tmart 2.4.11

Modelling and correcting for the adjacency effect in aquatic remote sensing

T-Mart: Topography-adjusted Monte-carlo Adjacency-effect Radiative Transfer Code

Description

T-Mart solves radiative transfer in a 3D surface-atmosphere system. It supports customizable surface models and enables simulation and correction for the adjacency effect (AE) in optical aquatic remote sensing. AE correction substantially improves satellite-based retrieval of water-leaving reflectance in nearshore environments (Wu et al., 2024).

Links

Home page: https://github.com/yulunwu8/tmart

User guide: https://tmart-rtm.github.io

Publications

Wu, Y., Knudby, A., & Lapen, D. (2023). Topography-adjusted Monte Carlo simulation of the adjacency effect in remote sensing of coastal and inland waters. Journal of Quantitative Spectroscopy and Radiative Transfer, 108589. https://doi.org/10.1016/j.jqsrt.2023.108589

Wu, Y., Knudby, A., Pahlevan, N., Lapen, D., & Zeng, C. (2024). Sensor-generic adjacency-effect correction for remote sensing of coastal and inland waters. Remote Sensing of Environment, 315, 114433. https://doi.org/10.1016/j.rse.2024.114433

Installation

1 - Create a conda environment and activate it:

conda create --name tmart python=3.9
conda activate tmart

2 - Install dependencies:

conda install -c conda-forge Py6S

3 - Install tmart:

pip3 install tmart

Quick start: adjacency-effect correction

T-Mart supports AE correction for Sentinel-2 MSI and Landsat 8/9 OLI/OLI-2 products. Correction is performed directly on level-1 products and can be followed by any amtospheric correction tools.

Minimal input:

import tmart
file = 'user/test/S2A_MSIL1C_20160812T143752_N0204_R096_T20MKB_20160812T143749.SAFE'

# NASA EarthData Credentials, OB.DAAC Data Access needs to be approved
username = 'abcdef'
password = '123456'

# T-Mart uses multiprocessing, which needs to be wrapped in 'if __name__ == "__main__":' for Windows systems. This is optional for Unix-based systems
if __name__ == "__main__":
    tmart.AEC.run(file, username, password)

The tool takes approximately 20 min to process a Landsat 8/9 scene and 30 min for a Sentinel-2 scene on an eight-core personal computer. See Instruction - Adjacency-Effect Correction for detailed instructions.

Known issue(s)

rasterio version 1.4.x leads to unprojected S2 image files when performing AE correction in Mac’s Terminal. You can specify rasterio version 1.3.9 in Installation step 2 to get around this:

conda install -c conda-forge Py6S rasterio==1.3.9

Others

T-Mart can calculate reflectances of various units, see Table 1 in Wu et al. (2023) for examples.

For questions and suggestions (which I'm always open to!), please open an issue or email Yulun at [e-mail]