2020 21 season Penguin Data - SteveCossy/IOT GitHub Wiki

Page last updated 1 October 3pm

• Andrew Hornblow who is an electronics genius who has designed, developed, and manufactures a variety of sensors and other electronics.
• Elise Smith a researcher who discusses requirements with Andrew
• LJ (Lei Ju) is a recent post-grad graduate who has created, and hosts, a Node Red site that gets subscribes to Penguin data. http://leiju.org:1880/ Andrew has done most of the site development
• Sherman Perry writes javascript and other code for the project
• Nate is a current post-grad student who has a personal interest in learning how to use data analysis languages.  Easiest way to do this is to have some real data to practise on.

For a long time, Andrew has been providing technical assistance to researchers interested in supporting the penguin population around the coast of New Plymouth.  Andrews role in the research is to monitor characteristics of penguin burrows in real time.  This season saw the introduction of RFID tags in some penguins, with Andrew making RFID readers in various formats.  Photos in this album show a lot of details of the project: https://photos.app.goo.gl/CCsQzZSRJTfXYcYb9
Andrew is continually adding photos and screen shots of recent work. There are also some video IRD camera penguin activity links.

Typically, Andrew's sensor devices (shown in some of the photos above) use low energy, short range, wireless (RF) devices to send sensor data to a gateway.  The gateway is typically a Raspberry Pi with an appropriate RF receiver, which gathers the data and sends it to a cloud provider 'Cayenne'.  Cayenne is an MQTT broker.  Their servers receive data from authenticated gateways.  Other servers can 'subscribe' and receive copies of any data that is 'published' by a gateway.  Cayenne also provides excellent visualization of live data, but Cayenne does not keep all data points for more than a day or so.  Older data is accumulated - Seconds are accumulated into minutes, minutes into hours, hours into days etc.

This is working fine and is meeting the needs of the people following and tracking the birds as they usually just need to know what happened last night for purposes of nest box checks and RFID tagging and some specialist temporary GPS tracking that is about to take place (written 02/10/21).
The thermal data plots are a good example of glanceable work as wriggles = birds. We do not have an ambient thermal probes in each mound or cluster of nests. Andrew is working on a Node Red Function Node and JSON to discriminate a ambient thermal diurnal drift of up to 10 degrees and short term fluctuations over 10 minutes to make a penguin detector. This is still in progress. This will minimise opening nests and disturbing birds while they are midst egg laying and chick raising. RFID reading of the 'chipped' birds at each nest box entrance is being commissioned at present. This will be available as a great aid for the scientists trying to pin-point bird arrival and departure esp the ones with expensive GPS trackers. A parent bird may only return for a few hours in middle of the night.

I have a server at Whitireia which uses Python files to subscribe to various Cayenne data streams, and keeps a permanent copy of that data.  On 1 October I downloaded all current data files and stored the copies here: https://w2shared-my.sharepoint.com/:f:/g/personal/steve_cosgrove_whitireia_ac_nz/Eua_C7BdzVBJpCL3ZXViBYgBGrijhxLtqKgaxlA1ka3B4A?e=HYhlVB

The code running on that server is described here: https://github.com/SteveCossy/IOT/wiki/CayenneMQTT

Filenames are of the format: <device id>_<Channel number>  It appears that current Penguin data all has the device id of adeef670-4bcd-11eb-b767-3f1a8f1211ba_6

Each line of a data file is in this format: <timestamp>,<device id>,<data value>. This provides a valuable detailed database for researchers. Cayenne is more useful for the day to day shareable quick view dashboard.

Each channel collects data from one sensor.  Some examples can be found by observing the live Cayenne data views:

• https://cayenne.mydevices.com/shared/60bb14fd2a964b08bc38200e
• https://cayenne.mydevices.com/shared/60bb14442a964b08bc381fcb

Channel    Purpose

• 1   Ambient Temperature outside a penguin burrow.
• 2   Temperature inside nest (burrow?) 14
• 3   Temperature inside nest 09
• 4   Temperature inside nest 08
• 5    Battery status on the same sensor

https://cayenne.mydevices.com/shared/60b815952a964b08bc333d91/

• Channels 30 - 33 - RFID values observed in each burrow

Data for each device will arrive at the main device page in the first instance

• This is the main point where data can be Subscribed to etc a kind of engineering access point.
• Data at this point can only be seen when logged into the Cayenne Account.
• When setting up each sensor, a Standard descriptor is added in general <ch X> - Description format.
• The number of decimal places and possibly the data Type and Units may be added.
• Although the data at this point Can be turned into a pointer or a plot it is best to leave data in Minimal Raw format.
• One or more 'Devices' may be present.
• A 'device' may be a Pi or other Micro 'Hardware' or a virtual device created by, for example, a python script.

Project Pages can then be developed from the above device data into meaningful groups with pointers or plots and importantly Shared using simple links etc. These will chop and change as things grow and new devices like GPS and RFID data start to stream or things like use of ambient temperature reference fall out of favour.

The data currently being collected includes channel numbers 1 - 26 for environment and battery data, and 30-33 for RFID data.  Andrew advises that details of what is being recorded in each channel is changing all the time as new equipment is being installed but at present (1 Oct 2021) Broadly...
4 x Each Mound has each one picaxe that transmits

• 4 x DS18B20 temperature probes to 1 decimal place (may be upgrading to 2 shortly)
• 1 x Solar PV / Battery status

3 x RFID readers Ch 31,32,33 (at present - 34th when parts arrive )

• 1 x RFID Mk II reader per mound on a power cord that can be placed over the nest box entrance of any bird(s) of interest
• 2 x Portable battery powered proposed Mk III reader that has directional sensing + RFID to track particular birds, situations
• Each RFID has integral unique MOFrequency of the RFID loop tuning and power supply battery status on One common Ch30 Sys Status

The penguin data can be broadly identified by observing values:

• 0 - 30 will most likely be degrees celsius.
• 3000 - 5000 will most likely be battery voltages in millivolts (3 - 5 volts). Practically the voltage is between 5 Volts (5000mV) Max and 3 Volts (3000mV) Min LiPO uSolarPV equipment e.g. temperature sensors shut down to protect battery < 3300mV
• Higher values will most likely be RFID values for passing penguins

This data could be analysed to get information such as:

• Which burrows are in use by which penguins (among the few that have RFID chips).
• How often penguins come and go from each burrow (indicated by changing temperatures).
• Which burrows have eggs being incubating (temperature never drops to zero as there is always a bird there incubating the egg).
• When bird roll the eggs over or swap incubating duties (each causes a different, distinctive. fluctuation in temperatures - Andrew might be able to supply examples, or there might be some in the Google Album above).
• How often, at what time, and other characteristics can be determined about penguins visiting each others burrows.
• Discriminate a ambient thermal diurnal drift of up to 10 degrees and short term fluctuations over 10 minutes to make a penguin detector. (Andrew's idea)

Note from Andrew: Elise will be the best person to fill in the facts from some of my woolly wishful thinking. My own thesis is that a heck of a lot can be monitored with the most basic setup and this all helps the science. The birds are an 'indicator' species.
Taking a Meta-Penguin view: Looking at the Whole population within well defined situation at the port i.e. when what bird numbers and stages per year: The timing of bird pre visits, arrival, lay eggs, fledge, return moult. This put alongside oceanic temperatures, climate etc is a potential gold mine of Living data and coastal marine environment health with regard to climate change. Least disturbance of wild birds is best and can save carbon, kWh'age and cold nights by watching and sharing the data from comfortable glow of our own devices :)

• What is a typical battery charge / discharge cycle (if there is a 'typical')
• What is the correlation between temperature and battery charging rate (high temperature should indicate more sun, and more charge). There are no real serious temperature sensors on the electronics / battery enclosure, so this is not an exact science! LiPO sci and Relationship probably well defined elsewhere ?
• Which batteries show signs of not being able to hold a full charge.

Note from Andrew: As per shared overview project https://cayenne.mydevices.com/shared/60bb14442a964b08bc381fcb

• There is a 'recycled' 250mAh battery from sound lure project that seems to have a happy max terminal voltage of 3950 mV.
• The performance of the tiny (solar garden style) uSolar PV is mounted inside the clear top case is outstanding.
• Most days sun or shine Mounds 1 and 4 do not drop below 4000mV.
• Normal operation saw no battery outage mid winter + weeks for grey / wet sky.
• Solar PV orientation is simple 'general North' and almost vertical so only getting a few hours top up per day.
• This points to potential use of infinite durability, low cost / ecco / super capacitors (on hand) as an experiment... None of this too surprising given the PICAXE-08M2 chip is idling at single uAmp digits 95% of the time.

Presently you will spot the faulty temperature lead that has been most likely chewed by a bird OR rat: Temperature probes have live +ve and -ve in them and any water getting in will discharge things!

Last few months we have discovered use of simple dehydrated / oven kiln dried MDF gear plate placed inside base of the hermetically sealed clear top boxes can absorb as much as 3gm (cc/ ml) of water. Each plate is cut weighed 'as is = shed damp' then kiln dried and re measured as installed etc. This is significant for electronics in the wild!
So in short RFID is a work in progress (Power issues) and other competing projects: Similar idea dumping to EEPROM 4 fathoms under the ocean 13km offshore South Taranaki...