Sonar data is commonly affected by noise due to the processing of scatter signals and interference of acoustic waves scattered from the seabed. To overcome this problem and to limit the noise in sonar images, the sonar operator can change the sonar settings (e.g. range, pulse length, modulation, inter-track distance, etc.) to acquire the best possible acoustic data. On board autonomous underwater vehicles (AUV), due to the low bandwidth of the communication with the robot, the real time definition of the best settings by an operator is nearly unfeasible. For these reasons, we have developed an analysis method for automatically assessing the quality of the data. The results of this process are then sent to the AUV planning module which can change the sonar settings (e.g. inter-track distance). The classical approach is based on the correction of the artefacts related to the wave propagation in water column and the characteristics of the sonar system. This approach requires strong a priori knowledge of the system and the conditions of acquisition of the sonar data. The main objective of this paper is to propose a statistical measure of quality of the sonar data acquired using AUVs. This statistical measure would be representing a quality map for the input sonar data. As no prior measurement of similarity or dissimilarity of sonar images is given, the decision to whether accept the quality of data as noisy/non-noisy will be based upon statistical hypothesis testing. To accomplish the quality mapping, spectral domain filtering is performed to extract the residual image representing the speckle. Based on Maximum Likelihood (ML) method, parameters are estimated from the data for Rayleigh distribution and its fit is evaluated using Goodness-of-fit (Gof) test. Experimental results show the viability of the proposed approach while mapping the data into quality matrix representing the acceptable regions on sonar data acquired using DAURADE.