By Munyaradzi Makoni
CAPE TOWN, South Africa, Sept 26 (Thomson Reuters Foundation) - It was in Mbeere, in rural Kenya, where Muthoni Masinde began to pick up on nature's way of signposting shifting weather patterns, as she helped her mother farm their land.
She learned that if thick swarms of crickets appeared in the evenings during planting season, it meant the rains were about to end - a sign that farmers should stop planting because their seeds would not germinate.
If farmers could hear the bird known locally as kivuta mbura - "the one that pulls the rains" - they should expect heavy rain for several hours afterwards.
Studies in computer science soon took Masinde away from rural life, but she never forgot the value farmers placed on observing the environment and its rhythms to grow their crops.
Using her expertise in computer science, Masinde has now developed a tool to make predicting drought easier for Africa's small-scale farmers by combining traditional know-how with scientific weather data.
"I was motivated by the realisation that in most African countries, rain-fed agriculture still accounted for over 70 percent of food production," Masinde told the Thomson Reuters Foundation.
The mobile application, which has been tested in Mozambique, Kenya and South Africa's KwaZulu-Natal province, works by pooling information on rainfall, temperature, humidity, atmospheric pressure and other climatic indicators.
Observations are collected from conventional weather stations and smaller, cheaper sensor-based stations that can be deployed in large numbers to give better coverage and more accurate results.
Farmers feed their observations into a mobile phone.
"With all this data, computer science models are used to predict the drought for short-term, medium-term and long-term," said Masinde, now a senior lecturer and head of the information technology department at the Central University of Technology in South Africa's Free State province.
Farmers are consulted to merge the technical information with their own forecasts before it is widely distributed, she explained.
"Indigenous knowledge ensures that the system is relevant, acceptable and resilient," Masinde said.
The information is simplified and disseminated to small-scale farmers via SMS and audio files in easy-to-understand messages, such as: "There will be adequate rain during the first two weeks of the season; you are advised to plant early to take advantage of this rainfall".
PREDICTIONS LIKE TOSSING A COIN
The World Bank says in sub-Saharan Africa, agriculture accounts on average for 35 percent of gross domestic product and employs 70 percent of the population. More than 95 percent of the region's agricultural areas depend on rainfall rather than irrigation.
Many African farmers still use knowledge passed down through generations to guide them on when, how and what to plant.
But such traditional know-how is becoming increasingly unreliable in the face of climate change, which has disrupted the seasons and led to longer droughts, greater flooding and erratic rainfall.
"Predictions have become more like tossing a coin," Masinde said.
Conventional weather forecasts are often of little use to small-scale farmers because they tend to cover huge areas and use unfamiliar technical terms.
"Near normal rainfall means nothing to rural farmers," Masinde said. "Some don't even have access to media used for forecasts, like televisions."
Masinde was last month named a winner in South Africa's annual Women in Science Awards, taking the prize in the category of Distinguished Young Women Researchers: Research and Innovation.
The award came with 75,000 rands ($5,255) in prize money which she said would go towards establishing a drought mitigation centre in South Africa's Free State province.
Last month, the South African Weather Service partnered with Masinde's institution to begin the process of rolling out the tool. Namibia has shown interest as well, she said.
She is working with a team of researchers and postgraduate students to put the tool to use in the Free State and some parts of KwaZulu-Natal province.
"In the future, anyone with access to the internet can access the forecasts generated from the tool," she said.
($1 = 14.2710 rand) (Reporting by Munyaradzi Makoni; editing by Katie Nguyen and Laurie Goering. Please credit the Thomson Reuters Foundation, the charitable arm of Thomson Reuters, that covers humanitarian news, women's rights, trafficking, property rights and climate change. Visit http://news.trust.org to see more stories)
Our Standards: The Thomson Reuters Trust Principles.