Great post. I'll probably be pointing people to it for years to come, as I hear these kinds of questions pretty often.

Frank,

The question that I keep getting over and over and over and over and over is how do we tell how old a particular image is. I keep seeing people refer to "oh that image is from 2006" or "that's 6 months old." How do they know?

I have to agree with the comment above, this is a great post, hopefully people will read it and learn that not collecting hi-res imagery from space is not easy.

Dude,

People use various things that do or don't show up in images to estimate date. For instance, if a new house is built in a neighborhood but doesn't show up on the imagery, You know that image can't be very recent. Or, if road construction is going on in the imagery, it's fairly easy to date it if you know when that construction would have been taking place. You can use other methods to determine these things too, such as looking for leaves on trees (winter vs. summer) to narrow it down to a specific time.

Great post, Frank!

Great Post Frank! It's always great to read what you are digging up here regarding Google Earth. I have a huge directory of KML files that need to be looked at in GE, and most of them have come from here. So, thanks!

I almost wonder if the attitude of people thinking that the imagery in GE is real time stems from popular culture. Specifically, I'm thinking of a part of a Simpsons episode where Marge zooms into her house and finds Homer sleeping in a hammock in the back yard in a scandalous fashion. Little references like probably influence people's perceptions. Just a thought.

@Dude: I think one way people can determine the date of the imagery is simple imagery interpretation. For example, if someone zooms into their house and sees a car that they don't have anymore but can specifically recall when they have it. Details like that would provide a useful range for when the picture was taken. Obviously, though, you couldn't get an exact date, but still.

Just my $.10.

Cheers!!

Agree with Dude: It continues to be a rotten shame Google won't bother to publish a vector layer like the Digital Globe set that would list the date each imagery "subarea" was taken, remote sensing source, and date added to Google Earth. They surely have a back office system that tracks this. What's the harm in letting users know more about what we see? In the absence of facts, everyone I know disregards it as just a dumb, unpredictable, and often misaligned backdrop graphic. But then again Google's entire system has ducked decades of work on metadata standards.

This is the first full explanation of the imagery process for Google Earth that I have read but it still annoys me when I load the Digital Globe layer and seeing rectangles of 0% cloud cover imagery over areas which do not have high-res coverage yet. I'm still not sure if Google has a whole database licence or if Google buys the imagery depending on what it is.

Well done setting the record straight, Frank. And as always, in clear language that us non-developers can understand!

I think people also assume that satellites are like digital cameras, continuously streaming ready-made photos down to Earth. As I understand it, a lot of of them are multi-spectral sensors picking up data that we can't even see with the human eye (infrared, for example). Many of those NASA photos of the San Diego fires were actually false-color images. That is, someone had to process the image to make it *look* like a photograph—it didn't just appear that way naturally.

Much of the imagery (at least for North America) has actual dates attached, right within Google Earth. Look in the layers, under More->DigitalGlobe. If you play around with various years/cloud cover layers, you'll see the rectangular sections of imagery as they were taken.

Now, the only trick is, a particular area might have several dates attached - you can click on the actual label and go to a preview of the imagery in your browser - DigitalGlobe uses this preview function to sell the original imagery, if you'd like. Take a close look, see if the seasons/cloud cover match to what's actually displayed in GE, and 90% of the time you will be bang-on.

I live in a city (Calgary, Canada) where we literally have changes to our roads and buildings daily, and this method has so far worked with every imagery update. Some of us who track development know to within a week when things change, and the DigitalGlobe dates are accurate if you are careful.

Maybe Frank could write an article for GEblog that covers this a bit clearer than I did :)

It's true that the new layer became available last December, but not forget that Aqua and Terra layers have been availale for a few years. They're updated daily too, the only difference is that Daily PLanet is a global mosaic, whereas Terra/Aqua have only some parts of the Earth surface - the rest is black/transparent.

Another interesting feature is the MODIS archive - you can go back in time and see the Terra/Aqua images from all the days back to 2004-12-25. So anyone can still see the California fires. To do so just use time parameter (time=YYYY-MM-DD) in WMS request - you need some WMS client program with time support or World Wind and this plugin http://www.worldwindcentral.com/wiki/MODIS_historical_image_layers_plugin
Maybe something similar could be done for GE...

If there's nearly real-time data available, why not crowdsource the whole quality control and review process by letting people mark rectangular spots on the imagery which is good quality - just like the NASA crowdsourcing project of marking craters and "interesting spots" on Mars imagery?

I was wondering who is in charge of changing the pictures. I have a project where we are building condo's however the old plant( now gone ) is still in the picture.
237 Monroe Street
Saline MI 48176

Any help for who to ask for an update photo would be great.

I've worked for a satellite image distributor and the problem with displaying dates is that when you provide imagery from so many different sources, the dates can be incorrect on a relatively regular basis. For example, if you do a large aerial survey, it may take days or even weeks to take all of the imagery. However, when the data is delivered, it usually has one date attached to all of the data. Depending on the source though, most dates are accurate to within a month or two. I would say that dates associated with Digital Globe satellite data are accurate to the day.

You also run into that issue when data passes through a number "middle men" before it reaches the customer. This seems to be the worst when private companies resell government data. I've seen differences of up to two years between two different companies selling government data when it was obviously the same picture.

A friend said he chatted online, he's in northwestern Canada, with a family member in Ohio, his brother said he read the name off a store sign (with Google Earth) across the road and also asked what the pipe in neighbors yard was, can this be possible? Was his brother looking at years old imagery? If it is possible even if not in 'realtime' then how does one access it?

Curious in Northern Canada

I was wondering if i could use an image from google earth for my High School year book cover?