You are here

How The Climate Corporation Built Big Weather Data

Back in 2002, David Friedberg walked daily
past a shop called the Bike Hut on his way to work at Google in San Francisco.  

On sunny days, business brimmed at the
bike rental shop. On rainy days, though, sales slumped.

This prompted Friedberg to ponder
about how much money weather-dependent businesses lose.
During cool summers, utilities lose money
because fewer air conditioners run. Ditto for manufacturers who try to sell
toasty warm coats during balmy winters.

“Seventy percent of businesses lose
money due to weather,” says Friedberg.

Farms make up a good share of those businesses.
“Agriculture is the most weather sensitive
industry,” he says.


Why It’s Different For Farmers

Other industries buffer
weather risk by location. For example, odds are that adverse weather won’t all
strike a clothing company with sites in San Diego, Seattle, and St. Paul.

Not so in
farming. Adverse weather could easily wipe out a farm’s crops that are spaced
just several miles apart.

Farming is also risky,
as high input costs often accompany high yields. Capping this is increasingly
volatile weather. A good example is the 2012 drought that was followed by the torrential
downpours in many areas during spring 2013.

“It doesn’t matter
if climate change is manmade, or if it is part of a natural cycle,” says
Friedberg. “Whatever it is, (weather) is getting less predictable.”

Farmers hedge this risk by buying federal crop insurance. However,

Friedberg and
Google
co-worker Siraj Khaliq also saw several niches to tap that federal crop
insurance could not reach.

“You can buy (federal) enterprise unit
coverage and get net revenue coverage across your farm,” he says. “But many farmers
still have high-risk and high-value fields.”  These fields are not fully protected under
federal crop insurance, he says.


Field
Specific

In 
2006, Friedberg and Khaliq launched WeatherBill, a firm that sold weather
insurance to supplement federal crop insurance policies. To support this
business, the company developed a huge array of weather and agronomic data on
individual fields across the United States. Investments in data management fueled
by heavy venture capital infusions make it possible to pull in over 10 million
daily environmental observations in the continental United States, say company
officials. In time, a $100 million investment enabled the firm to pinpoint
individual field characteristics, such as current moisture and 24-hour
precipitation levels.

“Our largest (former) employer (of
employees) is Google, so we know about manipulating large amounts of data,” jests
Jimi Crawford, the firm’s senior vice president of science and engineering.

In 2011, the company changed its name
to The Climate Corporation (Friedberg says one negative connotation of WeatherBill
is farmers would just think of getting a bill.) The Climate Corporation
developed a website (www.climate.com) where
farmers can register fields and receive field-specific weather updates. Earlier
this month, it also launched Climate Basic and Climate Pro products to help
farmers better manage agronomic risk.

Although they didn’t give specifics,
Climate Corporation officials say thousands of growers across 10 million acres
have used the website and the firm’s insurance products.

In time, its weather data capacity
attracted attention from buyers. On November 1, Monsanto finalized its $930
million purchase of The Climate Corporation.

Monsanto officials say most corn farmers leave 30 to 50 bushels
per acre on the table. Bettering the use of data science will help them glean
these yield boosts, say Monsanto officials.

Officials for The Climate Corporation
say the acquisition enables it to team its extensive weather data with Monsanto’s
crop genetics.

“As we get growers to tell us what
their farming practices and genetics are, we can do a better job giving
agronomic recommendations,” says Friedberg. The agreement also allows The
Climate Corporation to partner its weather data with other companies.


Seed
Selection

Today, complex computer software can help
farmers match seed with factors like soil
type, water-holding capacity, or past yield history. Some use satellite
imagery.

All these finely honed plans can go
away, though, if adverse weather rolls through fields. Officials for The Climate
Corporation note the firm’s technology can’t predict the weather.

“You won’t be able to project with a
high degree of certainly that it will be dry in the growing season,” says Jim Ethington,
vice president-product for The Climate Corporation. “There is too much random
variation that unfolds.”

Still, company officials say its data
can predict that a certain type of weather—such as drought—is more likely to
occur than not.

“You can make a stronger consideration
toward planting more drought-resistant hybrids than previously planned, maybe
up from 20% to 40%, says Ethington.


No
Corporate Suits Here

The Climate Corporation isn’t your
typical buttoned-down agricultural company. Casual but neat dress is the
everyday attire. Young twenty-something and thirty-something workers armed with
PhDs in disciplines ranging from physics to statistics stroll through the
collaborative open-office environment carrying laptop computers.  Co-workers hold meetings in rooms titled “Edison”
or “Doppler”.

“To build what we built, we had folks with
traditional large-scale computer experience who combined with people who had
backgrounds in applied physics and statistics,” says Friedberg. “They are able
to use those skills in uniquely solving problems that we have to build crop
yield predication statistics.”

The firm’s
products revolve
around three components.

· 

Hyper local weather monitoring.

“Rather than the
airport 20 miles away, we record (the weather) that is happening right in that
field,” says Jeff Hamlin, the firm’s director of agronomic research. “What is
happening in that field is different than in the field across the road.”

·     

Agronomic models.

Research teams built agronomic and
yield models partly based on decades of USDA crop-yield statistics for each
U.S. planting region. The firm acquired geospatial data, and also built soil
moisture and topography models for individual fields.

·     

Weather simulation models

. This includes a review of
past, current, and future weather trends for U.S. fields.

All this has enabled the firm’s yield
models to outperform those of the USDA-National Agricultural Statistics Service
(NASS), says Erik Andrejko, agronomy lead for the firm. He attributes that to
factors like accuracy-boosting factors like hyper-local measurements.


So
Who Owns The Data?

According to Monsanto and The Climate
Corporation, farmers do.

“Information that a grower enters into
the system is private to that grower,” says Ethington. “They have the ability
to share it with other people, like a farm manager, but it would not be shared
with others unless the grower gave permission. We would never share the
grower’s data without their knowledge or permission.”


Look
For The Science

Many high-tech products are coming
that promise precise agronomic information. Look for the science that’s behind
them first, says Friedberg.


It is very
easy to put a veneer on a website, and say this is cool, this is the best
product out there,” says Friedberg.

Dig deeper.
“Ask
the software provider to show where it will create value,” says Friedberg. “If
they cannot verify it, even if it is free, it will cost you later on.

“These are not simple problems to
tackle,” he adds. “They are easy to solve in aggregate on the national level.
But when it comes down to the field level, the system is difficult to build. It
requires a significant investment.”

Read more about