How to measure canopy height and volume using QGIS (Drones in Agriculture series, 6/7)
HTML-код
- Опубликовано: 24 фев 2021
- 1:42 Determining canopy elevation (with canopy-specific digital surface models)
3:26 Determining soil elevation (soil-specific digital surface models)
4:45 Extracting elevation information (writing to attribute table)
7:33 Canopy area extraction
9:16 Calculating height and volume based on canopy/soil elevation and area
13:27 Checking results
16:39 Vegetation index summary statistics and comparisons
If you found these methods useful, please consider citing our recent paper in the journal Remote Sensing:
Parker, T. A., Palkovic, A., & Gepts, P. (2020). Determining the Genetic Control of Common Bean Early-Growth Rate Using Unmanned Aerial Vehicles. Remote Sensing, 12(11), 1748.
![[Paris Olympics] Performance by Australia's Olympic Break-dancer Rachael Gunn breaks the internet](http://i.ytimg.com/vi/i1rl1KhKIjE/mqdefault.jpg)
I seriously don't have enough words to thank you.
I'm glad the videos have been useful for you!
If you ever write a publication using the methods, consider dropping a citation to our citation where they were described, Parker et al. 2020 Remote Sensing, 12(11), 1748!
Well explained, Sir - thank you for this superb series, it is of tremendous help for me trying my leg in ag UAV monitoring. Would be nice to be able to watch more content from you!
Thanks Péter, I'm glad it was useful for you! Planning to put out a few more videos soon, but I always get caught up in other things!!
Very nice work and greatly helpful. Thanks for sharing :)
Thanks Ahsan!
@Travis Parker, Hi Travis, Could you please explain how to get DSM of plots?
I have a drone image consisting of 7 bands (RGB, NIR, SWIR,...). Can I calculate/estimate plot volume in the absence of LIDAR data? If yes how?
really really useful
Thank you! I am glad it was useful for you!
Excellent
Hi Travis, and thank you for sharing the method. I found your videos super interesting, and thanks for that. I tried to apply it on forest (area partly covered by forest), used grid 50X50m, but he "Mean Canopy Height (m)" for some grid cells got negative meaning. Cannot say why that happened so. Could the terrain be the reason for that? It is not flat and even. "Canopy volume," calculated using the Mean Canopy Height, also got the negative meaning in the same grid cells. Talking about the canopy volume, I have another question: when applied to forest area, what your call "canopy volume" it is actually forest stand volume (including space in between the trees), while canopy volume would be the volume of the upper part of trees (i.e. leaves), right? Did I get you correct? Thank you.
Hi, thanks for your comment. Yeah I don't go into a ton of detail about how to deal with complex terrain here, but one thing that helps is to use a grid of smaller polygons. If you are dealing with a deciduous forest where trees lose there leaves, the best thing to do is fly in winter (with ground control points) and then compare to summer when trees are leafy. Negative values mean that your canopy is on average lower than the forest floor or other non-canopy areas. As far as canopy volume vs. forest stand volume, that might be, I think it depends on the resolution of your orthomosaic, maybe among other things
Hello Travis,
I wanna to know the values we get after calculating plant height are in feets or cm?
Hey! The defaults in Pix4d and QGIS are all in meters, so your results should be in meters. If you want to check some individual points in your DSMs, drop them into QGIS and you can use the "identify features" tool to check some examples. To access that tool, click the symbol at the top of QGIS with a blue circle with letter "i" in it and cursor, or alternatively click Ctrl + Shift + I (if you're using Windows). Then click some points in your field. The results show up in the "Identify Results" panel (bottom right, for me). The difference in the values between a roof of a car to the ground next to it (roof elevation - ground elevation) should have a value of something like 1.5, representing 1.5 meters difference between them, for example. I hope that helps!
Hola!! gracias por tus videos! 1. cual fue la altura de vuelo y 2. el tipo de dron o cámara usada? Muchas gracias!!
Hola! Gracias por mirar! 1. La altura de los vuelos en estos videos es entre 15-20m, 2. El dron de este video fue un DJI Phantom 4, pero usualmente uso un Matrice 100 con el Zenmuse X3 o Zenmuse XT y MicaSense RedEdge-M (multiespectral)!
Pero los metodos funcionan con cualquiera altura y muchos tipos de dron y camara!
@@travisparkerplantscience Hola muchas gracias! muy amable por sus respuestas, me acabo de suscribir, excelente material!
@@eddisonjose Gracias por suscribirte y por los comentarios!
Great method for calculating the canopy area and volume, big thumb. Did you try the point cloud map to get higher accuracy?
Hi, yeah I have tried methods with the point clouds, but there are typically a lot of extraneous points that are artifacts. The smoothing in the DSMs can actually be really useful in eliminating some of that noise. But by all means, if working with the point clouds is more useful for your application, go for it!!
good job
Hello Travis, thanks for a great video. I would like to ask if it is possible to compute canopy volume using available DSM and DTM online like Copernicus GLO-30 DSM and FABDEM DTM? If it is, what could be the possible methods? Thank you!
Hi Collen, it certainly might be possible, although I haven't done that myself. If you do this kind of thing, I would strongly recommend using the same data source for DSMs and DTMs, so that you don't have imprecision in their overlap and whatnot. You could certainly try loading those into QGIS, potentially cropping the outputs, using the raster calculator, and then extracting data. I can't guarantee anything though since I usually am using UAV-collected data!
@@travisparkerplantscience thank you! I'm just so worried because our evaluators suggested to use these data to derive a canopy volume but I also haven't tried it yet. Our topic was all about assessing the mangrove damages brought about by a typhoon here in the Philippines and they have also suggested to correlate it with the canopy volume that's why I'm really worried on how to derive canopy volume with these data. Anyway, thank you for the information, it did answer my question thank you!
Hi Travis, Thank you for these great videos.
Did you use GCPs when generating the DSM? Can you get accurate height information if you don't use GCPs and rely only on the sensor geotagging?
Hi Jose, thanks for your comment! I didn't use GCPs, but you could. The images are already geotagged, because there is a GPS unit on all standard drone cameras, so Pix4D automatically determines the scale of the images and the DSMs and other outputs are automatically georeferenced. So basically you could drop the orthomosaics into QGIS and then click the option for adding a basemap (e.g. Google maps, etc.) and the orthomosaics and DSMs show up in the correct positions. In my experience, measuring plant canopy heights and other things relative to the soil is far more accurate than comparing between flight dates with GCPs, so I typically bypass use of GCPs altogether. I hope that helps!
I used '(Threshold < 0.5)' as denominator for sDSM. However, the sDSM values were higher than the cDSM. How to calculate canopy height in such case? The images were acquired from corn trials.
Hi, I would use the information tool (little "i" symbol) to click around your field at areas of interest to check on that. I also use median soil values, rather than averages, in case you have furrows. But that would be a very strange pattern to have soil higher than canopies, I would check everything very carefully. If the plants are truly tiny, maybe < 3 weeks old, they might be too small to display properly, otherwise I would just thoroughly check everything
Hello travis
Can you please make a video calculating crop biomass using the terrain model
Thank you
The volume estimates from this video are often very predictive of biomass, but there is no way to directly measure biomass (weight) from the air. So you have to harvest and weigh, and compare how those data match with your volumes. We have seen some excellent correlations for this kind of thing in alfalfa, with R^2 values of 0.9 or greater
Hi, Thx. for the interesting video !. How did you get the Grid layer ?
Hi David, check out this video: ruclips.net/video/SQbq2cxAk7k/видео.html
Thx. Travis ! Can you pls. let me know what is the recomended order of viewing your QGIS clips ?
@@davidbehr7865 The playlist is here: ruclips.net/p/PLifbxiLZAb-QHX2QjBqCNKysSHM3gPuTX
...and the QGIS videos go from videos 3-7, although I recommend all of them!
nice
Good day, sir.
I'm a geodetic engineering student. I humbly ask for your kind assistance regarding my thesis entitled “Estimating tree height using UAS Images and Photogrammetric 3D Point Clouds. I’m very much interested in estimating tree height using drone imagery. Is there way to estimate tree using 3D Point Clouds? And classify them into classes such as ground, non-ground, man-made, and vegetation?
Thank you for your time. I hope will give this comment a positive response. Thank you very much.
Hi Marie, the methods I usually use involve pulling the height data from raster files (DSMs) rather than point clouds, so that is what I would do in your situation, if you have raster DSMs from something like Pix4D. Certainly all the data required to do these calculations would be found in your point clouds, if they have been classified into soil vs. tree based on color, for example. Would you be able to generate raster DSMs of your fields/orchards?
Hi Ma'am. What processing/orthomosaicking software are you using? If you can generate 3d point clouds using UAS images, then for sure you could also generate DSM and/or DTM. We are also using drone images for our farm applications.
Thanks for sharing your acknowledgment,
I have some questions about calculating canopy height.
1) Can I understand sDSM same as DTM (Digital Terrain Model)?
2) Do you have any reason for choosing a median value for sDSM?
Great questions!
1) sDSM is not the same as DTM. In the sDSM, a pixel where there is plant canopy would be "NA", or no data. In a DTM, the value would be a numeric value of where soil was predicted to be under the canopy. So there shouldn't really be "NA"s in a DTM. The soil pixel values should be the same in both.
2) Median value or mean could be used. In my fields we often have big flat beds with furrows or grooves in between, and I care about the big flat part on top, so I tend to use median.
@@travisparkerplantscience Thank you for answering! It helped a lot!
@@user-jo9bt4is1y You're welcome!
Hello Travis, Thanks for these great videos. I do drone surveying for the forest section, all these videos you have posted are related to plantation, Can i use same methods for forestry data?
Hi Madhusai, yes, the methods should be applicable to forestry. At some point you could run into complicating factors such as the density of your canopies, clustering of trees, complex terrain, etc., but that is to be expected. My recommendation would be to capture some images, get some ground truth data (if possible), and see how they correlate. If needed, you can make fine-tune adjustments from there
Hello Madhusai, just wondering if you have used this method in forestry. I am trying the same.
I just found your channel @Travis Parker.
Thanks for sharing this great video class. I loved your content to understant more about drone image processing.
I have some questions about it. Is there any way to contact you by email?
Thanks in advance.
New suscriber, greetings from Perú.
Hi Harry, feel free to contact me: trparker (arroba) ucdavis (punto) edu
Hi...can u please tell me from where I download DSM Data for Indian cities??
Como vc aprendeu português??
Sou de California, e tem muitas pessoas que falam espanhol aqui, entao aprendi espanhol. No laboratorio aqui, tivemos muitos brazileiros, ate 5-6 durante um verao, entao aprendi portugues com duolingo, podcasts, youtube, e sem duvida, com amigos!
Hi, is there any way to convert single drone image of a farm, to DSM/ DEM ?
Hi, unfortunately there isn't. The way that the DSM is generated is that it can determine spatial relationships based on the differences in the geometry between images, kind of like how it is sometimes easier to judge the distance of an object if you are moving, or use both eyes instead of one
When I run multiple zonal statistics they all separate into their own layers. How do you get them to append to the same Attribute Table?
Interesting... to be clear, they aren't in different columns in the same attribute table? Instead, they are making new vector shapefile layers or something? If so, I haven't heard of this :/
@@travisparkerplantscience Exactly. It’s not arraign to the attribute table of the vector that I run the statistics on. It’s creating an entirely new layer and putting them in that. Then if I run further statistics, it creates another layer still.
@@s7ewie That's a really strange and frustrating issue. I haven't heard of it before. I would recommend checking that there isn't any box checked in the zonal statistics tool window saying "write to new layer" or similar (granted, I don't have any box like this on my device), but otherwise it might be worth doing a small test on a different device to see if you get the same issue. If you don't, you could carefully check the vector grid and zonal statistics steps to see if there are discrepancies between them. If you do have the same issue between devices, you could try watching through this video very closely again to see if there are clear differences between what you see here and what you have in front of you, but otherwise I am not really sure since I haven't heard of the problem before. Sorry to not be more helpful :/
same here!! did you sort it out?
@@capsulaproducciones3370 not really no. Although if you run the next set of statistics using the previous zonal static layer as the vector layer, it should just create another later and all the data onto the existing append the data onto the existing table.
Hi Travis. Do you by any chance have an email address that I can contact you on? I’ve got a project and could use your expertise. Might have to send you the project files.
Hi Travis, when I copy and paste the attribute table from qgis to excel, automatically excel shows thousand instead of hudreds, for example: Qgis number: 713.45263671875 when I paste in excel looks like this: 71,345,263,671,875
I already tried to fix it from excel but coudn`t find the way to do it.
Hi, that's definitely a strange problem... In your example, the decimal is after the three but the comma is before the three... that is how it is showing up for you in excel? I wonder if it has to do with using a Spanish language version of Excel, since sometimes commas are used en vez de decimals? You could try 1. changing the formatting options in Excel (general vs. text vs. number, etc.), 2. pasting values only (possibly in combination with different formatting options), 3. try pasting into Google Sheets, possibly adjusting the formatting or using Ctrl+Shift+V to paste values only, or if needed 4. Use the Ctrl+F function to remove the commas, then re-insert with an equation, if possibly (not really recommended, depending on range of values you are working with)