Landsat in Action - Google's Earth Engine with Noel Gorelick

Video Transcript
Download Video
Right-click and save to download

Detailed Description

Google's Noel Gorelick talks about the value of Landsat data in the Earth Engine project.


Date Taken:

Length: 00:03:56

Location Taken: Sioux Falls, SD, US


My name is Noel Gorelick and I work
for Google.

I'm one of the founders of Google
Earth Engine. Back in around 2008

we were working with some people
who were trying to do deforestation

mapping in Brazil. Very
specifically people looking for

people trying to cut down trees
that shouldn't be doing so. And the

very first idea for Google earth
engine was Google trees. We would

actually make a tree counting tool,
but that was the the basic impetus

was to be able to take data like
the Landsat archive and count

trees, map cities and provide the
ability for people to access that

data and and use it easily. And
quickly turned into something much

larger. We’re going to do more than
count trees, right. If we gonna get

all this data were going to a lot
more than that and so over the

course of literally a week, me
and my partner, the first 2 people

on the team said okay we know we
can do all kinds of land cover

mapping with this.

People use Earth Engine a lot for
processing Landsat data but the

really nice thing about earth
engine is that all the other data

sets are there and you can merge
together. So it's actually somewhat

uncommon for a user to just use
Landsat because it's so easy to

bring in MODIS and Sentinel and a
bunch of other data sets. Most

applications sooner or later turn
into handling many different kinds

of data sets along the way. Until
this year Landsat was by far the

most used data set on the platform
Sentinel 2 is catching up very

fast. The really nice thing is you
can use them together. So almost

all Sentinel 2 users are also
Landsat users.

Until that happened, being able to

combine old calibrated Landsat 5
with newly calibrated Landsat 7

both geometrically and
radiometrically was questionable.

People did it, they weren't always
happy with the results. And now

they're sort of a stamp of approval
and a consistent calibration across

the whole archive that make that
vastly easier. So I think that's

the best thing Landsat has done for
their users since the data became free.

I am excited about it working. I
think it's definitely the next stop

for everybody that wants to be
using Landsat data. One of the

outcomes I hope is a per image land
cover classification. So every

single image will have the cover
classification attached to it and

that's really what users end up
wanting. They just work with the

raw data because that's what they
have to do but if you could just

say this is an urban pixel I think

happy with those results and not
bother with the raw pixels.

The exciting thing about where the

Landsat program is going and the
fact that Earth Engine gets to help

is that the 5 most advanced
scientists in the world doing

remote sensing analysis can do some
work and then in a few weeks,

months, maybe years, everybody else
can do the same thing and that's

quite revolutionary for everybody
else. The cutting edge of science

moves on at the pace it has more us
always moved on but, the ability to

bring the the rest of the world
along for the ride at a much more

accelerated rate is the part we’re
excited about. So when the Landsat

science team members who are
arguably probably the best in the

world at doing this kind of
analysis make breakthroughs we can

transfer those breakthroughs to the
Peruvian minister of forestry and

beekeeping in a matter of weeks or
months. I'm excited that 8 years

ago the things that we thought were
going to be the future are in fact

the future. The future is now and
and we're quite happy about it.