 |
<< Back to WCRP CMIP3 Subprojects
Applications of nonlinear time series analysis methods
| PI: |
Jiri Miksovsky |
| Institution: |
Charles University, Prague, Czech Republic |
| Abstract: |
The data are intended for use in the research project entitled ~Applications of nonlinear time series analysis methods for study of spatial relations between climatic variables~, supported by the Czech Science Foundation (ID 205/06/P181), specifically for statistical downscaling analyses.
The aim of the project is to assess the possibilities of application of nonlinear time series analysis methods for constructing mappings between different variables of the climate system. Several nonlinear techniques will be examined (multilayer perceptron neural networks, RBF neural networks, method of local linear models) and their ability to serve for implementation of transfer functions between predictors (reanalyses of climate variables and outputs of climate models) and predictands (temperature and precipitation at several Czech and European stations) will be evaluated. Aside from nonlinear downscaling techniques, attention will also be paid to possible application of nonlinear methods for filling in missing data in the existing time series, using the records from nearby stations as predictors. The outcomes of the project should contribute to the improvement of the results of statistical downscaling and to completion of the existing records.
|
| Publications: |
Add Publication
|
<< Back to WCRP CMIP3 Subprojects
|
|
