Scientists at USDA’s Agricultural Research Service have developed a new way of using satellite imagery to predict cattle weight gain on rangelands. The method fuses multiple images together over a period of time to monitor how forage quality changes over time and its relation to cattle’s weight gain.
The study was published in Ecological Applications. “This study is probably the first time high-quality datasets have been used to predict cattle weight gain directly from satellite imagery,” said Sean Kearney, a postdoctoral research associate in Fort Collins, CO.
Managing grazing on rangelands is often difficult due to variability in temperature and rainfall, and it’s important to know when forage production and quality are changing, USDA said.
Scientists used satellite images and field observations for 40 pastures grazed over 10 years to predict the performance of cattle grazing in eastern Colorado in the summer. Weight gain predictions were made by looking at forage quality and quantity in the satellite images.
“We observed that in years when satellite images showed forage greening up earlier, before cattle began to graze, the quality of the diet declined more rapidly and cattle weight gain was lower, especially toward the end of the grazing season,” Kearney said.
“In some years, plenty of biomass was still available late in the season, but a large portion of the high-quality forage was missed because it peaked (reached top quality) so early in the season. This resulted in cattle feeding on lower-quality grass, which reduced their performance.”
Lauren Porensky, research ecologist, said they knew forage quality mattered but not to what extent. “Now we can estimate diet quality across space and time and have a better idea of what is causing changes in diet quality throughout the season.”
Moving forward, USDA researchers will link the imagery with GPS collar data to study what drives cattle foraging behavior. They will also work on a new way to predict diet quality in near-real time. — Anna Miller,WLJ managing editor
