Statistical downscaling skill under present climate conditions: A synthesis of the VALUE perfect predictor experiment
VALUE is a network that developed a framework to evaluate statistical downscaling methods including model output statistics such as simple bias correction and quantile mapping; perfect prognosis methods such as regression models and analog methods; and weather generators. The first experiment addres...
| Autores: | , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2019 |
| País: | España |
| Institución: | Consejo Superior de Investigaciones Científicas (CSIC) |
| Repositorio: | DIGITAL.CSIC. Repositorio Institucional del CSIC |
| OAI Identifier: | oai:digital.csic.es:10261/213840 |
| Acceso en línea: | http://hdl.handle.net/10261/213840 |
| Access Level: | acceso abierto |
| Palabra clave: | Bias correction Evaluation Regional climate Statistical downscaling Validation |
| Sumario: | VALUE is a network that developed a framework to evaluate statistical downscaling methods including model output statistics such as simple bias correction and quantile mapping; perfect prognosis methods such as regression models and analog methods; and weather generators. The first experiment addresses the downscaling performance in present climate with perfect predictors. This paper presents a synthesis of the VALUE special issue, with a focus on the results of this first experiment. This paper presents a synthesis of the results. Model output statistics performs mostly well, but requires predictors at a resolution close to the target one. Perfect prog performance depends crucially on model structure and predictor choice. Weather generators perform in principle well for all aspects that can be expressed by the available model structure. Inter-annual variability is underrepresented by both perfect prog and weather generator approaches. Spatial variability is poorly represented by almost all participating methods (inherited by model output statistics from the driving model, not represented by the perfect prog and weather generator methods). Further studies are required to systematically assess (a) the role of predictor choice for perfect prog; (b) the performance of spatial weather generators, to study the performance based on GCM predictors; (c) downscaling skill in simulated future climates; and (d) the credibility of simulated predictors in a future climate. |
|---|