Printable Version of Topic

Click here to view this topic in its original format

Unmanned _ Earth Observations _ Atmospheric Rendering using Real Observations

Posted by: scalbers Jan 12 2014, 09:03 PM

This might be near the edge of the UMSF scope, so if needed it can be moved to Chit-Chat. I've been working on taking a 3-D real-time cloud analysis, fed by Earth orbiting satellite and other types of remotely sensed and in-situ observations. This set of 3-D fields are then rendered to produce an all-sky image from a given vantage point. Finally this rendered image is being compared with actual camera images.

While this is done right now on Earth's surface, including the top of a 14000 foot mountain, it can be extended to work higher in the atmosphere, near-earth space, or other planets/satellites with atmospheres. The idea is to take into account scattering of sunlight, moonlight, etc. by various components of the atmosphere, including clouds, precipitation, and aerosols. Twilight is interesting to render as well.


Posted by: nprev Jan 12 2014, 11:31 PM

Seems like a good place for this to be to me, anyhow (these are Earth observations, after all), and thanks for sharing your work, Steve! Cool stuff!

Any chance you can apply this work to other worlds like Mars? Results might be interesting

Posted by: scalbers Jan 12 2014, 11:54 PM

Thanks, and good question nprev about Mars application. Here are a few considerations. First would be to hook up the software to information about Mars' atmosphere instead of Earth's. I also have topography data from Earth that would need to be replaced with Mars info. Atmosphere wise, Mars is mainly different in having much less Rayleigh scattering from air molecules, yet a good amount of aerosol (Mie) scattering from dust. The color of the dust would have to be included, as this is different than typical aerosols on Earth. Then comes the task of setting up a cloud analysis for Mars, maybe from a combination of Mars orbiter images and atmospheric modeling...

Perhaps a simplified experiment that neglects clouds and simply looks at the gas and dust components would be easier. This would just use a more limited set of subroutines in my software. One of the main variables then would be the aerosol optical depth.

Posted by: nprev Jan 13 2014, 12:20 AM

Understood. Could have some interesting applications for future solar-powered landers/rovers. Seems like Mars is the only place this work could be applied other than Earth.

Posted by: scalbers Jan 13 2014, 07:48 PM

Interesting point about the solar power aspect, as a sky brightness/radiance map can help calculate the power with a particular tilt angle. Thus there could be solar power applications to both Earth and Mars.

Depending on how much the software can be generalized (or emulated), I think imagery can be rendered for such wide ranging places as Titan or the atmosphere of Jupiter (for example). At least Mars should be a relatively straightforward case to try making an image based on aerosol scattering. It should be interesting to try and reproduce various times of the day, including sunset.

Posted by: scalbers Jan 26 2014, 07:36 PM

Here's my first simple attempt for the sky on Mars:

This could be refined with information from actual panoramas, particularly if we happen to know the solar elevation angle. I could try to fit the simulation with a panorama showing just a partial sky, then use it to fill in the missing sections.

Here's an animation with the sun dropping lower in the sky (from 60 degrees altitude down to the horizon), displayed using a cylindrical projection:

There is another thread called where I'm putting more info about Mars sky renderings.

Posted by: scalbers Feb 14 2015, 02:39 PM

The software package I'm building now allows a vantage point off the surface, tested to about 2000km altitude. Here is an animation showing a launch into the stratosphere up to 40km above Boulder, CO. The 360 degree azimuth range cylindrical panorama also shows a full altitude range from +90 to -90 degrees. Frames are in 1km altitude steps, starting at the surface. It's interesting to see the pace at which the familiar blue sky above fades into the blackness of space.

This is only shown to 40km altitude since the area where cloud data (from Feb 14 1730UTC) has been analyzed is relatively small. The still image below is from the middle of the animation at about 20km. Balloon flights having cameras are often done to the stratosphere for comparison.

Posted by: stevesliva Feb 15 2015, 05:09 AM

That's pretty neat! It's like the god's-eye view from google earth with cloud deck and haze added. It would be neat to have the google-earth type datasets combined with the meteorological ones.

Posted by: scalbers Mar 6 2016, 07:51 PM

Thought I'd mention some animated renderings of the 2017 solar eclipse (of course without the real clouds yet) that I've been working on. These are ground level and stratospheric level views.

Posted by: scalbers May 5 2016, 08:33 PM

Another interesting perspective is looking at the Earth at night from Low Earth Orbit (LEO) altitudes. Here is a recent rendering with a wide field of view:

The rendering was done without clouds at an altitude of 500km over the Yellow Sea. We can see the city lights, airglow/nightglow, twilight arch, and the inner zodiacal light (a.k.a. outer solar corona). The imagery can be compared with actual LEO views.

Posted by: JohnVV May 6 2016, 12:07 AM


if you are interested
i put together a earth lights map a bit back

from the VIIRS data

Posted by: scalbers May 8 2016, 06:31 PM

Good to see you are working with VIIRS as well. I've been using the monthly stable lights tiles from NGDC/NCEI. The GEOTIFF images available are broadband radiance in units of nanowatts / cm**2 / sr. For direct display these can be gamma corrected and then colorized. For my rendering software I still input the radiance, and then try a simple assumption to convert this to spectral radiance in passbands corresponding the RGB colors.

Interesting to consider what the colors should be. Lights are trending over time from more orange (high pressure sodium) to white (LED) in many cities (maybe more in their centers). Other cities have a history of more mercury vapor (greenish white). The scattering by clear air and clouds is something I'm handling pretty simply right now and will need more development. Emitted radiance can also vary with the look angle.

Posted by: JohnVV May 8 2016, 07:36 PM

i used the iss citys at night images for the painting the colors

that is after i removed clouds and noise but not lights -- not as easy as it sounds

i like the crescent , but shouldn't there be a bit more orange / red near the surface on the upper right

Posted by: scalbers May 8 2016, 07:54 PM

In terms of the crescent the orange enhancement would mainly as noted be in the very lowest portion of the atmosphere (i.e. the lower half of the troposphere). This layer would be just over 1 pixel thick. Here though the intensity is so high that it saturates the brightness to white, even though it would indeed be a more orange color. This is a good question and more careful comparison with observed imagery might be of interest, particularly near the left and right edges of the twilight arch with higher resolution renderings. Here is a start to exploring this:

In the above, some red is visible in a thin line above the horizon. I should redo this one with the latest software version though and check the aerosol assumptions. More is included in the link below.

Posted by: scalbers Jan 22 2017, 06:01 PM

A redo of a sunset, this time from space at 400km altitude.

QUOTE (stevesliva @ Feb 15 2015, 05:09 AM) *
It would be neat to have the google-earth type datasets combined with the meteorological ones.

These renderings use land surface spectral albedo information from the Next Generation Blue marble, along with USGS topography data (albeit with limited resolution during closeup views). Below is a depiction of what a launch would look like during early twilight starting at ground level and going up to 400km altitude. The sun is two degrees below the horizontal, though it rises above the limb soon after launch. (animation)

The hemispherical animated view looks up for the first 15km then looks down. 3D clouds are included and some cloud shadows (crepuscular rays) are visible.

Powered by Invision Power Board (
© Invision Power Services (