Farming by satellite

Delta, 2 June 2010

Remote sensing can be used to determine how much water is needed to grow a kilo of wheat, rice, maize or cotton. Dr. Sander Zwart developed a satellite application to make agriculture deal more efficiently with water.

The Food and Agriculture Organisation (FAO) has unofficially endorsed dr. Sander Zwarts’ model, called Watpro, by ordering that the same worldwide benchmarking Zwart has made for wheat should also be used for calculating rice and maize. As Zwart is currently in Benin, a colleague of his will handle this new assignment.
After eight years work on his PhD research, which he conducted alongside his regular job at WaterWatch remote sensing services in Wageningen, dr. Zwart (who has double MSc titles in both irrigation and remote sensing) is on his way to Africa to put his theoretical work on satellite-augmented agriculture into practice at the Africa Rice Center in Cotonou. “Chinese and Libyans are constructing large irrigated rice fields in Mali”, Zwart says. “They use water from the Niger to irrigate thousands of hectares of new rice fields.” Strangely enough, the Africans import their rice from Thailand, and no one is really sure to whom the Chinese will sell their African rice. The situation illustrates a globalised food market under stress. “In Asia, every small plot of land is in use and every drop of water is needed”, Zwart explains. “Africa has plenty of unused land and untapped water reserves. That’s what attracts China to Africa.”
The amount of water needed to produce a kilogram of crop is not only high but also quite variable. The water productivity – as the amount of crop per cubic meters of water per kilogram is called – varies from 1.52 kilograms of wheat per cubic meter in the Nile delta, to 1.39 kg/m3 in the Netherlands and only 0.54 kg/m3 in Pakistan.
The production per unit of water must be increased, Zwart argues, because the world population is rapidly growing. Besides, agriculture faces increasing competition for water from industry and domestic usage. On top of that, changing diets (increased meat consumption) accelerate the demand for food even more. To face that challenge, Zwart aims to establish benchmarks for the water productivity of important food crops and cotton. Such figures would enable local optimisation of agricultural water consumption.
By applying cunning algorithms, the water production value can be derived from satellite data, Zwart explains. Professor of Water Resources, Wim Bastiaanssen (Civil Engineering and Geosciences), developed the application Sebal (Surface Energy Balance Algorithm for Land) 25 years ago. He was Zwarts’ PhD supervisor.
The near infrared sensors measure the greenness of the crop. A sequence of measurements over the course of the season provides information about the biomass produced and thus of the harvest. The far infrared is a measure for the surface temperature from which the model derives the evaporated amount of water. But actually it gets much more complicated, since land use and weather data are also involved in the calculations. However, ultimately, the Sebal model presents for each pixel in the satellite image values for the biomass production and water use, from which the water production values can be calculated. This makes it easy to spot poorly performing plots where local improvement efforts should be focused.
As clever and powerful as Sebal may be for local applications, Zwart says it is just too complicated for making worldwide scans. He has therefore developed a simpler version, called WATPRO, which is based on Sebal and which immediately calculates the water productivity from remote sensing data for large production fields (minimum size: one by one kilometre) all over the world. This enables Zwart to establish worldwide benchmarks for water production. His map shows that wheat grown in France for example needs only half the water of that in Russia.
Comparisons with other methods of estimating water productivity show considerable differences among them (plus or minus fifty percent). Zwart argues that the accuracy of his Watpro method is comparable to that of Sebal (five percent over the course of a season). Other methods use different assumptions, leading to deviating values. Ultimately worldwide measurements are needed to calibrate the model, as indeed was done for Sebal.

S.J. Zwart, ‘Benchmarking water productivity in agriculture and the scope for improvement – remote sensing modelling from field to global scale’, PhD supervisor professor Wim Bastiaanssen, 26 May 2010.

Weblink to Delta