Proposition of a punctual estimator for vertical uncertainty of hydrographic surveys

Abstract

Knowledge of the submerged morphology of aquatic environments has always been a challenge, precisely due to the inherent difficulties in characterizing and exploring the non-apparent occurrences of these surfaces. The description of the characteristics of the oceans, rivers, lakes and other bodies of water makes it possible to obtain bathymetric information useful in several areas, such as maritime or fluvial navigation, civil works, prospecting of mineral resources, supply or generation of energy, etc.  From the depth information it’s possible to establish practices directed to the planning and execution of numerous hydrographic activities. The hydrographic surveys, currently used, are performed by acoustic systems such as single beam echo sounders, multibeam echo sounders and interferometric sonars. However, regardless of the technology used, the data collected will always contain uncertainties, that is caused by gross, systematic or random effect. If the data are subject to uncertainties not acceptable to a given standard tolerance, such information may not be accurate for certain purposes. One of the commonly used estimators, RMSE (Root Mean Square Error) is highly influenced by the presence of outliers in the samples, and may not be adequate to describe the statistical quality of the set of observations. Thus, the objective of this research is the proposition of a point estimator to quantify the vertical uncertainty of bathymetric surveys, called Robust Uncertainty, which, unlike the estimators used most of the time, is resistant to outliers and does not depend on the probability distribution of the sample. To establish the confidence interval of this estimator, the Bootstrap technique was used. Experiments were performed with simulated data, as well as the use of real data, referring to two study areas. From the results obtained, it was possible to verify the performance of the proposed estimator, which was clearly resistant to the possible outliers present in the data set. It was also observed that the presence of the outliers in the databases had little influence on the point estimates of uncertainty, showing their efficiency and robustness.