Estimating herbage standing crop from rainfall data in Niger.

Abstract

To help local Niger government authorities and donor countries ameliorate conditions in the advent of drought, a rapid yet simple means to assess annual herbaceous production at the end of the rainy season is needed. Several rainfall variables were tested as estimators of herbaceous production using weighted and logarithmic transformation in regression analysis. A computer program was developed in Basic to generate rainfall parameters from daily rainfall data. Input parameters used to calculate the rainfall variables were varied to minimize the regression standard error of the estimates. Selected regression models were compared using 80% confidence levels for mean values for each rainfall treatment class using logarithmic and weighted regressions. The selected weighted model involved the number of moist days and consecutive dry days as independent variables. The selected logarithmic model used total rainfall as the only independent variable. These models were tested by comparing an independent data set with the 95% confidence intervals for observations. Selected models separated rangeland production classes of 200 kg ha-1 confidence limits for mean values. The logarithmic model could only do so when biomass levels were less than 800 ha-1. Thus, these models only have application for predicting herbage biomass within rather large classes.