BY CALEB HAMPTON

AgAlert

For decades, experts have done their best to estimate the size of California’s almond crop prior to harvest.

With the state producing more than three-quarters of the world’s supply, the global market rises and falls with the fate of the Central Valley crop. Expectation of a shortage can stoke competition among buyers and raise prices, while a large crop forecast can achieve the opposite.

“It affects many lives,” Jasbir Sidhu, who grows almonds in Fresno and Madera counties, said of the yearly estimate.

Dan Sumner, professor of agricultural and resource economics at the University of California, Davis, said reliable forecasts are important because they help farmers make sensible investments.

“You make decisions based on what you think is going to happen,” Sumner said.

But experts said the rapid expansion of almond production in California—acreage roughly tripled during the past 30 years—made it harder to predict the crop.

Since the 1950s, the U.S. Department of Agriculture National Agricultural Statistics Service provided almond crop estimates. In recent years, USDA released two annual forecasts: a “subjective” estimate each May based on a survey of growers and an “objective measurement” each July based on counting nuts in hundreds of sample orchards.

Last year, the department stunned growers with a July forecast of 3 billion pounds.

“It signaled to buyers around the world that there was an oversupply, which was not accurate,” Sidhu said.

Almond prices dropped nearly 20% overnight, according to various sources, temporarily erasing roughly a billion dollars from the total crop value and eliminating the profit margin for many farmers.

After a few weeks, prices recovered as a consensus emerged that USDA had overestimated the crop, but the forecast still cost many growers. And it prompted a change.

In December, the Almond Board of California’s directors voted to stop funding USDA’s objective measurement, making this year the first in more than half a century that the industry will not look to the federal government to measure the crop ahead of harvest.

Bob Silveira, chair of the trade group’s board of directors, said almond production had simply gotten too big, with too much variation from one orchard or microclimate to the next, for USDA’s sampling method to produce a reliable statewide forecast.

“Getting a truly representative sample was just impractical,” Silveira said in a statement.

But is there a better way to count nuts?

Joel Kimmelshue, owner and principal agricultural scientist at Land IQ, a Sacramento-based technology firm that specializes in geospatial mapping, said the company was “in conversations” with the almond board about providing the industry with a crop forecasting calculator.

Embedded in the calculator are layers of data on the state’s almond footprint, including up-to-date bearing acreage and the age and location of every almond orchard in California.

Those variables are important, Kimmelshue said, as some regions generally have higher yields, while young orchards yet to reach peak production can double their average output from one year to the next. Some orchards are also classified as stressed or abandoned, affecting yields.

To map these variables, Land IQ used satellite data, remote sensing, artificial intelligence and a geographic information system, or GIS, to develop its own remote sensing technology. With near-total accuracy, the technology recognizes the visual patterns of different crops and conditions.

To calibrate, verify and continually refine its technology, Land IQ uses “heavy ground truthing,” Kimmelshue said, employing an agronomist to drive 20,000 miles a year—nearly the equivalent of circumnavigating the world—surveying California farmland. The agronomist is accompanied by staff who capture images of a subset of all crops.

For several years, Land IQ has provided official acreage figures for the almond industry.

“What we’ve chosen to do is give the industry the ability to develop its own yield estimate with really solid data behind it,” Kimmelshue said of the yield calculator, adding that stakeholders could use the tool to “run different scenarios” based on reports from growers.

If there were a hard frost in Butte County and perfect conditions in Kern County, he said, “you can turn this knob and turn that knob” to calculate a statewide figure that accounts for regional disparities.

Kimmelshue said it was “still up in the air” whether the almond board would use the Land IQ calculator to release a crop estimate this year. The almond board declined an interview request from Ag Alert.

Without USDA’s objective measurement, experts said the market would likely respond this year to the department’s grower survey, which last week estimated a crop of 2.7 billion pounds, as well as forecasts published by various private entities.

While counting nuts may not be rocket science, some agricultural technology firms have in recent years leveraged experience in aerospace engineering to bring new approaches to crop forecasting

Bountiful, a San Francisco-based agricultural data science subscription service, was founded by three former aerospace experts. The company uses machine learning combined with historical data, satellite images and weather analysis to provide crop forecasts for individual farms and a statewide estimate.

“It is rather different than counting nuts on a tree,” Bountiful founder and CEO Megan Nunes said in a presentation at the annual Almond Conference in Sacramento in December.

Sidhu, the almond grower, said he agreed with the almond board’s decision to eliminate the USDA forecast. But he said he worried other forecasters may also struggle to account for increasingly volatile growing conditions as farmers contend with unpredictable weather, water shortages and financial pressures that may reduce fertilizer or pesticide applications in some orchards, affecting crop yields.

“I think our challenges are moving faster than what our technology can cope with,” he said.