Introduction

R3 represents a new and innovative approach to provide climate services at local scales in Norway. It will accomplish this through implementation of an integrated approach that aims to account for a range of possible outcomes and extensive interaction with stakeholders. Norway is characterized by large regional differences in geography and climate. An integrated modelling approach - one that can encompass a range of outcomes and represents detailed spatial structure and dynamics - is required to produce reliable climate change projections. Our approach addresses these requirements and also addresses the "usability gap" between climate information and decision-making. The need for improved, robust and credible estimates of local-scale climate change in decision-relevant contexts is pressing. We identify four challenges that currently hinder production of relevant, reliable and robust climate projections on local scales: a) Insufficient resolution in existing multi-model ensembles; b) Persistent biases in driving general circulation models; c) Dynamical approaches sample a small range of outcomes and under-represent uncertainty; and d) Lack of effective two-way communication between scientists and user communities often results in outputs/products do not match users' need for decision-relevant climate information.

R3 will make use of existing capabilities of the Norwegian Centre for Climate Services (NCCS) to develop a new framework for producing reliable climate change projections on local scales. We address the challenges described above through a number of actions, which culminate in an integrated modelling strategy for local scale climate projections. The added value of this approach will be evaluated, both from the stakeholders' perspective and in terms of scientific outcomes. The project's methodological innovations and scientific advances will benefit Norway and will provide a framework that will be generally transferable to other regions of the world.

Objectives

Primary Objective:

Develop and implement innovative, integrated approaches to produce reliable local-scale climate projections for communities and stakeholders in Norway.

Secondary Objectives:

1. Continuous collaboration with users and stakeholders in the co-production of climate know-how; tailor the provision of climate information to their needs and opportunities in ways that are relevant to their organizational structures and cultures;

2. Extract and downscale robust climate change signals from global climate models in a way that mitigates systematic biases over the Northeast Atlantic region;

3. Integrate this with empirical-statistical downscaling to obtain temperature and precipitation statistics for local scales and expand the ensemble size to cover a range of outcomes;

4. Develop decision-relevant climate information in collaboration with project stakeholders;

5. Demonstrate the added-value of the integrated approach from both the stakeholder and scientific perspectives.

Work Packages

WP1: Understanding the communication and co-production of climate services

LEAD: Simon Neby (Uni Research Rokkansenter)

To ensure effective communication and learning processes throughout the project, through identifying bottlenecks and constraints both in the production process and in the application of the data within local climate planning activities. The more overall aim is to contribute to understanding how local climate modelling can be integrated into local level planning activities.

The research in this work package has two main purposes: 1) Study the coordination efforts and translatory process of innovation from science into practice, and 2) Identify bottlenecks and problem areas in order to facilitate and improve cooperation and communication within and beyond the project.

WP2: Upstream bias correction and dynamical downscaling

LEAD: Erik Kolstad (Uni Research Klima)

All climate models have biases, and for Norway the most relevant of these are excess amounts of Arctic sea ice and an erroneous North Atlantic jet stream (too zonal, too far south). If a ‘brute force’ approach is used, where climate model data are used directly as boundary conditions for the regional model and/or ESD approach, these model biases can lead to misrepresentations (the ‘garbage-in-garbage-out’ issue; Giorgi and Mearns 1999).

Research questions:

• What is the impact of biases in sea ice, sea surface temperatures (SSTs) and the jet stream in the driving GCM data on the regional model, and how can we reduce these biases?

• Can we identify model configurations optimized for reproducing the most important climate variables in all three regions (The West, The East and The North) of Norway?

Approach: To address the first task (bias-correction), we will build on the methodology in Bruyère et al. (2014) to correct the GCMs’ biases in the jet stream and North Atlantic storm track.

In the second task, we will first identify the best possible regional model configurations for each of the three regions of Norway, by testing the sensitivity of the model to different factors, such as the horizontal and vertical resolution, and the parameterization schemes for the planetary boundary layer, microphysics, radiation and convection.

Once the model configurations are determined, we will downscale the ERA-Interim reanalysis dataset for the period 1986–2005 to serve as our baseline climate state control run. We will also downscale three bias-corrected GCM model runs for a future time period (to be determined with stakeholder input) under the Representative Concentration Pathway – namely the RCP8.5 scenario – provided by the Intergovernmental Panel on Climate Change (IPCC), yielding high-resolution training data for the hybrid empirical-statistical downscaling approach in the next WP.

WP3: The hybrid downscaling approach and local scale ESD

LEAD: Rasmus Benestad (met.no), Abdelkader Mezghani (met.no)

The aim in this work package is threefold: i) Adapt the hybrid approach developed by Walton et al. (2015) to Norway; ii) Extend the approach by including precipitation (P) as well as temperature (T); and iii) Employ weather generators to produce local-scale projections of daily time series. We hypothesize that the hybrid downscaling approach will allow for computation of robust statistics and full sampling of uncertainty in the CMIP5 ensemble.

It is important to validate the downscaling strategies and put them to the test to see if they are able to reproduce similar aspects in the past as they are being used to project for the future (Benestad et al., 2007).

WP4: Extension to hydrological applications

LEAD: Deborah Lawrence (NVE)

The hydrological impacts of climate change are one area where the link between science and areal planning are already being forged (Vasskog et al., 2009; Lawrence and Hisdal, 2011). However, there are still challenges that need to be addressed to increase the skill, credibility and robustness of future assessments.

This work package will consider two alternative and complementary approaches for extending the output of climate models to produce locally relevant hydrological information. The first approach will entail the application of ESD using the monthly or seasonal temperature and precipitation outputs from the hybrid dynamical-statistical downscaling approach to estimate seasonal runoff and the probability of over-threshold flows. The second approach will use the monthly {P,T} statistics in combination with a weather generator (Mezghani and Hingray, 2009) to produce a large ensemble of hydrological model runs for given locations, such that the hydrological response to the statistical characteristics of the {P,T} input data can be assessed.