South African Scientists Think Software and Tribal Knowledge Can Predict Drought

Could it work for Canada too?

Earlier in January, the UN announced that 14 million South Africans may go hungry thanks to reduced crop production—the result of a severe drought that's drying the region out.

The next time a drought hits, a team of researchers at the Central University of Technology in Free State, South Africa want local governments and peoples to be prepared. To that end, they're working on software that mixes modern and ancient knowledge to predict the onset of droughts.

"The research is born out of necessity," Adeyinka Akanbi, one of the researchers leading the work along with Dr. Muthoni Masinde, told me over Skype. "Right here in sub saharan Africa, we've been dealing with drought for the last couple of years and it's getting worse. We're looking for much more innovative ways to forecast and predict complex environmental phenomena."

The first step in the researchers' system is to collect the traditional wisdom of local indigenous peoples about environmental events and what they mean—the blooming of a particular tree, for example. These correlations make up what's called an "ontology." By feeding this information into what's known as a complex event processing engine, the idea is to give weather sensor data some additional context.

The Libelium sensors used by the team. Image: Adeyinka Akanbi

"For example, with the blooming of the tree, when it is captured in the ontology, the inference engine will be able to predict that the rain is going to fall soon," said Akanbi.

The trick is to build software that combines this information with a continuous stream of sensor data on soil moisture levels, temperature, and more. "The program puts them together and refines and improves the accuracy of the prediction," Akanbi added. For the current drought, Al Jazeera reports that no end is in sight, and the hot and dry conditions are expected to last for at least a few more months.

So far, the team in South Africa has set up a wireless sensor network made up of Libelium sensor boards in Thaba 'Nchu, a fertile region in Free State that is home to Tswana and Sotho peoples. The university has also already sent researchers out to 13 different villages in the region to make contact with three different tribes, in order to learn from them. The way Akanbi tells it, this has been the most challenging part of the project so far.

"There's a lot of literature on scientists recognizing that indigenous people have a lot to tell scientists, especially in ecology"

"When you try to go to these remote places, if you're not part of the tribe, they are not so welcoming," Akanbi said. "We had to get someone to act as an intermediary for the researcher, and we were able to gather this knowledge through workshops, questionnaires, and one on one interviews."

The process took months, Akanbi said, and the researchers offered monetary incentives to the tribespeople in return, and cell phones so they could stay in contact.

Akanbi also emphasized that the project is global in scope—the team hopes that when all is said and done, their software could be just as useful for governments (and individuals) in Canada, for example, as it is in South Africa.

Canada is an ideal candidate for this approach to weather prediction because First Nations peoples in Canada have a long history of watching for signs in nature to predict the weather, from the migration of birds to the position of the moon.

"There's a lot of literature on scientists recognizing that indigenous people have a lot to tell scientists, especially in ecology," said Glen Aikenhead, a professor at the University of Saskatchewan's college of education specializing in cross-cultural science education.

A workshop with locals. Photo: Adeyinka Akanbi

Compensating indigenous peoples in return for their knowledge, which Akanbi's team attempted to do, would be one problem for any governments attempting to survey them, Aikenhead said. The bigger issue, however, would be ensuring that they also have a stake in how their knowledge is used to make policy decisions in a country that has a long, dark history of oppressing and exploiting its indigenous peoples.

For example, the Canadian Environmental Assessment Act gives authorities the discretion to consider indigenous knowledge in any environment assessment, but that input doesn't necessarily translate into action.

"It turns out that these elders would share this knowledge in good faith, and the scientists would listen, but then when a decision was made, the question is who has the power to make that decision?" Aikenhead said. "It was always the scientists back in Ottawa, or back in the corporate boardroom, that said okay, we've learned this from the elders, but this is what we're going to do."

Even so, Akanbi said, the knowledge is there, and so is the technology to make use of it—but more importantly, South Africa needs to get ready for the next drought, once it makes it through the current one.

"These sensor networks, thanks to the Internet of Things, are readily available," Akanbi said. "You can easily get rainfall and soil moisture, and temperature, even on the internet nowadays."

"The only thing that's missing is the indigenous knowledge."