• English 
  • Spanish 


MULTIivariate Statistical Downscaling Methods (spatial and/or multi-variable): Contribution to the international initiatives and to Escenarios-PNACC National Program

logo_plannacional.png Tipo de proyecto: Proyecto nacional
Programa: Spanish R+D Program 2013-2016
Código: CGL2015-66583-R
Periodo: Enero 2016 - Diciembre 2019
Estado: En progrso

Regional climate change projection (from global to local projections) is one of the current key priorities for both the Intergovernmental Panel on Climate Change (IPCC) and the World Climate Research Program (WCRP) due to the increasing demand of these products by the impacts and risk assessment communities. Statistical Downscaling Methods (SDMs) is one of the existing methodologies to produce local climate change projections from the output of Global (GCM) or Regional (RCM) Circulation Models, based on a variety of techniques developed so far. However, there are still limiting factors that prevent the comprehensive worldwide application of SDMs, including the different limitations and assumptions of the different techniques.

Several international initiatives have emerged in this field in the last years to explore the strategic challenges: 1) CORDEX-ESD and the European branch EURO-CORDEX (http://www.cordex.org, sponsored by the WCRP) and 2) the action COST VALUE (http://www.value-cost.eu) from the VII-FP of the UE. The first validation and inter- comparison experiments have been recently proposed in the framework of these initiatives to analyze the limitations and applicability of these techniques, with a relevant participation of the research group of this project. One of the interesting problems in this area is the development of multi-variate models providing results spatially consistent (multi-site) and/or with coherence between variables (multi-variable). These requirements are needed by different impact communities (e.g. hydrology).

The aim of this project is to develop new multi-variate downscaling techniques, working in the framework of these initiatives and inter-comparison experiments enhancing the divulgation, application and impact of the results. The capabilities of the probabilistic networks to build simple and computationally cheap models working at daily scale will be explored. These techniques have been applied successfully to other fields to build multi-variate discrete, gaussian, exponential or mixture models from a local factorization of the corresponding join probability, which is automatically obtained from the available data using appropriate learning algorithms (data mining).

This project will not only contribute to the international initiatives, but also to the Plan de Adaptación al Cambio Climático (Escenario-PNACC 2012), actually based on the IPCC-AR4. First, the existing products (the corresponding to statistical downscaling methods) will be updated taking into account the improvements achieved in the downscaling methods in the last years and the models of the last inform of the IPCC-AR5 (CMIP5). Second, according to the results obtained in this project, a new update of these scenarios will be developed, including, when possible, of MOS and multi-variate products.
The objectives of the project are:
1) Develop new multi-variate downscaling method with spatial and inter-variable consistence.
2) Contribute to the international initiatives CORDEX and the inter-comparison experiments, strengthening the international position of the group.
3) Contribute to the update of the Escenarios-PNACC, in collaboration with the Oficina Española de Cambio Climático (OECC).