You are here: Home Project description Activities Improved Modeling and Uncertainty Assessment

Improved Modeling and Uncertainty Assessment

The general goal of activity 5 is to improve existing modelling techniques building on new data of crucial importance and on the testing of important modelling assumption (both developed in activity 1 through 4). Activity 5 is based on different approaches (niche models, DVM, dispersal models, coupled models) that have been used so far to derive global change impact projections, each with advantages and drawbacks and often with no sufficient attention to topics like uncertainty estimates. In this context, A5 will try to compare and combine these approaches into an integrated modelling framework.

Download Briefing Sheet on Work Description of Activity 5.


Over the last two decades, several modelling techniques were developed and used to evaluate the potential impacts of global change on biodiversity. However, despite major methodological improvements, some major limitations remained that required new data and developments to be overcome. This activity precisely builds on the previous four activities where new data of crucial importance are generated and important modelling assumptions are being tested, to further improve modelling techniques and derive global change projections and integrated assessment.


Activity 5 aims at improving a range of modelling approaches, especially:

  1. Statistical niche-based models of species and diversity distributions.
  2. Dynamic models (including DVM and landscape models).

Besides improving each approach and model per se, key efforts here will be to:

  1. Develop a dispersal module for species or functional groups that can be further incorporated into the different modelling approaches;
  2. Evaluate all types of models and their projections; in particular, we will test the ability of models to predict the future by assessing how they predict the past from the present (hindcasting), or the present from the past (forecasting).
  3. Quantify uncertainty along the whole modelling process to provide more informative projections of global change impact on biodiversity in Europe at various spatial scales.


Two core activities take place within this activity: (i) an array of models are being developed and improved at three geographical scales over currently existing approaches; and (ii) thorough testing and uncertainty analyses of the models are carried out.

The three scales used for modelling are Europe (continental scale), regional study areas (e.g. Alps, Finnmark, Central Europe; regional scale) and local study areas (landscape scale). Three types of predictive model are being improved over currently existing approaches. First, niche-based models of species distribution, ecosystem boundaries, and biodiversity pattern are calibrated based on new spatial predictors of climate, topography and land use. Scale effects are also assessed and multi-model projections (ensemble forecasting) developed with these models. Second, a hierarchical hybrid model, combining a spatially explicit dynamic landscape model and a population viability model is being developed and tested. And third, biogeochemistry models (LPJ, CARAIB) are being improved and will be applied to scenario simulations in order to evaluate expected ranges of ecosystem processes and functioning for a range of plant functional types, biomes and bioclimatic affinity groups in Activity 6. A transversal effort is also dedicated to developing a dispersal module that can be incorporated in all these modelling approaches. The new dispersal routine will allow simulating forecasted migrations in all three types of models, by considering either species or functional groups.

Testing and assessing uncertainty in models and projections is a crucial task, yet not sufficiently considered so far in global change ecological studies. A thorough evaluation procedure is being developed to assess predictions. Here, paleoclimatic data obtained in A1 will serve at projecting models back and forth between past and present times, as a way to evaluate our ability to predict the future. Projections in the past will be checked against paleodata on past species distribution from the European pollen database and, where possible, using the novel paleo DNA identifications from A2. Uncertainty assessment is as important. It is assessed at all levels of the project, by gathering and storing error and uncertainty estimates associated to all data, modelling techniques and projection tools used or developed throughout the project.

Results and Publications

available for

Task 5.1.2 Scaling between fine and large scales

Task 5.1.3 Optimizing biodiversity scenarios by multi-model predictions

Task 5.2.1 Develop and improve tools

Task 5.5.1. Protocol for evaluating models

Document Actions