RedEdge reflectance values seem relatively ok but absolutely wrong

PIX4Dmapper PIX4Dmapper Questions/Troubleshooting
1

Hi all –

I flew over a mixed forest in Alaska with RedEdge and produced a 5-band reflectance image that was calibrated using calibration panel. My result looks generally ok and NDVI is ok as judged visually. But, the absolute reflectance values seem very, very low. See below (true color image, in theory). This spectral profile is of a bright, broadleaf tree where I’d expect NIR reflectance above 40% most likely. Even more strange, the white lines on the road only show visible reflectance of between 1% and 3%.

Anyone have any thoughts on this? Am I just off by a power of 10 or something or is there a more nefarious issue? For reference, my calibration panel reflectance panel values are shown on the panel as, e.g., 49.2 but I enter that into P4D as 0.492. I think I’m doing that right but maybe not…

Thanks!

Mike

 

2

Hi Mike, The reflectance factors you are using is okay. The values should be between 0-1. However, you can contact the manufacturer from where you bought the target and they can give you the vales. They will just need the serial number of the target.

Low reflectance values can result if the flight was carried under cloudy conditions, or just the targets were taken with different light conditions. It is always because of some issue with radiometric calibration. I have replied to your support ticket. We can investigate more when we get your images and quality report.

3

To add closure here. Momtanu examined my images in an interesting way (shown below) and demonstrated that the light conditions under which the calibration target was imaged were pretty different than the light conditions of the flight even though they were separated by only 15 minutes or so. The curse of patchy clouds. Ultimately, I got somewhat better results by running the radiometric calibration with Camera and Sun Irradiance correction but without using the ground target. My final spectra look much better but it’s not a sure thing as to whether I have usable reflectance values throughout the image.

 

Figure shows my reflectance target values (bottom left of each plot) as outlier.

4

Hi Mike,

The calibration target enables to have an absolute reference, which would allow you to get absolute reflectance values and make it possible to compare data coming from several cameras. However, it is not necessary if you would like to compare data from the same camera (e.g. several different missions acquired with this camera), as there is the sunshine sensor. The latter serves as a reference for the camera. In this case, the reflectance is given on a relative scale, and it is necessary to process with the irradiance data from the sunshine sensor.

Therefore, it is possible to process projects without using a calibration target. These can then be compared with projects captured with the same camera. However, if you have several cameras from which you would like to compare data, or if you require absolute reflectance values, it is necessary to have a radiometric calibration target.

Radiometric targets make the calibration more accurate. Does you reflectance values seem better now?

5

Hi. Can i know how to see the spectral reflectance value for each tree after i have generated the ndvi map?

6

@carlmarlyana You can draw regions in the NDVI map (https://support.pix4d.com/hc/en-us/articles/204769419-How-to-create-a-Region-with-the-Index-Calculator), but the issue is in Pix4DMapper is that you can draw only one region at a time to get all the statistical values (mean, median, min, max). I would say using the zonal statistics option in QGIS would be a better workflow (Zonal Statistics Plugin)