This work package will be carried out at the Swiss Federal Research Institute WSL.

WP1 provides subsequent WPs with high-resolution data needed for the two dynamic vegetation models (DVMs); LPJ-GUESS (WP2) and Fate-HD (WP3), as well as for calibrating partial biodiversity models (WP4/WP5). WP6 will produce large amounts of climate data from the COSMO-CLM² climate model, yet at a 0.11° grid resolution, which is too coarse to capture climate variation in complex terrain. Global circulation and weather models are already able to run at high horizontal resolutions close to 1 km, but are still heavily constrained by computational limits. Currently, numerical km-scale models are computationally too expensive for coupling them with DVMs and BD models. Therefore, we will apply a downscaling to the COSMO-CLM² (see WP6) output based on the CHELSA algorithm. We will do so for a reduced set of climatic variables needed for impact studies, with a computational efficiency that is 745 times faster than a numerical model, and therefore suitable for downscaling COSMO-CLM² output to a 1 km, or higher resolution. For each climate variable, we will develop a module to downscale the input data at daily resolution to 1 km for Europe and 100 m for the European mountain ranges. In addition to the high-resolution climate data, WP1 will also provide remotely sensed data from Sentinel2 to provide surface albedo to parametrize their relationship to climate.

Tasks

1. Downscale bias corrected climate data: CLM modelled precipitation, temperature, and wind speed will be bias-corrected using the ISIMIP3b method following, which is a parametric quantile mapping method designed to adjust biases in all percentiles of a distribution of a climate variable while preserving trends in these percentiles including extremes. The downscaling of precipitation, temperature, wind speed, and solar radiation for DVMs and BD models in subsequent analysis will be performed using an updated CHELSA algorithm. This algorithm applies a series of statistical- and mechanistic downscaling routines to data provided at coarser (here CLM) resolution. Temperature and radiation are freely scalable and the method works on either 1 km resolution (Europe) or 100m (Mountains) resolution alike. Cloud cover and precipitation are only downscaled to 1 km, as further downscaling rather creates noise.
2. Coupling of downscaling algorithms with DVMs: In this task, we develop, benchmark, and implement a routine that integrates COSMO-CLM² with CHELSA, the DVMs, and the BD models, and explores feedbacks into the COSMO-CLM² on an HPC system. This will allow better and more directly exploring BD effects on CLM- modelled climate data.
3. Parametrization of BD climate relationships: Here, we replace the parametrizations in ESMs with more detailed schemes using the downscaled climate data, in combination with remotely sensed data (from Landsat, Sentinel2), LPJ-GUESS simulations in WP2 and classifications from WP4 to better represent BD-sensitive climate parameters (e.g. albedo, vegetation structure, etc.)
4. Synthesizing the overall results of the project: This task will focus on the coordination and synthesis of the project to an integrated, high ranking scientific publication.

Milestones

1. Bias correction & downscaling implemented
2. Coupling of bias correction and DVMs achieved
3. BD-climate relationships parameterized from RS data

Deliverables

1. Data paper on downscaled climate data;
2. Climate datasets on an open data portal;
3. Paper on coupling between bias corrected, downscaled climate data, dynamic vegetation and earth system models;
4. Paper on influence of remotely sensed vegetation properties on climate.
5. Overall synthesis paper of the influence of biodiversity on climate.

Cover image: Jay Mantri, Unsplash.