not only from low to high resolution but also from high to low resolution, this is the upscaling. Upscaling methods use local data, proxies (ice cores, tree rings, ...) to reconstruct large scale climate evolution.

The SAMA's up/downscaling group is focused on understanding, developing and using statistical upscaling and downscaling tools. This page is dedicated to this research field at IPSL and is a link toward the up/downscaling scientific community.







Latest news

up arrow






Seminars, workshops, conferences

up arrow






Links

up arrow

up arrow