In a recent work described in Ref. [1], an angle-independent methodology was developed to use the multi-beam echo sounder backscatter (MBES) data for the seabed sediment classification. The method employs the backscatter data at a certain angle to obtain the number of sediment classes and to discriminate between them by applying the Bayes decision rule to multiple hypotheses. This method is adopted and applied to very shallow-water applications. There are two issues when dealing with riverbed classification in shallow water. Shallow water depth results in a small beam-footprints and hence a small number of scatter pixels, which makes the classification results to be less discriminative. The significant bottom slopes will also affect the backscatter data and hence the classification results. We aim to handle these issues using the high resolution bathymetry and backscatter data. A methodology is developed to estimate the precise bottom slopes using the high resolution bathymetry data. Corrections are then applied to convert the arrival angle of the signal into the true incident angle and to compensate for the effect of the ensonified area. The high resolution backscatter data allows one to reduce the statistical fluctuations using an averaging procedure. The methodology will then be tested on a MBES data set from the river Waal, the Netherlands. The acoustic classification results are correlated with the mean grain sizes of the sediment obtained from core analysis of the grab samples. The dependence of acoustic backscatter classification on sediment physical properties is verified by observing a significant positive correlation coefficient of 0.70 between the classification results and sediment mean grain sizes of the grab samples.