A Common Factor Analysis Based Data Mining Procedure for Effective Assessment of 21st Century Drought under Multiple Global Climate Models

Continued global warming has increased the risk of drought all over the world. Therefore, effective drought assessment, in conjunction with accurate drought characterization and its long-term evaluation, is essential. In recent developments, the use multi-model ensemble data of various specific sets of Global Climate Models (GCMs) is common in climate research. This research provides a new drought index – Multivariate Weighted Ensemble Standardized Drought Index (MWESDI). The procedure of MWESDI uses Common Factor Analysis (CFA) based data mining approach for handling multiplicity and specificity structural problems in the data sets. The proposed procedure aims to enhance the effective use of GCMs by addressing the data problems associated with dimensionality reduction and important feature extraction. In application, we used observed and simulated time series data of 18 GCMs distributed across the Tibet Plateau region of China. Based on error performance measures, the proposed index is a more reliable and precise measure compared to its competitors. Our findings related to drought assessment indicate that the Tibet Plateau will probably face more severe and frequent droughts during the 21st Century.