What's the preference: shallowest depth, average depth, or somewhere in between?
Improving Bathymetric Modelling in Multibeam Echosounder Systems
In the realm of Multibeam Echosounder (MBES) systems, assigning a depth to the centre of a grid cell is a crucial yet challenging task. Traditional methods often rely on a simple average, but new approaches are emerging that incorporate statistical properties and robust statistics to better represent the seafloor within each cell.
One such approach is the combination of the mean depth with statistical measures, such as the standard deviation or variance. This method helps quantify the uncertainty or heterogeneity in the measurements, providing a more accurate representation of the seafloor.
Another technique is the use of robust statistics, like median depth or trimmed means, to minimise the influence of outliers or extreme values. Weighted averages, where individual soundings are given different weights based on their quality or spatial distribution, are also employed to improve the assigned depth's reliability.
Probabilistic or Bayesian methods model depth as a distribution rather than a single value, considering both the mean and higher moments (skewness, kurtosis) or prior information. Interpolation-informed depth assignment, which incorporates geostatistical properties like semivariance or other variogram statistics, is another method that improves the depth estimate at the grid cell centre based on spatial correlation among points.
While these methods are not directly addressed in the provided search results, they are consistent with common practices in geospatial data processing and ocean mapping, where mean values are complemented by statistical descriptions to enhance data representation and mapping accuracy.
Figures 8 and 9 depict the seafloor profiles along designated lines in two areas, showcasing the profiles of the shallowest, mean, and depths based on the mean and (corrected and uncorrected) standard deviation. The use of the corrected standard deviation reduces the percentage of cells with shallower obtained depths compared to the actual shallowest measured depths to around 1%.
However, it's important to note that using the mathematical shallowest depth based on the uncorrected standard deviation can still result in depths that are shallower than the actual shallowest measured depths for nearly 6% of the cells. Similarly, using the mathematical shallowest depth based on regression coefficients can lead to unrealistic depth values for some cells due to the points being spatially not well distributed.
Using the mean depth can help avoid an unrealistically shallow grid, but it might leave hazardous objects undetected. Research on more realistic measures to account for the depth variation in a cell is still needed, especially in regions with depth variations occurring over a relatively small distance or the existence of man-made trenches.
In conclusion, these alternative approaches for assigning depth to the grid cell centre in MBES data combine the mean depth with statistical measures and robust or weighted statistics to better represent the sounding data within each cell, improving the reliability of the bathymetric model. The introduced alternatives are applied to data derived from two surveys, one in the vicinity of the Eemshaven seaport and one in the Westerschelde estuary.
- In the exploration of Multibeingham Echosounder (MBES) data, environmental science finds substantial importance in combining mean depth with statistical measures such as standard deviation or variance, while leveraging technology like data-and-cloud computing for a more precise representation of the seafloor within each cell.
- As research progresses in improving bathymetric modeling in MBES systems, technology plays a pivotal role in embracing new methods like probabilistic or Bayesian approaches, using robust statistics and geospatial data processing techniques to better analyze and represent ocean mapping data.
