By Dr. Paul Monaghan, Fish Welfare Initiative Research & Development Lead
Summary: In order to increase the scalability, impact, and cost-effectiveness of our farm program (the ARA), we are interested in leveraging satellites to detect water quality issues remotely, which would lessen our current reliance on visiting farms individually. This post provides an update on the work we conducted in this area in 2024, and our plans going forward.
Specific updates include:
We discovered some issues while attempting to replicate the findings of our previous proof-of-concept satellite study, conducted in the first half of 2024. This has led us to be more skeptical of that specific model and to pursue other models instead. Publishing our initial findings before critically analyzing them further was a mistake on our part.
We remain excited about this technology and have now recently published an innovation challenge, in which we will pay companies for developing satellite-powered predictive models that meet our specifications.
Background
FWI’s current core program is the Alliance for Responsible Aquaculture (ARA). This program centers on FWI field teams collecting water quality data from member farms and providing the farmers with corrective actions when critical water quality parameters indicate that fish may be exposed to poor conditions.
The current ARA model requires FWI personnel to visit fish farms physically. This dependency on field teams, and the limitation on the number of farms that FWI personnel can visit each day, means the current model does not meet our minimum program thresholds for scalability. This limit to scalability, coupled with the low frequency of visits per farm, limits how impactful the program can be. Given these challenges, we aim to make improvements to the ARA such that it is more scalable, impactful, and cost-effective.
In 2024, we began exploring if satellite-based remote sensing is a viable option to allow us to remotely monitor water quality. If this approach works, it would open the door to more frequent collection of water quality data at ARA farms, while simultaneously reducing the requirement to actually visit farms. This could allow us to significantly scale the program by facilitating data collection at many more farms.
What We Did
At the outset of this work, FWI had no prior experience in satellite-based remote sensing. To test the feasibility of this approach, we partnered with a private partner in India to leverage their experience with accessing and analyzing satellite imagery. Together, we conducted a proof-of-concept study—conducted between February and March 2024—which compared water quality data determined by analysis of satellite images (provided by our partner) with empirical water quality data determined by direct measurements at the same 20 ARA member farms (provided by FWI).
In June 2024, we published a blog post summarizing the key findings from this proof-of-concept study. At the time we were excited, as the results indicated a strong correlation between the satellite-predicted and the empirical data for four out of six key water quality parameters we studied, suggesting that this technology had potential for us to integrate into the ARA.
However, concerns about the findings came to light when we subsequently engaged three separate independent consultants to probe into the data (we are grateful to Animal Ask, amongst others, for helping us here). Efforts by the consultants to develop predictive models of their own—using the same empirical dataset we collected as part of the proof-of-concept—were unsuccessful. Additionally, when we attempted to replicate the findings on our own by using the same previous model that had given us the initial results in the study, we were unable to do so.
It remains unclear why our initial results couldn’t be reproduced by the independent consultants, or by our own internal team. This experience has provided us with valuable lessons about the importance of additional validation before drawing conclusions.
Path Forward
Despite these challenges we remain optimistic, and we believe that the potential of satellite-based remote sensing is sufficiently exciting for our programming that it warrants further investigation. To boost our chances of success, we recently launched an Innovation Challenge. This Innovation Challenge offers a financial reward to any party—technology companies, academic institutions, non-profit organizations, individuals/groups. . . anyone!— who can successfully show that they have models matching our needs. We hope that offering financial rewards incentivizes people to work on this problem.
Learning lessons from our earlier proof-of-concept study, we will independently conduct the validation exercise to determine if the models predict water quality parameters with sufficient accuracy for our needs. We expect to be in a position towards the start of the second quarter of 2025 to know whether we can continue pursuing satellite imagery as an intervention.
While we’re disappointed that our 2024 work didn’t produce models that we could take forward, and also displeased that we communicated results with more confidence than we now believe was warranted, we believe we have learned a lot. We continue to be excited by the prospect of leveraging this technology, now through our Innovation Challenge, to improve our programming and help us to improve the welfare of millions of fishes.
Comments