I am a farmer in India, with an 8 hectare plot. It is quite wooded. My objective is to map the boundaries, contours, internal roads etc., and then prepare spatial database of various fruit trees planted in the farm. I plan to keep this updated in the future.
Since I am on a very limited budget, my question is this:
what is the minimum equipment/s that I have to buy to do cm level survey grade mapping? Can someone list these for me, please?
Can I buy two M2 receivers plus helical antennas and use one as a Base and the second one as Rover?
I dont mind doing solely PPK, if that will save me money on buying LoRa. I do not have much knowledge on mapping or GPS technology, so will appreciate any and all help in pointing me in the right direction. Since there are lots of tress in the plot, I am worried about losing gps fix often, hence I am gravitated towards M2.
thanks
Roshan
What makes you believe you need centimeter-level accuracy?
I would propose that you start with a drone first and that if you determine you really need more accuracy then look into improving the drone mapping with survey equipment.
First reason for this is the sheer amount of information that you will collect and the visual nature of it. We routinely perform analysis on large tracts of land to develop conceptual designs that eventually lead to final development. Terrestrial survey and centimeter-level accuracy is not required for land development, repetition is more important. What it is required for is other Civil characteristics such as underground utilities, drainage systems (not drainage analysis) and structures that must be built with a tight tolerance.
The second reasoning would be the frequency at which you can collect the data that then determines the amount of data you can collect. In my experience it would take approximately two days to sufficiently survey 8 hectares with point collection. Locating property lines, feature grade breaks, structures and a 10 meter grid of the topography is quite time consuming. Even if done to that level there is allot of missing information in between. If done correctly it would take approximately 15 minutes of air time to collect 100,000x more data and the backend is much faster as well because the work of creating a map, contours and surfaces is done as part of the photogrammetry process. This leaves you with time to do other things while the processing is taking place instead of sitting in front of the computer for hours trying to create the geometry yourself.
What does centimeter-level point accuracy do for you when you have to collect it sparsely? The beauty of a drone map and point cloud is that you are collecting data at a relative centimeter level instead of on a grid and feature lines. At the end of the day things like drainage analysis, volume measurements and design to existing features are going to be much more accurate with a drone map than what they are with single points collected in the field and it is very unlikely that you will have to return because you missed something.
If you determine that you really do need centimeter-level accuracy then hire a Surveyor to set Ground Control Points (GCP’s) for your drone flight. Make them permanent monuments on your site for perpetual use.
If you are good with sub-meter global accuracy and centimeter relative accuracy then create GCP’s from the first map you fly and all future flight will be relative within centimeters.
Thanks for taking the time to respond to my query, this is very much appreciated. You have really set me off in a direction that I haven’t given much thought to, basically for the following reasons:
I am not much familiar with the process of drone mapping, hence was not aware of its capabilities (I was under the impression that UAV surveys are effective only in areas with very scarce tree cover, which is not my case), and
A drone is another major piece of kit that I have to invest in (but coming to think of it, probably I can find one to rent from a photographer and can hook it up with the M2).
I know that manual work is laborious and will take several days to complete (my farm is with small hillocks and valleys) it in some detail. Manual surveying is unavoidable for me because one of my main objective is to take the locations of my 2000 odd trees in the farm, and then geotag it with the age, type, yield and such details including photographs. I don’t think this can be achieved with a drone alone, but would certainly require manual collection of point data from the location of each tree. However, I am not sure whether I need cm level accuracy for this purpose or not. But I certainly can see the benefit of using drone mapping to collect overall data which will give me a far better starting point - thanks very much.
@roshanahmd, thank you for the thorough description of your use case. @michaelL thank you for the very reasonable proposition.
Indeed, I can agree with Michael that the drone mapping will definitely ease the job in such tasks as mapping the boundaries, contours and internal roads on such a scale. You can use the drone equipped with a camera with hot-shoe access if it is possible to borrow one. With a properly collected dataset and M2 on that drone, it should be easy to achieve the centimeter accuracy of your orthophoto plan. Also, you can place GCPs in PPK mode with two M2s. I recommend checking this thread on that.
Regarding the tree location task, I’d say that it might be a relatively challenging in the dense forests even for multi-band receivers like M2 and RS2 because they would still encounter the signal deflection from the trees from time to time. Also, during PPK it is a bit more tricky to predict whether you will achieve a fix solution on the post-processing during the survey itself. Plus, as Michael noted, you need to decide whether you really need to have their position to be centimeter-accurate.
Additionally, I need to clarify, that the LoRa signal as any radio signal is prone to deflection too, that is why using LoRa in the forest might be not really stable. As an alternative, you can send the corrections over the Internet through a free caster like RTK2Go.
A pair of RS2 will be way more suitable for the surveying tasks as it is built for such applications. It could be easily mounted onto the tripod and the survey pole, it is fully autonomous as it has a robust battery and the shock-, dust- and waterproof case. The built-in antenna does not require an additional ground plane. We recommend using RS2 in survey projects which should cover areas of such size due to it is already a complete package with the list of pros I’ve listed above. The M2 is capable of doing the same job, but it involves way more setup preparations like power supply, fastening to the survey pole, preparation of the ground plane and etc.
Summing that up I’d say that the minimal setup would look like this:
I agree that if you truly need to tag trunks that the drone will have trouble modeling the ground, but it works perfectly well for locating trees. I have worked with an Arborist a couple of times to not only provide map plots of trees, but to identify Oak Wilt and plot a GPS track for them to efficiently find and treat those trees. In most occasions the trunk is in the approximate center of the canopy, but there are always exceptions as is with the case of Live Oaks that can lean over time and go off in several directions. One thing to consider though is that if the area is that heavily wooded that GPS ground surveying will also not work. You will need a total station or robotic instrument to shoot with a laser underneath the canopy. In addition regarding the topographic survey it is possible to clean up the 3D point cloud to create a Digital Terrain Model (DTM) of the ground that will be very accurate.
As for the drone being an additional cost, it would not be any more expensive of an initial investment if you consider the entire package for land surveying that would be required and the software you would need in order to process that land survey. Most cloud photogrammetry processing solutions offer some level of analysis, but a software that I would suggest as your base analysis and record keeping software would be the free GIS solution called QGIS.
As I stated earlier you can always get one-time GCP’s set and consider adding land survey equipment later. Renting the equipment and the learning curve to do what you intend to do would be more expensive than just having someone do a small part to get you going.
Drone does not work for forested area, trees of the understory layer do not appear. Species identification on drone maps can be very difficult.
Cm level accuracy (the fraction of a leaf size !) does not make sense for tree position and contour mapping. The method is setting primary reference points where the canopy is wide open using PPK GPS and setting secondary reference points under the canopy using a total station (two persons are needed). In that case tree will be mapped using the total station and set reference points. Road contours can be mapped either using the total station or using a Fish-eye objective on a camera fixed on a 3-4 m hand-held pole instead a drone. In that case, below-canopy referencing is made using the set control points. For tree mapping, I personally use two reach modules (single channel, post processing) and a total station. I tried the method of a camera on a pole with Agisoft Metashape to make a map but it is expensive in time and equipment (software + camera + computer (min 32 GB ram) + graphic card…). I would go only with a pair of M2 and a low-cost total station. Also consider a durable and efficient tree tagging system. A high resolution map is of little help for tree recognition in a forested area if there are no identifiable field markers (roads, parcels numbers, tree tags…). I invested in a metal stamping numbering head to make durable aluminium tree tags for tree management. On a very limited budget, i would classify by order of importance
A tree tagging system/ durable reference points for recognition in the field and future updates
A very coarse mapping with common GPS
Precision relative mapping using a total station
Georeferencing the relative precision map using two M2
