FAO home page
Contact References Search in FAO Français Español
About Climpag Related links
Climpag home page
Climate monitoring
Climate risk
El Nino
Southern Oscillation
Health aspects
Rainfall Variability
Methods and tools
 
Rainfall Variability Analysis of rainfall variability and drought in the 1961-2002 period
The global rainfall pattern by classes

Global analysis  Annual rainfall | National rainfall index | Global rainfall pattern
Case studies  Burkina Faso | Cambodia | Nepal | Tanzania
Downloads  Images (IDA) | Images (GeoTiff) | Movies (AVI) | Windisp | Quantum GIS

Global map  All profiles  The driest  Intermediate classes  The wettest  Coefficient of variation

Classification

Method
To get an overview of global rainfall variability, a pixel-by-pixel classification has been performed. On that way, areas with similar climatological characteristics can be identified. The analysis is based on the yearly series of the annual rainfall images, expressed in mm/month, drawn from the NOAA data base. Using ADDAPIX software, a spatial and temporal analysis is achieved. After submitting the yearly series to a principal component analysis a non hierarchical cluster analysis follows. The principal component analysis allows the reduction of the data sample with a negligible loss of information. Mostly, a small number of principle components explain an overwhelming majority of the overall varialbility of the input data. Subsequently, clustering of the pixels on the basis of the chosen first principle components is performed in order to define regions with similar characteristics in rainfall variability between 1961 and 2002. In this analysis we restricted to a number of 12 classes. The global map (see menu) shows the spatial distribution of the classes.
Results
As anticipated, the global map with the geographical distribution of the 12 rainfall classes shows the driest classes in the subtropics, the wettest in the tropical region and intermediate rainfall amounts in great parts of the midlatitude regions. But superimposed orographic effects and circulation systems also result in exalted spatial variations of yearly rainfall amounts (see the rainfall classes of Burkina Faso, Cambodia, Tanzania and Nepal).
Furthermore, the profiles of the annual rainfall amounts, expressed in mm/month, are plotted for every rainfall class. In order to compare the rainfall variability of the rainfall classes, each rainfall amount of a single year is divided by the average monthly rainfall amount of the respective class and multiplied with 100. Obviously, the drier the rainfall class the higher is the interannual variability. Semi-arid and arid climates show a higher variability than moist climates. Less variability can be found in countries with high rainfall or in large countries, which include several agroecological zones. For a clearer presentation the profiles of the driest, the intermediate classes and the wettest classes are plotted seperately. Apparently, the presentation of these profiles also makes the detection of drought-cycles possible. The coefficient of variation allows a comparison of the variation of populations that have significantly different mean values. Plotting the coefficient of variation versus the average monthly rainfall underlines the statements above (see coefficient of variation).


  Home |  About |  References |  Links |  Contacts |  NRC |  NR |   
  Comments? Please contact us