Water Resources Management

, Volume 20, Issue 6, pp 899–915

Large-Scale Assessment of Drought Variability Based on NCEP/NCAR and ERA-40 Re-Analyses

  • Isabella Bordi
  • Klaus Fraedrich
  • Marcello Petitta
  • Alfonso Sutera
Article

DOI: 10.1007/s11269-005-9013-z

Cite this article as:
Bordi, I., Fraedrich, K., Petitta, M. et al. Water Resour Manage (2006) 20: 899. doi:10.1007/s11269-005-9013-z

Abstract

The impacts of different spatial resolutions and different data assimilation schemes of the available re-analysis data sets (NCEP/NCAR and ERA-40) on the assessment of drought variability are analysed. Particular attention has been devoted to the analysis of the possible existence of a linear trend in the climatic signal. The long-term aspects of drought over the globe during the last forty years have been evaluated by computing the Standardized Precipitation Index (SPI) on 24-month time scale. The SPI, in fact, seems to be a useful tool for monitoring dry and wet periods on multiple time scales and comparing climatic conditions of areas governed by different hydrological regimes. To unveil possible discrepancies between the analyses carried out with the two data sets, we studied the leading space-time variability of drought by applying the principal component analysis (PCA) to the SPI time series. Results suggest that on the global scale, the two re-analyses agree in their first principal component score, but not in the associated loading: both re-analyses capture a linear trend, though the areas where this feature should be most likely observed are not uniquely identified by the two data sets. Moreover, while the ERA-40 unveils the presence of a weak net “global” trend towards wet conditions, the NCEP/NCAR re-analysis suggests that the areas in the world characterised by positive/negative trends balance to zero. At large regional scale, a good agreement of the results with those obtained from the observations are found for the United Stated, while for the European sector the two re-analyses show remarkable differences both in the first loading and in representing the timing of the wet and dry periods. Also for these areas a linear trend, superposed on other short-term fluctuations, is detectable in the first principal component of the SPI field.

Keywords

drought assessmentre-analysis datalarge-scale variabilitystandardized precipitation indexprincipal component analysis

Copyright information

© Springer Science+Business Media, Inc. 2006

Authors and Affiliations

  • Isabella Bordi
    • 1
  • Klaus Fraedrich
    • 2
  • Marcello Petitta
    • 1
  • Alfonso Sutera
    • 1
  1. 1.Department of PhysicsUniversity of Rome “La Sapienza”RomeItaly
  2. 2.Meteorologisches InstitutUniversität HamburgHamburgGermany