Computer model gives early warning of crop failure

Crop_computer_model_ Ecoagreiculture partners
Copyright: Flickr/Ecoagreiculture partners

Speed read

  • Climate-induced crop failures were predicted by the model in up to a third of the crop area
  • Rice and wheat yields were more predictable than those for maize and soybean
  • Predictions are helpful, but they must be put to good use

Send to a friend

The details you provide on this page will not be used to send unsolicited email, and will not be sold to a 3rd party. See privacy policy.

An international team of researchers has developed a computer model to predict global crop failures several months before harvest.

Since 2008, widespread drought in crop-exporting regions has resulted in large increases in food prices on global commodity markets. With climatic extremes also expected to become more common, being able to predict global crop failures could help developing nations that are reliant on food imports — making them more resilient to spikes in food prices.

The study, published in Nature Climate Change this week (21 July), involved analysing 23 years of climate forecasts and satellite observations to develop a computer model for predicting crop yields. The researchers then tested how well their model predicted the actual yields at the end of each season for four staple crops: wheat, rice, maize and soybean.

They found that climate-induced crop failures were reliably predicted in up to a third of the global crop area. The results suggest that computer models such as this could be used to produce crop estimates up to five months before harvest and help establish a system to predict global crop failure.

"This presents the first assessment of the reliability of cropping prediction on a global scale," study co-author Toshichika Iizumi, a researcher at Japan's National Institute for Agro-Environmental Sciences, tells SciDev.Net. "It demonstrates that we can predict food production ahead of the harvest, which is a valuable food security tool for dealing with changing climates."

Yet the reliability of the model's predictions varied substantially by crop, with wheat and rice yields being the most predictable. For the major wheat-exporting countries, the model's forecasts were reliable for up to 35 per cent of the harvested area.

“Even though these cropping predictions may not be perfect, they still have the potential to improve the efficiency of activities for ensuring food security.”

Toshichika Iizumi

However, soybean and maize yields showed little predictability. Maize is a key crop across much of Africa and Latin America, suggesting more work is required to improve crop predictions for many developing nations.

But Chris Funk, a research geographer at the University of California Santa Barbara, United States, says these findings could still help the developing world mitigate at least some food price shocks.

"In most of the world, wheat and rice are the dominant food source for rapidly expanding populations of urban poor," he tells SciDev.Net. "These populations, who may spend up to 70 per cent of their income on food staples, are highly vulnerable to rapid price increases."

The model also showed varying predictive powers between regions and countries. For example, reliable crop predictions could only be made for three per cent of the harvested area of Thailand, the world's second-largest rice exporter.

"Even though these cropping predictions may not be perfect, they still have the potential to improve the efficiency of activities for ensuring food security in food-dependent countries," says Iizumi. "This includes diversifying the countries from which food is bought, balancing combinations of imports and stockpiles of food, and prioritising emergency food imports."

Lindsay Stringer, director of the Sustainability Research Institute at the University of Leeds, United Kingdom, says that, while the information provided by predictions is helpful, the challenge for many African countries is putting it to good use.

"Focusing on food production is only part of the puzzle," she tells SciDev.Net. "There are also issues in terms of capacity gaps and institutional communication challenges. These cause problems in mobilising personnel and resources to act upon forecast information, and having necessary structures and processes in place to facilitate urgently needed action."

Link to study abstract


Nature Climate Change doi: 10.1038/nclimate1945 (2013)