There are various methods/ algorithms used to extract water quality data from satellite images. Most of them are applied for large water bodies. I am looking for a suitable method to extract water quality data from ponds and lakes using Landsat 8 satellite images.
"Water quality" is a broad category. You may be able to retrieve optically-active variables of "water quality" such as turbidity, surface temperature and perhaps algae concentrations. But you will not be able to measure any of the chemical constituents, of course. Unfortunately for measuring inland waters, Landsat 8 spatial resolution is 30 m for spectral band data and 15 m for panchromatic data, which limits you to relatively larger water bodies or channels. Some papers claim to have developed regressions between Landsat data and macronutrients nitrogen and phosphorus, but these must be correlational, not causal, and their usefulness would be limited to very productive (eutrophic) settings where nutrient concentrations are closely related to optically active phenomena such as algal absorption or turbidity scattering.
Dear researcher
Thank you for your clear explanation. There are so many research papers developed algorithms to extract water quality data. They have used ENVI and Erdas software. I want to study the water quality (BOD, COD, pH, Ec, Chlorophyll a, Salinity, Nitrogen and Phosphorous concentration...) of small inland lakes and ponds using Arc GIS. Can you explain further regarding this?
I'm glad my earlier answer was clear; that's not always true :-). Yes, it is true: There are many research papers seeking algorithms to remotely sense many different water quality parameters. These are highly desired data, and remote sensing offers larger coverage at lower cost. However, the constituent has to have some kind of optical signature to be detected and quantified directly.
Of BOD, COD, pH, EC, Chlorophyll-a, Salinity, Nitrogen and Phosphorous concentration, only chlorophyll-a has any kind of direct remotely-sensed optical signature: The chlorophyll-a pigment absorbs light in the red (around 441 nm) and blue (around 676 nm) parts of the visible spectrum. All the rest of those listed are "colorless" and have to be inferred indirectly from measurements of water constituents that DO covary with optical "observables". Usually this is done by a laborious process of regressing optical measurements against in-situ measurements (the field measurements are the laborious part). Unfortunately, we can't go back in time and collect these field data to compare to archival LANDSAT pictures in your specific study systems.
Chlorophyll-a and turbidity can both be remotely sensed. Turbidity often reduces light transmission into deep water and thus limits algae (chlorophyll) production. If turbidity is low (10 ug/L), there may be enough algal uptake of N and P to affect them, but again you are not measuring N and P directly. High algal production (a) produces oxygen during the day but (b) causes respiration oxygen demand at night, affecting BOD, COD and pH. pH will be driven up during the day as algae remove CO2 (carbonic acid) from the water, and will decrease during the night as respiration increases CO2 in the water. EC and salinity are not affected. It is these indirect mechanisms that the journal articles rely on, along with MANY assumptions.
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts