Determining improvements in Landsat spectral sampling for inland water quality monitoring
Inland waters are optically complex and provide an ongoing challenge to effective water quality monitoring through remote sensing. Imaging satellites with spectral sampling designed for this task often have coarse spatial resolutions, preventing any capture of information from small lakes. Medium re...
Main Authors: | , |
---|---|
Format: | Article |
Language: | English |
Published: |
Elsevier
2020-06-01
|
Series: | Science of Remote Sensing |
Subjects: | |
Online Access: | http://www.sciencedirect.com/science/article/pii/S2666017220300043 |
id |
doaj-9c3d12d7b25949a39999cd071dd55982 |
---|---|
record_format |
Article |
spelling |
doaj-9c3d12d7b25949a39999cd071dd559822021-06-10T04:57:07ZengElsevierScience of Remote Sensing2666-01722020-06-011100005Determining improvements in Landsat spectral sampling for inland water quality monitoringRyan T. Ford0Anthony Vodacek1Corresponding author.; Chester F. Carlson Center for Imaging Science, Rochester Institute of Technology, Rochester, NY, USAChester F. Carlson Center for Imaging Science, Rochester Institute of Technology, Rochester, NY, USAInland waters are optically complex and provide an ongoing challenge to effective water quality monitoring through remote sensing. Imaging satellites with spectral sampling designed for this task often have coarse spatial resolutions, preventing any capture of information from small lakes. Medium resolution satellite systems such as Landsat 8 have the appropriate spatial resolution and sensitivity required to resolve these waterbodies, but the spectral sampling is not optimal. This work uses system simulation to explore potential changes to Landsat spectral sampling to determine if its ability to monitor inland waters could be improved. The HydroLight and MODTRAN radiative transfer models are used for simulation in a Look Up Table and spectrum matching approach to provide maximum flexibility intesting spectral sampling scenarios. To isolate the testing to the impacts of spectral sampling, all simulations were performed based on the known system noise characteristics of Landsat 8. Spectral sampling changes tested include the addition of yellow and red edge spectral bands as well as conversion to an imaging spectrometer. Simulated spectra of inland waters undergoing a cyanobacteria bloom, including atmospheric effects and sensor noise, were implemented with the Look-Up-Table retrieval process to extract estimated concentrations of waterbody components. The retrieval accuracy of each potential system is compared to that of a modeled Landsat 8 baseline. All potential systems show an increase of retrieval accuracy over the baseline. The best performing system design is an imaging spectrometer, followed by the addition of both a yellow and red edge band simultaneously, and the addition of either band individually. Testing also demonstrates that resampling an imaging spectrometer with 20 nm spectral resolution to the Landsat 8 band responses produces outputs matching those available from Landsat 8. Our results indicate that future Landsat missions should aim to add as much spectral sampling as is feasible, while maintaining at least the same sensitivity. The minimum change to improve water quality monitoring capability is the addition of a red edge spectral band.http://www.sciencedirect.com/science/article/pii/S2666017220300043Remote sensingLandsatSpectral samplingLakesCyanobacteriaWater quality monitoring |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Ryan T. Ford Anthony Vodacek |
spellingShingle |
Ryan T. Ford Anthony Vodacek Determining improvements in Landsat spectral sampling for inland water quality monitoring Science of Remote Sensing Remote sensing Landsat Spectral sampling Lakes Cyanobacteria Water quality monitoring |
author_facet |
Ryan T. Ford Anthony Vodacek |
author_sort |
Ryan T. Ford |
title |
Determining improvements in Landsat spectral sampling for inland water quality monitoring |
title_short |
Determining improvements in Landsat spectral sampling for inland water quality monitoring |
title_full |
Determining improvements in Landsat spectral sampling for inland water quality monitoring |
title_fullStr |
Determining improvements in Landsat spectral sampling for inland water quality monitoring |
title_full_unstemmed |
Determining improvements in Landsat spectral sampling for inland water quality monitoring |
title_sort |
determining improvements in landsat spectral sampling for inland water quality monitoring |
publisher |
Elsevier |
series |
Science of Remote Sensing |
issn |
2666-0172 |
publishDate |
2020-06-01 |
description |
Inland waters are optically complex and provide an ongoing challenge to effective water quality monitoring through remote sensing. Imaging satellites with spectral sampling designed for this task often have coarse spatial resolutions, preventing any capture of information from small lakes. Medium resolution satellite systems such as Landsat 8 have the appropriate spatial resolution and sensitivity required to resolve these waterbodies, but the spectral sampling is not optimal. This work uses system simulation to explore potential changes to Landsat spectral sampling to determine if its ability to monitor inland waters could be improved. The HydroLight and MODTRAN radiative transfer models are used for simulation in a Look Up Table and spectrum matching approach to provide maximum flexibility intesting spectral sampling scenarios. To isolate the testing to the impacts of spectral sampling, all simulations were performed based on the known system noise characteristics of Landsat 8. Spectral sampling changes tested include the addition of yellow and red edge spectral bands as well as conversion to an imaging spectrometer. Simulated spectra of inland waters undergoing a cyanobacteria bloom, including atmospheric effects and sensor noise, were implemented with the Look-Up-Table retrieval process to extract estimated concentrations of waterbody components. The retrieval accuracy of each potential system is compared to that of a modeled Landsat 8 baseline. All potential systems show an increase of retrieval accuracy over the baseline. The best performing system design is an imaging spectrometer, followed by the addition of both a yellow and red edge band simultaneously, and the addition of either band individually. Testing also demonstrates that resampling an imaging spectrometer with 20 nm spectral resolution to the Landsat 8 band responses produces outputs matching those available from Landsat 8. Our results indicate that future Landsat missions should aim to add as much spectral sampling as is feasible, while maintaining at least the same sensitivity. The minimum change to improve water quality monitoring capability is the addition of a red edge spectral band. |
topic |
Remote sensing Landsat Spectral sampling Lakes Cyanobacteria Water quality monitoring |
url |
http://www.sciencedirect.com/science/article/pii/S2666017220300043 |
work_keys_str_mv |
AT ryantford determiningimprovementsinlandsatspectralsamplingforinlandwaterqualitymonitoring AT anthonyvodacek determiningimprovementsinlandsatspectralsamplingforinlandwaterqualitymonitoring |
_version_ |
1721385595330101248 |