Downscaling techniques aim at resolving the scale discrepancy between climate change scenarios and the resolution demanded for impact assessments. Requirements for downscaled climate, to be useful for end users, include reliable representation of precipitation intensities, temporal and spatial variability, and physical parameters consistency. This report summarizes the results of the proof of concept phase in the development and testing of a novel data reconstruction method and a downscaling algorithm based on the multiplicative random cascade disaggregation method using rainfall signals at different spatial and temporal resolutions. The Wavelet Transformed-based Multi-Resolution Analysis (WT-MRA) was used for reconstructing the historical daily rainfall data needed as input for the downscaling methodology, using satellite-derived proxy data. Comparisons with presently used software showed that in all the cases; that is, the reconstructed, generated daily or downscaled daily data, the products developed outperformed the control test by either generating more accurate outcomes or by demanding significantly less parameterizing data.
Quiroz, R.; Posadas, A.; Yarlequé, C.; Heidinger, H.; Raymundo, R.; Carbajal, M.; Cruz, M.; Guerrero, J.; Mares, V.; Silvestre, E.; Jones, C.; de Carvalho, L.V.; Dinku, T. Application of non-linear techniques for daily weather data reconstruction and downscaling coarse climate data for local predictions. CCAFS Working Paper No. 21. CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS), Copenhagen, Denmark (2012) 32 pp.
Application of non-linear techniques for daily weather data reconstruction and downscaling coarse climate data for local predictions. CCAFS Working Paper No. 21