Hydrometeorological downscaling and bias correction techniques

February 5, 2018 – February 7, 2018 all-day
Statkraft AS, Oslo, Norway

Course responsible: Prof. Asgeir Sorteberg, Geophysical Institute, UiB and Prof. John Burkhart, Department of Geosciences, UiO

Instructors: Rasmus Benestad, MET Norway and Ethan Gutmann, National Center for Atmospheric Research (NCAR)

Target Group: Doctoral students with hydrological, meteorological, or climatological background with an interest in regional atmospheric and hydrological downscaling and bias correction methods.

Credits: 1 ECT credits.

No. of participants: Max 15 students; 7 CHESS PhD students and 8 non-CHESS participants.

Registration form here. Deadline: 15 January 2018

Submitted participant list

Background: Downscaling and bias correction are important steps in the processing data for the development of hydrometeorological forcing datasets for impact analyses of climate. Renewable energy forecasting, geo hazard warning, climate mitigation, and numerous other impact studies require downscaled or bias corrected data as forcing data to drive models. Presently a wide variety of techniques are in use. Each of the specific fields requires an understanding of the driving geophysical conditions, but also requires a high degree of specialization. A short course on hydrometeorological downscaling and bias correction techniques is offered to provide an improved foundation for students working in these fields.

Course objectives: The course will guide in the development of competence for a new generation of researchers and characterize the strengths and weaknesses of different hydrometeorological downscaling and bias correction approaches. Specifically, the course will provide an overview of the  application of statistical and dynamical approaches to downscaling and bias correction. Students will be introduced to the `esd` package in R that is widely in use for statistical downscaling and the Intermediate Complexity Atmospheric Research (ICAR) model that provides a quasi-dynamical approach to downscaling. Theoretical and practical background to statistical and dynamic approaches will be covered and assessed for applicability in varying use cases. This is a unique opportunity to gain instruction of the two software packages provided by their authors.

Outcomes: Participants will gain advanced knowledge on statistical, dynamical and pseudo dynamical downscaling and their role in providing relevant information for different applications. Following they will obtain:

  1. knowledge regarding different downscaling and bias correction techniques
  2. critical ability to assess and characterize the main challenges with the different approaches
  3. practical introduction to the `esd` package and the ICAR model, and knowledge of how to critically judge the outcome of different techniques and their usability for various applications.

Learning module: The course will consist of 3 full-day lectures and practical modelling exercises, followed by a project due 1 month after the course.  Attendants are expected to allow at least two days to prepare for the course. Students will use their own laptops and will be required to test software installations prior to arrival. Further, they are to read relevant literature from a provided reference list before arrival. The final project will be due 1 month following the course. It is beneficial for students to work with their own ‘case’ and datasets. Optimally, students will have these data available to work with during the course practicals. Further details regarding what type of data to prepare and the software installations will be provided once students have registered.

Course Outline:

Day 1:  Introduction to statistical and dynamical downscaling approaches for climate analyses; Statistical downscaling with `esd`
Day 2:  Dynamical and Quasi-dynamical downscaling with ICAR
Day 3:  The use of bias correction methods in hydrometeorological applications

Prerequisites: Students should have fundamental knowledge of meteorological and climatological processes. Familiarity with bash, linux console, and python or R scripting is helpful.