GaugeMap: Thousands of river gauges, each with a Twitter account | Hacker News Reader

GaugeMap: Thousands of river gauges, each with a Twitter account

gaugemap.co.uk

94 pointsColinWright32 comments

This is amazing! I've wanted to do some "citizen science" project in waterways around me. For instance, building cheap sensors that detect temperature/water level/salinity/particulates/etc that are so cheap as to be basically disposable.

The sensor itself could probably be made for less than $10 these days using various arduino modules, but I'm stuck on how to have the sensors communicate back to a "base station" to collect the information. Cellular-based would probably have great range but who wants to have a phone plan for each sensor! Zigbee may be an alternative, but it may require line of sight. Obviously it'd also have to be low power to work on a LiIon battery, or possibly have a solar panel to harvest power.

I believe the only way to make change is to show our impacts on our environment. And local government doesn't have the tech expertise, money or political willpower to suggest that "local business" may be destroying our health (when it's creating jobs!).

Anyone have any tech suggestions or organizations that do this sort of thing?

Search for "M2M" (machine to machine) communications. There's a lot of this going on over old 2-way paging networks. It's how the non-vaporware "Internet of things" works. There are thousands of air conditioning compressors, refrigeration units, elevators, pumping stations, and other boring but important pieces of infrastructure sending their short messages every few minutes. Often they're plain text, something like "Pump 1: OFF. Pump 2: ON. Inside temp: 71. Outside temp: 52. Faults: NONE." There are little M2M modules for such communication. They tend to be rugged, reliable, and cost a few hundred dollars.

Some of these use cellular service. There are special low-data-rate M2M plans for this. AT&T charges $3.33 a month for 1MB/month if you pay by the year. If you have a protocol that doesn't bloat its data traffic by using HTTP, HTTPS, XML, JSON, and other data hogs to send a few bytes of payload, this works out well.

There's something new called LoRa.[1] This is a standard for low-bandwidth wireless network devices with a claimed 10 mile range. Data rates range from 300 bps to 50 kbps. They run in the 900MHz ISM band, so they have good building and foliage penetration, although the range then drops to a claimed 2-3 miles. Units start around $56 on Digi-Key. There are Internet gateways available, so that many of the little cards can be connected to the wider world.

[1] https://www.lora-alliance.org/

Satellite is also an option for remote locations:

https://www.sparkfun.com/products/13745

$250 for this module, and $12/month for the satellite usage.

For the sensor itself, that's a simple micro controller for under $10, and a $5 sensor, a battery pack, solar panel, and a most likely, a few resistors. Costs add up quickly if you don't want to run out and replace the batteries, and if you want it to phone home.

It's pretty amazing that AT&T charges that much per MB; that's close to market for satellite M2M. Of course, the terminals for cellular M2M are considerably less expensive.

There are cellular network providers that specialise in IoT applications, where you pay monthly per device and data transferred. This is the first one from a search for "IoT Cellular": http://www.aeris.com/technology/aerconnect/

Particle run their own, but by the looks of it you have to use their Electron hardware: https://www.particle.io/cellular

If you could get people to install an app on their phone, you could use Bluetooth LE to passively collect the data as they walk past. There's a few asset tracking solutions that use this approach, such as Tile: http://www.wired.com/2013/08/tile-a-better-way-to-find-your-...

I'll throw this out there as well. . .

Windows 10 IoT OS is available too, which is built for something like this.

https://www.microsoft.com/en-us/WindowsForBusiness/windows-i...

I'm looking at something similar at the moment, but in agricultural areas where line-of-sight is inhibited only by foliage and terrain. For my purpose, it's OK to collect an SD card every so often, but I've toyed with the idea of an ESP8266 and a directional antenna to e.g. a mobile broadband router, probably again with an antenna upgrade.

Earlier today, I found [1] which - amongst other things which you might be interested in - calls out a venture [2] to establish a network of 'citizen scientist' type flood sensors across the UK. Their scheme looks to be a network of radio-to-Internet gateways which support a number of independently-deployed sensor nodes.

[1] http://oomlout.co.uk/blogs/news/81763329-an-overview-of-ardu... [2] http://flood.network/

I am a hydrographer in the NT working out in very remote areas- Phone reception is rarely possible for such monitoring sites. Currently we use satellite Comms. to relay water data home, but this is very expensive and prone to failure.

I'm wondering if you, or anyone else, has some ideas on long range communication strategies for large areas (NT is 1.4 million km2).

The area we are required to cover generally means sites are chosen based on accessibility rather than operational importance — it would be great if we could expand the network with low maintenance loggers capable of communicating long distances.

Which operator are you using? We've found L-band to be very reliable and does not suffer from any of the attenuation issues found in Ku. And heavy spreading on the return path provides for an even more resilient link.

Disclosure: We'll be offering an L-band satellite gateway for IoT this summer.

We currently use Iridium. On Ku band I beleive. I'll look further into this and see what our options are for L-band.

Outernet looks like a great project. How small of a satellite can be used to achieve the connection? Any more info/documentation on the L-band gateway?

What update frequency is required?

It's heavily dependent on river/climate conditions. During the dry most loggers will only need to be polled once every hour. However, during the buildup and wet, the stations would ideally be sending back every ~5 minutes.

Thank you, you people are awesome!

I just needed to get the correct keywords to google and came across:

http://postscapes.com/cellular-internet-of-things-developmen...

The dataplan on some look very reasonable, like $1/MB/Month, but the hardware itself seems cost prohibitive.

I mean, the Dash costs $59 or the Electron at $39 just for the board to send over 2G? You can get a disposable cellphone for < $10 complete with LCD screen and that worm eating game!

I guess I'll just need to wait another year or two until economies of scale kicks in...

Oxford Flood Network http://oxfloodnet.co.uk/ "Making a flood detection network in Oxford, using low cost sensors and volunteers"

Sort of related, this website crowdsources finding and fixing polluted waterways: https://creeklife.com/

I've been supporting hydrological data collection for over two decades. When I first started in the early 90's, there was no clear defined trail on 'how to do it'. I learned over time what works and the right way to do it. I'd like to make a few comments.

You asked about organizations that engage in 'citizen science', I would start with the NOAA Co-ops program https://www.ncdc.noaa.gov/data-access/land-based-station-dat... . This would give you the basics. Another route taken is through a college based program. http://iowafloodcenter.org/projects/stream-stage-sensor/rive... As you can see, Iowa has a program that is on the lines of what you are looking at.

You mention about cheap 'disposable' sensors. Truth be told, no reliable sensor is truly cheap. In sensing, we look at repeatability, accuracy and resolution. In the design of a sensor we also look at interfacing and field calibrations.

Field data collection systems are comprised of several parts. Power, Sensing, Data Logging and Telemetry. It's interesting, different vendors of this equipment take different approaches. Everything revolves around the abilities of the data logger. Let's look at several:

Hobo: http://www.onsetcomp.com/products This is a pretty good company. The sensors have good accuracy. The products are reasonably priced. However, you are locked into their products and limitations. I've seen Hobo used by students in course work.

What does the USGS, USACE, NOAA and BUREC use for data logging? They use capable data loggers that can accept SDI-12 and ModBus based sensors along with Analog based sensing. CS: https://www.campbellsci.com/dataloggers and Sutron: http://www.sutron.com/product/xlite-9210b/

As brought out in this thread, Telemetry is the weak link as the trade offs are range verses bandwidth verses data charges, etc.

The notion of Real-Time data is also a factor. Is Real-Time 'right now', every 15 minutes or hourly? All play a factor in telemetry.

@RiverKey2640

Just out of curiosity, what's the probability that Twitter would shut down these river gauge accounts because they're not people? Does Twitter have a history of leaving projects like this alone?

There are beyond thousands of twitter account bots that say they are bots—never heard of Twitter shutting one down.

Twitter actually encourages / supports these types of accounts, as detailed here, where they describe a bot that responds to airfare requests and another one that shares information about school closures: https://support.twitter.com/articles/76915

This is a brilliant presentation of data. It's straight forward to make out which parts of the country are at current risk of flooding. It's a shame there are aren't more sensors of river flow but what there is, is useful.

FWIW, the US Geological Survey has an API (plus their own interactives and reports) for river flow http://waterdata.usgs.gov/nwis/rt

Yup -- they even have configurable SMS & email alerts you can set up to let you know when the water level at a particular point goes below or above a specified depth: http://water.usgs.gov/waternow/

I use it to let me know when my local river is running deep enough to make for a fun day out on a kayak.

http://bearriverbasin.org/rivers/rivers/

I made a similar map (though much less pretty) for the US a few years ago: http://whitewatermap.com

Also, the links to go to the American Whitewater pages are broken: It gives americanwhitewater.org//content/... instead of americanwhitewater.org/content/... But thanks again, this is really helpful.

I put something together for the US a couple of years back as well. I've been meaning to add some logic around deviation from the mean, etc. but haven't gotten around to it.

http://wwos-bitmason.rhcloud.com/

Some discussion of the app and parsing the USGS data for those interested:

https://blog.openshift.com/use-openshift-to-map-river-levels...

http://bitmason.blogspot.com/2013/11/over-summer-i-wrote-pos...

This is great! It would also be cool if you could have polylines representing the rivers for each given segment (the putins and takeouts are now on AW, not sure how long they've been up). But nonetheless this is a great resource, thanks!

Compare to the usgs site or this US army corps site: http://riverstatus.usace.army.mil

Excellent site and very timely - the UK is currently undergoing a series of serious flooding incidents, especially in the north of the country.

Amazing!

If we had this in our rivers prior to the recent Chennai floods, several lives and property could have been saved.

Is it possible to limit zooming out and panning with Google maps? It strikes me that it would make for a better UX if you could limit that behavior to avoid shifting the map out of the relevant area for some of the less savvy users.

Is there any information on the sensors themselves out of interest.

I'm curious if they use something like LoRa with CSS. Or FSK. Or some kind of cellular system.

Reply on news.ycombinator.com