A review of approaches for water depth estimation with multispectral data

Abstract

The purpose of this paper is to give an exhaustive assessment of water depth estimation in shallow inland and coastal water environments with the assistance of optical remote detecting procedures, utilizing satellite/airborne multispectral information. Remote sensing has been used to map bathymetry for several decades. Inaccessibility, large‐scale depth mapping, very shallow areas, and cost constraint are some of the reasons why remote sensing technique is a boon for bathymetry. Researchers have come a long way since the remote sensing‐based bathymetry has been used. Fourfold increase in spatial, spectral, radiometric and temporal resolution of sensors have also increased the accuracy. There are two types of remote sensing techniques; Active and Passive. Active remote sensing includes the use of ground penetrating radar and bathymetric lidar, which are expensive but precise surveys successful up to 70 m depth. On the other hand, passive remote sensing deals with water photogrammetry and radiometric methods done by capturing imagery with space borne/air borne platforms having depth penetration up to 25 m. The latter technique includes finding of a logarithmic relationship between river depth and image values. This technique is most widely used because of its cost effectiveness. In this paper, the emphasis has been given to derive bathymetric techniques using multispectral data as these methods have proved reasonably successful when water quality, bottom reflectance, and atmospheric effects are all invariant. At a specific area, the detection of possible changes in water flow and levels indicates flow path changes which can be used to predict potential surface level flooding for better water management and policy decisions.

In-situ manipulations of aquatic optical depth and its effect on small unoccupied aerial systems–derived spectral reflectance

“…Conversely, the only bands that did not exhibit a significant difference (p-value > 0.05) for the M and BM vessels were the blue and NIR bands (Table 3). Reflected NIR energy was not significantly different because water strongly absorbed NIR energy, causing a shift from lower to higher wavelengths (Shah et al 2020).…”

“…Gholizadeh et al (2016) state that in clear waters, the maximum light penetration at 475 nm (e.g., RedEdge sensor blue band center point) can be as deep as 55 m and as shallow as 60 cm when TSS concentrations approach 400 mg L -1 . Jensen (1989) and Shah et al (2020) support this by stating that if OAC concentrations were low, the optimum wavelength range to assess bathymetric properties is 440 -540 nm. Within a similar spectral region (e.g., 450 -600 nm), bottom reflectance has been shown to significantly increase reflectance values measured (Cannizzaro and Carder 2006;Zeng et al 2017).…”

“…However, the peak reflectance measured for the C and B vessels was from the NIR band. NIR energy can help distinguish between physically shallow and deep waters because between 0.3 and 0.9 m, NIR energy can be readily absorbed (Shah et al 2020). Studies have also used NIR energy to develop single band (Shafique et al 2003) and band ratio (Larson et al 2018) relationships with TSS, suggesting the high NIR values observed were the result of interactions with the bottom substrate (PreC and PreB vessels) and substrate in solution (PostC and PostB).…”

“…The MChl-a models exhibited significant differences in both slope and constants. The significant increase in Fe from PreM to PostM likely caused the spectra to shift and change shapes at higher wavelengths impacting the region in which Chl-a could be readily estimated (Shah et al 2020).…”

A Preliminary Assessment of a Newly-Defined Multispectral Hue Space for Retrieving River Depth with Optical Imagery and In Situ Calibration Data

“…Slope and intercept depend on optical properties of the atmosphere, water, and substrate at each wavelength but, if they can be considered invariant across the image, Equation (4) provides a predictor for depth. It has been used in numerous studies (see Shah et al [11] for a recent review), either with the calibration of slope and intercept against in situ depth measurement, or without calibration; in the latter case, attenuation coefficients K(λ) have to be modeled physically [12,13].…”