Automatic measurement and computation of stream cross-sectional area using ground penetrating radar with an adaptive filter and empirical mode decomposition

An automatic and non-contact method for measuring cross-sectional area by using the ground penetrating radar (GPR) with an adaptive algorithm is proposed in this study. The adaptive algorithm is composed of an adaptive filter, maximum amplitude method and empirical mode decomposition (EMD) method for estimating the cross-sectional area of a stream. It has been widely considered dangerous, labor-intensive and time-consuming to measure the cross-sectional area of a stream; moreover, difficulties in measuring the water area increase as the stream beds changing drastically with strong current. Compared with the conventional measurements by sounding weights, a ground penetrating radar (GPR) without contacting the water is used in this study to measure the water area, which obtains the continuous measurements of the stream cross-sectional area accurately and promptly. This non-contact measuring method can keep hydrologists away from risk and dangerous situations, especially during high water. However, signals of raw data obtained from a GPR are quite complicated and rely on experts to interpret. Nevertheless, data interpretation by human is still more subjective than a machine. In this study, the measured electromagnetic signals were converted into digital signals. The adaptive filter approach was applied to reduce biased error of the GPR signals. The EMD was then used to smooth the cross section shape. The method for interpreting the GPR data was developed to reduce the subjective judgements and to improve the interpretation of data for obtaining the accurate cross-sectional area of a stream. Compared with the results given by the conventional method, only minor differences exist. However, the time and efforts needed for measurements were significantly reduced. It demonstrates that a GPR integrated with the proposed calculation method could be practically employed to accurately measure the cross-sectional area of a stream. Moreover, the procedure proposed in this study can be integrated more easily into an automatic and non-contact streamgaging for continuous gaging during a flood event.