Analyzing Imagery and Change Over Time

Peter Becker and Vinay Viswambharan demonstrate how Dubai has changed in the last 40 years.

Aug 30th, 2012

Embed
Download
Transcript
480x270
960x540
Custom
Width:
Height:
Start From:
Player Color:

Right-click on these links to download and save this video.

Transcript

00:01So let me tell you about my trip to Dubai last year.

00:03I was riding through the streets of Dubai and looking around and realizing, why are all these buildings so modern?

00:10So I turned to my friend Hossam and said, "Why are there no old buildings in Dubai?"

00:15And he said, "Well, the answer's simple: the United Arab Emirates is a country that's only 40 years old."

00:21The city, the country really is that modern.

00:25So that created the challenge and the question for our next presentation, which is to answer the question...

00:30...How much has Dubai changed in the last 40 years?

00:34To answer this, please welcome Peter Becker.

00:41Thanks, John. So let me show you.

00:42Here, a map of Dubai, with recent Landsat imagery.

00:49Let me zoom in to the old part of Dubai and change time back to 1975.

00:56Dubai, a small, coastal town, a single runway.

01:01By the year 1990, it had expanded, the town as well as the runway.

01:06By the year 2000, further expansion and more vegetation.

01:12Let's highlight that.

01:15We can see there's coastal development, and by the year 2000, they had built massive islands in the form of palm trees out into the gulf.

01:27By the year 2010, further development along the coast.

01:32But what if we want to actually measure this change?

01:35Well, I can use this tool to find an area of interest.

01:43And the system now computes the areas, returning for me that, in 1975, it was all barren land.

01:52By the year 2000, about 4 kilometers of urban growth; by 1942, over 40...by the year 2010, over 40 kilometers of urban growth.

02:03Vegetation had increased, and interestingly, the water area had actually gone down by about 12 kilometers...

02:10...due to the development of the palms.

02:13Now, what is the basis of this application?

02:17Well, there are three key things here - one, the world Landsat services...

02:22...two, the ability for ArcGIS Server to perform dynamic classification...

02:27...and the integration of these geoprocessing tools into a web application.

02:32So let me show this to you in ArcGIS Desktop 10.1, that many of you already have.

02:38Here, I can search online for Landsat imagery.

02:44This returns me the newly released Landsat services.

02:49They have been extended to include the new GLS 2010 data from the USGS...

02:55...as well as additional radiometric corrections and additional visually enhanced services.

03:00I can add these directly to ArcGIS Online, and they're available for you to use today.

03:08This service is a service that actually shows one of the capabilities of 10.1 to enable a single service to have multiple products.

03:16Here, I can pick the visualization analysis, and you will see the enhanced visual quality of these services.

03:26Let's zoom in to Dubai.

03:27The server performs the processing, returning me the required product.

03:32And from within Desktop, I can add additional enhancements, for example, add a dynamic range adjustment.

03:40As in the web application, we can use time and go back and have a look at what did Dubai look like in the year 2000?

03:49Or, move back forward to the year 2010.

03:55Now, although this is a seamless image service, I can actually get information about the actual pixel values...

04:03...as well as meta information about the individual scenes.

04:08Now, a new capability that we're adding in 10.1 is for ArcGIS Server to perform dynamic processing or dynamic classification.

04:19And to show this, I'm going to bring in this local server connection, and we'll see similar services based on the same imagery.

04:27If we now look at the products available, mind you, I can get to see, for example, this land/water interface...

04:37...or go and have a look at a full class land use.

04:43These same services I can then use in a geoprocessing tool.

04:47Here, I bring it up.

04:49It allows me to define the service's input, allows me to define an area of interest, and then, this will be processed.

05:03The tool is making multiple requests to that service.

05:06The server is dynamically calculating that classification, returning the classification results to the geoprocessing tool...

05:13...which then gets the statistics and generates the required reports.

05:19As soon as these are completed, they can be displayed.

05:22This same geoprocessing service is then shared with the web application to enable the dynamic reports.

05:30So I've shown how the extensive imaging capabilities within ArcGIS 10.1 can be used to measure the growth of Dubai...

05:39...but we know Dubai has also grown with a lot of tall buildings.

05:43And to show you how we can measure the vertical growth, let me introduce you to Vinay Viswambharan...

05:48...from the software development group.

05:51Thanks, Peter.

05:57So you just saw analysis and classification done at a city scale.

06:02I want to show you characterization of change at a finer level of detail.

06:06To do this, I need high-resolution imagery.

06:10So here's a WorldView scene over the area of Dubai.

06:14I receive this as raw, unprocessed TIFF files on disk.

06:17Now, one of the biggest improvements we've made in ArcGIS 10.1, the system leverages the metadata that's provided...

06:24...and it can define products automatically.

06:28So as I drag my pan-sharpened product into map, the system pan sharpens, orthorectifies...

06:35...and creates a visually good-looking product on the fly.

06:44The best is yet to come.

06:48So this is the Burj Khalifa, the tallest building in the world.

06:52It's an amazing structure, and that's some beautiful imagery to complement it.

06:57But that's obviously not the most intuitive view of the tallest building in the world.

07:02Using the sensor model information, we can now rotate my view and get a better perspective.

07:15Leveraging the sensor model information and using the meta information...

07:20...we've improved the analytical capabilities of ArcGIS even more.

07:24You can now make measurements off of your imagery.

07:28You can base it off the sensor geometry from the bottom of the structure to the top of the structure...

07:35...or you can make shadow base measurements from the tip of the shadow, tip of the spire, to the tip of the shadow.

07:46When characterizing change, you typically work with historical imagery.

07:51A common problem faced when working with historical imagery, your imagery doesn't line up.

07:58Let's navigate over to another place in Dubai.

08:03So here I have an older image.

08:05As you can see, the change is not quite obvious, but a problem with the image misalignment is a lot more obvious.

08:14These roads, they don't quite line up.

08:18ArcGIS 10.1 can fix this.

08:21We have this magic button.

08:24No, seriously, you'll like this.

08:27I'll bring up the Georeferencing toolbar, I'll hit it, and I'll tell you what it does.

08:32You're going to like this.

08:33The system is computing tie points between the two images.

08:37It's adjusting the images to fit automatically.

08:41Now, this is great for the historical imagery that I'm dealing with.

08:44And like I told you, it's a very simple workflow.

08:47Press a button, pick your images, no tie points, no control points, no check points.

08:52There is no complexity the user needs to deal with.

08:57Just like that, it fits.

09:08You can now compare the two images, automatic georeferencing.

09:14We can now navigate back over to the Burj, compare the two images, lines up perfectly.

09:21For five years, that is a lot of change.

09:25But if you notice, most of the change, much of the change is tall buildings, a lot of vertical growth.

09:32Vertical growth equates to massive change in floor area.

09:37Now, to compute change in floor area, or to quantify change, you need to measure building heights.

09:47So as I zoom out to the extent of Dubai, you can see that's a lot of data collection that I have to do...

09:52...and a larger collection of images to deal with.

09:55So here's where the mosaic dataset comes in.

10:00Now that's a lot of imagery.

10:03It's processed on the fly, and it's fast.

10:11As I zoom out to the extent of the Middle East, 2,500 scenes, GeoEye scenes, two and a half terabytes of imagery.

10:24It's processed on the fly.

10:33I've shown you all of this functionality.

10:36I've shown it to you in Desktop.

10:38But ArcGIS, it's a system, which means we can wrap all of this functionality into a rich, focused web application...

10:46...and I can crowdsource my data collection task to a larger group of people.

10:53So here's my web application.

10:56I've hosted it on my organizational account on ArcGIS Online.

11:01As I bring it up, the application lets me measure building heights and other attributes.

11:07And just like in Desktop, from the tip of the building to the tip of the shadow, I can attribute my building.

11:17Image services, like all of ArcGIS, are time aware, which means I can actually step through these images in time...

11:28...and you can see, the structure was first observed in 2009.

11:35On the right-hand tab, I have my reports populated; I have other statistics as well.

11:43So to answer John's question. How much has Dubai changed over the years?

11:48The city has changed a lot.

11:52With 96 percent of buildings captured by my organization, the approximate increase in the total floor area - 190 square kilometers.

12:03But John, why just numbers?

12:06Using these measurements, I can actually show you what it looks like.

12:13So here's the beautiful city of Dubai at night.

Copyright 2014 Esri
Auto Scroll (on)Enable or disable the automatic scrolling of the transcript text when the video is playing. You can save this option if you login

Comments

No comments. Be the first to write one below.

Comment on this Video