REGIONALIZATION OF THE PRECIPATION REGIME IN THE SOUTH-EASTERN PART OF MADAGASCAR

Climatology

The amount of rain during a given climatological period, within a specific region, determines the amount of agricultural production and the biodiversity life on that area. To know more about the rainfall class assigned at each point in the region, efficient and valid scientific techniques must be implemented. The main objective of this study is to look for homogeneous rainfall areas in the southeastern part of Madagascar; at 21°S to 24°S of latitude and 46.5°E to 48.5°E of longitude. We’ve chosen to use two different classification methods that are applied to rainfall data in this region. The first method is to classify linearly the data by using Principal Component Analysis. The second method is to use KOHONEN's competitive artificial neural networks to perform this classification. A single factorial axis, representing 92.693% of the variance of the axes, was selected for the Principal Component Analysis, thus generating three subfields. In the case of the Kohonen networks, composed of 6 by 7 neurons and a hexagonal topology, three distinct subfields could also be identified. In each subfield there are distinct monthly rainfall patterns, and the distribution of this last depends on the topographic position as well as the vegetation of the region considered.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.