discussion / Human-Wildlife Conflict  / 13 December 2023

Early Warning Systems (HWC) - successes/failures, recommendations? 

Hi everyone, 

I'm getting a lot of questions come my way about what early warning systems are available and effective -in any sort of environment. I'm keen to crowdsource some experience here - are you using Early Warning Systems? Can you tell me about how they're working - species you're working on, successes/failures, recommendations?  

Thanks a lot! 

Steph 




We had reasonable success predicting elephant crop raiding from movement patters during the day preceding the nights crop raid Staging behaviours identify spatial and temporal risk of human‐wildlife conflict - Hahn - 2023 - Journal of Applied Ecology - Wiley Online Library (colostate.edu)

Hi @StephODonnell , that's one of the ideas behind our 'WID' tag:

We are currently working on an updated version of this sort of tag, with much better range and flexibility, and I am hoping to organise a release to open source, although it involves a partner organisation, so they'll also have to agree. I think there are some similar systems with bears and sharks that I can, ahem, track down if you like. Is this sort of what you are after or am I way off?

There's also the emerging AI camera systems of course (e.g. as discussed by @kimhendrikse here: 

Apologies if I am way off...

Cheers,

Rob

 

Hi @StephODonnell , HWC applications are coming up on our radar pretty soon. We currently use the Sentinel devices and AI models we build in-house for biodiversity monitoring but our current satellite provider's technical limitations mean we can't send alerts or filtered information from the camera traps in under 2 hours. We are coming out with a LoRa version of the devices in a month or two, and we're building the cellular version to be released by the end of the summer. Those should open the door to sending information fast enough to respond to. 

These kinds of applications are at first going to be more successful in areas that already have functional LoRa/cellular infrastructure, so it leaves out dense rainforest habitats (where LoRa is going to be a less viable comms solution). However, we're working with another satellite provider in the coming months to speed up the current latency on our satellite data delivery so that they can be used for HWC scenarios. 

On a question of scalability, we've done ~200 devices in 6 countries over the past 10 months and are working on building and deploying 1000+ by the end of the year. We also do custom AI models so species are not a limitation - so long as there is enough training data to work with for the species of interest!

Cheers,

Henrik