SkyBathy Lee/SA Lite

Bathymetry Preview

Load GeoTIFF bands to start.
Language:

Lee / SA Lite Model Guide

This tool implements a simplified Semi-Analytical (SA) model based on Lee et al. (1998, 1999) theory to derive bathymetry from multispectral satellite imagery (e.g., Sentinel-2) in optically shallow waters.

1. The Physical Model

Unlike empirical models (like Lyzenga or Stumpf) which require in-situ calibration data, this model relies on inverting the radiative transfer equation. The reflectance measured by the satellite ($R_{rs}$) is modeled as:

R_{rs}(\lambda) \approx R_{\infty}(\lambda) + C \cdot \rho_b(\lambda) \cdot e^{-2 K_d(\lambda) \cdot z}

Where:

2. How the "Lite" Algorithm Works

Instead of solving for all Inherent Optical Properties (IOPs) pixel by pixel (which is computationally expensive and unstable without hyperspectral bands), this "Lite" version makes specific assumptions:

  1. Constant $K_d$: It assumes water turbidity is relatively homogeneous across the scene. The user defines an average $K_d$ value for each band.
  2. Bottom Mix: The bottom is not a single material but a linear combination of Sand and Vegetation. The model searches for the best mix factor $w$ (from 0 to 1) for each pixel.

For each pixel, the algorithm:

3. Input Parameters

Parameter Description Typical Values
Input Mode You can choose between:
  • Separate Files: Load distinct GeoTIFFs for Blue, Green, Red bands (common for Sentinel-2/Landsat downloads).
  • Single Multi-band: Load one GeoTIFF containing multiple layers and map them using the dropdowns (common for Planet/Maxar or drone data).
Switch via Radio Buttons
Rrs_490, 560, 665 Blue, Green, Red bands. Must be atmospherically corrected and preferably deglinted. GeoTIFF Files
Kd ($m^{-1}$) Light attenuation. Higher values indicate more turbid water. Blue: 0.05-0.15
Green: 0.05-0.1
Red: 0.2-0.4
Albedo (ρ) Wet bottom reflectance. Sand: 0.2-0.4
Vegetation: 0.05-0.15

4. Operational Tips

If you get noisy results or incorrect depths: