Digitising rural land records, one drone at a time | Hacker News Reader

Digitising rural land records, one drone at a time

bloombergquint.com

65 pointsharias9 comments

Wow! This is brilliant.

The other good news is that with Indian Government's de-regularization of geospatial data[1] up to 1-meter resolution; the initial mapping of topographical divides can be done much faster. Start with Satellite imagery and then zoom in for finer details with drones.

We work with farm owners in remote corners of India, there are areas where farmlands are sandwiched between natural forest. So, some interesting problems crops up -- prevent animals from destroying farms but also help them not trap in electric (solar) fences. If cows dies, your business is pretty much dead.

I really hope ISRO's data becomes easily for public access through an easy to use API, without having to navigate through their paperworks and bureaucracy. Last time I checked, some of their data (they say is for Public) comes in a downloadable that goes through a Flash Player (so, no longer works).

Disclosure: The link is from an article I wrote.

1. https://valinor.earth/blog/india-geospatial-guidelines-2021/

This made me smile:

If the project succeeds, boundaries currently defined by a tree or a stone, will stand transformed into something more tangible — a record on paper.

I believe that in Japan, the record on paper says your boundary is defined by a peg in the ground, so you're back to square one.

https://japanpropertycentral.com/real-estate-faq/land-bounda...

> An inadequate number of drones is another problem, as is the focus on picking the lowest bidders rather than the best technology.

This sounds exactly like the kind of project which shouldn't be lowest bidder first, given the cost involved when mistakes are made.

Any thoughts on getting this right using satellite data alone instead of drones? What are the challenges?

Using satellites come with their own problems.

First, this requires high resolution imagery to make out fine details. You have providers who can provide imagery with 0.3-2m resolution, but these can be expensive and not so easy to obtain. Satellite imagery for true colour is also affected by atmospheric conditions such as presence of clouds, which can occlude the view. Infrared bands can peer through this cover, but then they are not much use in surveying. Besides, you will need to wait for the satellite to make a pass over the target area to obtain the image, so the imagery can be out of date or captured at the wrong time e.g. if land is submerged under floods, so details are not visible. There is of course the upside that satellite imagery can cover huge swathes of land at once.

Drones are cheaper and can be flown over the target area on demand to generate true orthophotos from overlapping photos (https://en.wikipedia.org/wiki/Orthophoto).

Are drones really cheaper for large areas?

The problems you mentioned about needing to wait for satellite passes etc, would exist and may be worse with drones.

Drones can be economical for smaller areas, such as villages as mentioned in the article. For larger areas spanning hundreds of square km/miles, small UAVs won't do. Long range fixed wing UAVs or regular planes equipped with hi-res cameras or even satellite imagery are more suited. In the end it's a question of which approach is more fit-for-purpose, given the requirements and constraints.

In a country like India, cloud coverage during the rainy season can last for months, rendering satellites (unless you use something like SAR with false color composites) effectively unusable. Drones may not be able to fly during periods of rainfall, but can still be operated in daylight during breaks in weather.

While I was discussing the same with my friends in the industry, came to know about the following facts: 1. Even though resolution is high, drift of min 5m is there for the best of the satellites i.e. relationship between image and gps location. Also given the drift is random, harder to correcct. 2. Indian Surveys require Ground sample distance of around 3mm, drones can get this flying at 50 m height, getting the same from satellites would be too hard and expensive.

Reply on news.ycombinator.com