I need to check the similarity of two sets orthophoto from UAV/drone with different location and extent but in the same time frame. The data are in the form of indices (NDVI, LCI, NDRE, and GNDVI). What is recommended method to do it? I consider to use Cohen's d in R. But is it appropriate?
I also consider diffeR package in R as well. But this package is only for comparing two data in the same extent, isn't it?
Maybe a screenshot of the two extents and locations. As written, since any two flights will never be exactly the same, the 'how different in which ways is left unanswered for the reader. One could be contained within the other, they could over lap slightly. or they could be half a continent away from each other in a different biome and still be in the same time frame. Any comparative statistics would have to have some common sub-area and/or attribute present in both extents - the classic 'fixed, controlled, and measured'. And that attribute(s) should be a close proxy to some degree for both those indices and whatever you are trying to discern.
Thank you Michael John Patrick
Actually there is no overlap between the two rasters. They are in the same biome as well. Please fins the image attached
What is the ground resolution ( cell size ) your imagery? And the nature of the target ( Crops ( what stage of growth?)? Trees? What's the ground distance between their centroids, they look to be in the neighborhood.
The cell size is around 8x8 cm. It was an oil palm plantation. The age of the trees were various, but we divided into two classes based on its planting year: before 2013, and after 2013. These two images are only sample, actually we have several other blocks. To be more specific, we need to monitor the performance of this area after it was burnt by fire at the end of 2013. There are some other approaches, and UAV is one of them.
So we have divided the blocks based on two parameters:
1. Planting year or the age of the plants -> 2 classes
2. Burnt status (whether it is burnt or not) -> 2 classes
From those division, then we get 4 categories:
A. Planted before 2013, burnt
B. Planted before 2013, not burnt
C. Planted after 2013, burnt
D. Planted after 2013, not burnt
In order to have apple-to-apple comparison, we will compare categories with the same condition. Or in this case is the same age. In short, A compared to B, and C compared to D.
The distance between the two images in the attachment was about 1.3 km. But, as I stated above, these are just sample. We have several other images, located at the different places, no overlap between them. Maybe we will merge images in the same categories to become one raster file, then compare it with the correspondent categories.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Artificial Intelligence