Estimation of Rainfall in Burkina Faso Using the ESOC Precipitation Index

View More View Less
  • 1 European Space Operations Centre, Darmstadi, Federal Republic of Germany
© Get Permissions Rent on DeepDyve
Restricted access

Abstract

Rainfall is estimated in Burkina Faso for a full year using the ESOC precipitation index (EPI), a statistical cloud indexing method based on satellite data from METEOSAT. The EPI is converted into rainfall with the linear regression calculated between the EPI and the observed rainfall from a dense network of rain gages. Only one regression line based on the largest possible sample is used. The purpose of the paper is to assess the accuracies of the yearly and seasonal rain estimates to find out whether a single EPI-rainfall relation can be applied.

The results show that the yearly precipitation can be estimated to a high degree of accuracy. On the other hand, the precision of the seasonal estimates exhibits large fluctuations. While the dry season estimates are reliable, the transition between the dry and rainy season is characterized by a considerable overestimation, caused by the abundance of cold, nonprecipitating cirrus on the northern side of the intertropical convergence zone. During the rainy season the method suffers from a slight underestimation.

To resolve the major problem, that of nonprecipitating cirrus, a lower temperature threshold of 220 K instead of 235 K is applied in the determination of the EPI. The rain estimates for the transition period do improve slightly, but the gain is offset by the deterioration of the rain estimates made for the whole year and the rainy season.

The results suggest that the rain estimates made with a single EPI-rainfall relation are useful, but that they could be improved with some type of seasonal adjustments.

Abstract

Rainfall is estimated in Burkina Faso for a full year using the ESOC precipitation index (EPI), a statistical cloud indexing method based on satellite data from METEOSAT. The EPI is converted into rainfall with the linear regression calculated between the EPI and the observed rainfall from a dense network of rain gages. Only one regression line based on the largest possible sample is used. The purpose of the paper is to assess the accuracies of the yearly and seasonal rain estimates to find out whether a single EPI-rainfall relation can be applied.

The results show that the yearly precipitation can be estimated to a high degree of accuracy. On the other hand, the precision of the seasonal estimates exhibits large fluctuations. While the dry season estimates are reliable, the transition between the dry and rainy season is characterized by a considerable overestimation, caused by the abundance of cold, nonprecipitating cirrus on the northern side of the intertropical convergence zone. During the rainy season the method suffers from a slight underestimation.

To resolve the major problem, that of nonprecipitating cirrus, a lower temperature threshold of 220 K instead of 235 K is applied in the determination of the EPI. The rain estimates for the transition period do improve slightly, but the gain is offset by the deterioration of the rain estimates made for the whole year and the rainy season.

The results suggest that the rain estimates made with a single EPI-rainfall relation are useful, but that they could be improved with some type of seasonal adjustments.

Save