[Paul Morin] We're the last one for tonight. So, we're about to talk about the Impossible Project.
This is a project that required--
This is a project that required multiple science agencies
and also a Department of Defense and intelligence agency.
[You can't hear us? Is this closer? Okay.]
What we did was to use high resolution, sub-meter, commercial imagery.
So this is imagery with the resolution of about this much,
collected over a period of about five years,
to create an elevation model that's publicly available for the Arctic--
[Compton Tucker] from about two million scenes--
[Morin] with millions of scenes of imagery.
[So next slide.]
We used imagery in this case
from six satellites that were licensed to the US government
by the National Geospatial Intelligence Agency.
They're basically orbiting telescopes
but instead of something like Hubble that's pointed out,
these are pointed down.
It's pretty much the same thing.
Over a period of about four years, we took this imagery,
that was collected forty-five seconds apart,
ten gigabyte strips and and we moved about a hundred terabytes of this
imagery into a place where [what? a hundred terabytes]--a petabyte of imagery--
that allow, into computers, that allowed us to process this into high-resolution
elevation models. [Next slide.]
One of the most amazing things about this ...
the satellite constellation is their very high inclination for orbit.
They're going basically from 84 or 85 all the way around.
So, at the poles, we can see
any one location in the Arctic, or the Antarctic,
as frequently as about 15 times a day.
What was delivered in September of last year was this.
This is a 2-meter posting--so, basically the height of a tall man--
elevation model. Publicly available. No copyright at all.
It includes, you know an area that's about 20 million square kilometers.
But to be able to make up this 20 million square kilometers,
it took imagery of over 200 million square kilometers to make this.
What you're seeing here is extremely good elevation for Greenland and Siberia.
We have poorer elevation here that's soon to be corrected.
This is what's now available, public.
[Next slide, please.]
The first delivery of this information was Alaska.
In 2015, President Obama proposed this project
and we had 12 months to not only create
Alaska from basically raw imagery,
but we had 12 months to create the factory that
created Alaska.
What we have here is the next generation of Alaska that
will be delivered at 2 meter resolution in about August or September of next year.
[Next slide.]
The North Slope [, Alaska].
We can...
We've created the elevation for the North Slope, no less than six times.
Unlike elevation that we have in the lower 48,
where you only have one collection every five years,
every ten years, or maybe not at all,
we have the North Slope once a year.
That allows an investigator to go in and look at thermal karst collapse,
to look at deforestation, to look at tundra fires.
[Tucker] So for vegetation this is remarkable because you can look at the
volume of vegetation, because these are three-dimensional images that are being produced.
It also enables you to look at ice stream velocities
and many elements of the cryosphere.
You can do this year after year after year
at the spatial resolution of 2 meters by 2 meters in the XY
and a vertical accuracy on the order of 15 centimeters.
So, this is remarkable for our study of the polar
regions in the Arctic and in the Antarctic.
[Morin] And so now we get in where the rubber meets the road,
because these are polar-orbiting satellites--
because we don't need fixed-wing logistics to bring lidar in--
we can see anywhere, all the time.
In this case what we're looking at is
Yakutsk along the Lena River in Siberia.
This is an auto-generated, 2-meter posting elevation model of a braided stream
and one of the most largely, one of the most highly-populated areas in Siberia.
You can see in this elevation road cuts.
You can see deforestation and just timber harvesting.
You can see airfields, and of course you can see the buildings themselves.
[Next slide, please.]
We're, again, independent of geography.
This is the Kronotsky Gulf in Kamchatka
so actually North is this direction
and you just have volcano after volcano after volcano after volcano
and because these satellites are not just simply collecting as they orbit, they're actually tasked,
we're tasking every volcano along the Kamchatka Peninsula monthly.
So these elevation models can be laid upon one another and then subtracted.
Then you can see where the deforestation is,
where also the eruptions are.
[Next slide.]
Here's where Jim's work starts coming into play. [You want talk about it?]
[Tucker] Thanks, Paul.
So one of the really cool things, if you study the carbon cycle, is this is a three-dimensional problem.
If you only deal with the carbon cycle in two dimensions, you have no idea about volume or biological mass.
Because these data are three-dimensional, you can look at the volume of vegetation which is removed
and that translates into a weight because you have three dimensions to work with.
Here's one image from November of 2012,
here is another image of the same area from June of 2015,
and what you notice, all of these rectangular areas are areas where the boreal forest was cut and removed.
[Eric, the next slide.]
When you combine these images, you're actually able to look at
volumes of forests which have been removed.
For the carbon cycle this is massive.
This is what we hope to do in tropical and subtropical areas where the
combination of Landsat data and lidar data or SAR data.
With commercial satellite data, in the Arctic regions you can do it directly
from data at a 2 meter by 2 meter XY with plus or minus 15 meters in the vertical.
This is revolutionary for our study of the carbon cycle.
[You want to finish up?]
[Morin] What you saw before, that you could see the individual trees being
removed for the logging roads,
but now also this data is accurate enough that we can say that on this part of this clear-cut,
the trees were 12 meters tall before they were cut, and on the other side they were 20 meters tall.
So it's actually volume at that point and, remember,
we're not dealing with a Landsat pixel, which is 15 meters for panchromatic.
We're dealing with a high-resolution elevation model where the DEM is 2 meters.
[Next slide.]
Here's the scary one.
This is the Vavilov ice cap, if you go to the center of Siberia, all the way to the top,
right before you hit Canada, it's this ice cap.
So this ice cap is about 75 or 100 kilometers wide
and what we're gonna look at is a time series of this elevation model over a period of about four years.
[Next slide.]
Here's the next year. This is March 2014.
You can see a surge beginning here.
[Next slide.]
The surge continues.
[And the next slide.]
This surge completes.
That's sea level rise in action.
That's one hundredth of a millimeter of sea level rise per year.
Just from this one surge.
[And if you go to the next slide, please.]
Here's the difference.
So we're looking at these surges on a 2 meter by 2 meter pixel, year by year, month by month.
[Next slide.]
So if you're doing one end of the planet, you might as well do the other end of the planet at the same time.
Sometime this winter we'll be releasing an 8-meter resolution elevation model of the entire Antarctic as well,
a hundred thousand DEMS.
So, now,
after the Antarctic is released in about February or so,
the Antarctic will have better elevation than we have for the western United States
including California and Colorado.
[Next slide.]
[Thats it? Ok, ok.]
So, thank you very much.
Không có nhận xét nào:
Đăng nhận xét