Digital Image Colorimetry With API Ammonia Test

Dan_P

New member
Recently I asked a question about digital image colorimetry for reading colorimetric test results. (http://reefcentral.com/forums/showthread.php?t=2566871). Since then I looked more deeply into the subject and now report how digital image colorimetry can be used with the API ammonia test kit to detect levels of total ammonia below 0.1 ppm.

Digital image colorimetry involves taking a picture of the colored test sample and a blank under the same lighting conditions. An app, like Photoshop or ColorMeter, is needed to read the RGB values of the colored and blank solutions. For this test the R or red value is of interest because it is strongly absorbed by blue-green test solution. The deeper the blue-green color, the smaller the R value. The absorbance is calculated and compared to a calibration curve to determine the total ammonia concentration (see photo below). For best results, the solution should be clear and free of particulate matter. For this reason, the API test was modified.

The standard API test gives a very cloudy solution when the test sample is saltwater because the analytical reaction requires a high pH. This precipitate makes detecting low concentrations of ammonia impossible. To prevent the precipitate, about 0.1 mL of solid sodium citrate is dissoved in the 5 mL test sample. In addition only half the amount of each reagent is used and the development time is increased to 10 minutes. This gives a clear, precipitate free test solution. This protocol would make possible the use of a Hanna Checker in place of a camera to detect red light absorption.

The limit of detection, the lowest amount of ammonia this test procedure can detect is estimated to be about 0.05 ppm. Since only three samples were tested at each concentration, many more sample tests are needed to provide an accurate estimate of the limit of detection.

I will be repeating this study with API and Salifert phosphate tests and Salifert nitrate test.

API%20NH3%20Calib_zpssnpkyivv.png
 
Dan, did you check for the linearity of the camera used? Did you measure 10 times the standard deviation of a blank and compare that LoD to your stated LoD here? Also, are you using 32 bit color depth in the images to ensure plenty of decimals to prevent calculation rounding errors? No offense, but I'm highly skeptical of your values (but not of you or your work).
 
Dan, did you check for the linearity of the camera used?

Excellent question. Prudent action. I have not checked the linearity of my iPad camera. However, as long as any nonlinearity is relatively constant, i.e., I use the same camera and light source, I will be able to trend my water chemistry data and identify any major changes.

Did you measure 10 times the standard deviation of a blank and compare that LoD to your stated LoD here?

Ten standard deviations is probably a bit excessive for the work I am doing. Being 95% certain will be fine. The emperical approach I used can be found here:http://http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2556583/pdf/cbr29_s_pgs49.pdf

Also, are you using 32 bit color depth in the images to ensure plenty of decimals to prevent calculation rounding errors?

I am not sure what the iPad camera has.

No offense, but I'm highly skeptical of your values (but not of you or your work).

I am skeptical of the LOD myself. IF I recall, 21 measurememts are needed to establish it. I did three, so, let's take it as a provisional value.

Skepticism is a good thing. Thanks for your thoughts. They gave me some new ideas.

Dan
 
Dan very interesting stuff. I have been trying to work my contact at Hanna to see if they can get a unit made to read low NO3.

Dan while I think what your doing is great, I see two potential problems, one of them might not be a problem but it sort of relates to what NewReefer42 said.

First of all I am not sure if this is just a personal project or one that you want to see others be able to use. If it's the latter then the type of camera and settings etc are going to become critical and even then without known samples almost impossible for another person to do unless they duplicate your equipment or calibration experiments. That's why I hinted that a brand name TV colormeter might be the way to go since it is consistent across the board in it's settings. If your not interested in other people using your idea then no need to worry about any of that.

2) My really big concern is that your using an API reagents for this setup and I am pretty sure that they do not have the accuracy to read down to the 1ppm range. What I mean is that if you put in effluent with 1ppm or 0ppm into a test I am pretty sure that the color will be a fairly random shade of very light yellow and not repeatable. You really should look at using a test kit with reagents designed to read very low nitrates.

BTW the reason I mention Nitrates only is because for anything other than Nitrates and Mg I would suggest to eveyone to just get a Hanna Kit.
 
Last edited:
Dan very interesting stuff. I have been trying to work my contact at Hanna to see if they can get a unit made to read low NO3.

Dan while I think what your doing is great, I see two potential problems, one of them might not be a problem but it sort of relates to what NewReefer42 said.

Thanks for the comments. I am happy to share everything that I am doing, including all the shortcomings in my work. Just need a "forum" to download that type of info.

I already know API lower detection limits are too high, but the kits are so cheap, so, I can do lot's of experiments. Also, my pore water and skimmate study is looking at nutrients levels between 1-15 ppm, so API works for me. Also, I can't afford a Hanna for every chemical species I might want to measure when it dawned on me that any camera might a Hanna.

I like your discussion about Hanna and nitrate testing. I am going to look at Salifert PO4 and NO3 next to see what I can see and will share all.

Maybe we can get a test kit maker to put out a series of digital image colorimetry tests for those who might have a camera :)

Dan
 
Back
Top