A reduced-rank prediction model of stream-channel size response to traditional agricultural land-use practices in southwestern Nigeria

Abstract

Stream-channel morphologic responses are found to be related to different parameters measuring traditional agricultural land-use patterns and practices in 50 small headwater basins in southwest Nigeria. The problem of intercorrelations among these parameters made it initially difficult to establish their precise channel enlargement effects and to calibrate an impact prediction model. Through factor analysis of the 22 land-use and morphometric parameters, six factors identified as measures of traditional land-use practice, farm size, planting activities, shortened fallow, relief and overland flow, were found to account for 86% of the variance in the data. The factor-defining variables are length of cropping period, areas in short fallow, farm-plot size, length of farm preparation, relief ratio and overland flow. In a multiple regression analysis, only the first three variables were found to be statistically significant in explaining stream-channel morphologic responses. Thus, areas in short fallows, average farm size and length of cropping period adequately described those aspects of the traditional farming practices that affect basin hydrologic and channel responses. Since these variables were orthogonally derived, they formed the basis for the evaluation of the channel impact status of traditional land-use activities. The duplication of information and effects in the original 22-variable full-rank model were removed while utilizing the three-factor reduced model.