Ragweed survey

It's the new season

Saying the word "ragweed" results in mixed emotions. Specialists said that today's technology, such as walking, cycling and helicopter surveys are not efficient, but no one could say what would be the better method or technology. You always need proof to prove things that are and the things that aren't. 

With our experience in multispectral imaging we didn't quite understand how you can spot a certain plant or the amount of a certain plant in a parcel from a satellite image, when sensor's sold today lack the resolution for this even when used from a small airplane at a short distance. 

Survey

With these feelings we went to the border. Corn fields, harvested crops and sunfields covered the area surrounded by typical dirt roads. Our device was a Sony NEX 7 with a 24 megapixel RGB light-domain sensor and a MultiSpek NDVI attached to our medium copter. 

Test flight

We started by recognising our "weed guy" on the ground and the started taking a series of photos at the corner of a small parcel with 5m ascends up to 50 meters. Our goal was to find out how high we can fly while our cameras are still able to differentiate between plants on the ground. How each sort looked on the photo, with what color, tone, resolution is it visible? 

Up top you can see the variety of species in the corn! On the image below we highlighted the blossoming ragweed's typical yellowish crop-nodes. As we go higher, ragweed's thin leaves start to disappear first. Corn's leaves are the longest that can be seen clearly divided. The dried out leave is over the circled area on the top of the picture. 

5m

20m
When you zoom in, the best differentiator is the smudged, dark-green color of ragweed. All the other plants are typically lighter colored. 

40m
At this height you start to see those typical "grayish" spots in corn lines. In the middle of the field they are visible in rows. When we went in to check in person, they were clearly ragweed. 

On the same image shot of the stubble at 5 meters with the infrared domain NDVI camera you can clearly see the fresh ragweed, but you can also see that where ragweed is mixed with other plants it's basically impossible to differentiate.  

We concluded from our tests that when our "weed guy" indentified the plant in the visible domain, it can be further identified with the given color and tone. 

The "washer" - mixed test area

With our experience so far we could confidently spot the smudged greenish areas from up to 100 meters of height, with 100% success of what we found was truly ragweed. To prove this we let our copter loose over an area of mixed plants, so called "washer" zone, because this northern area was once covered with concrete and used to wash the farming machines. 

We found the following using typical points. The lea-land next to point4 looked slightly infected, when in fact ragweed was there in masses. There are traditionally clean tables such as corn and wheat and we thought that sunflower was always problematic. It was also clear that the area where roads and parcels meet was always infected!

The "sunflower" - a concrete table surveyed

With our findings about sunflowers from the mixed area we went ahead to apply our new knowledge on a 16 acre sunflower field, which was deemed 40% infected by a bicycle survey. We found this: 

Ragweed until the eye can see! Ragwee is clearly visible on the images and can't be mistaken with the other furrow-weed, even distinct sunflower heads can be identified. If someone asks how we can prove that the images were shot here, the figure on the right shows the image's location in space. They follow each other with precise overlaps. It starts on the ground with ragweed identified by our "weed-guy" and then it continues. 

The following picture was categorized by color and undeniably shows how wrong the cyclist's 40% was. 

Results

We proved that we can genuinely show if an area is infected by ragweed or not by visible-domain pictures that show infection with mathematical precision.