Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Lunar maps in WinJUPOS, a quick method to accurately align the reference frame

  • Please log in to reply
16 replies to this topic

#1 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • topic starter
  • Posts: 2,114
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 15 February 2020 - 01:37 AM

Many people are aware of the utility of WinJUPOS in planetary imaging, and a growing number are now aware that it can also be used to create remarkably accurate maps of the Moon using amateur images, as has been showcased in previous threads by several users.  For those interested in this topic, I wanted to draw attention to a very quick and easy way to accurately align the WinJUPOS reference frame to any lunar image, no matter how small the ROI, or when it was taken.  When you select "Moon" as the object in WinJUPOS and then click on "Recording" and "Image Measurement", you then open your image and select the appropriate time, date, and location.  The exact time and location are very important because they determine the perspective view of the Moon.  

 

This is the point at which most people would just resize the reference frame until it fits around the limb of the Moon and the terminator, assuming you have a full disk image of the Moon.  However, getting a very precise alignment is not as easy as you might think, because the outline provided of the Moon and terminator just represents the mean lines assuming the Moon was a smooth sphere, which it is not.  In particular, the lunar limb is not smooth, and some mountains stick up in relief.  Do you include those within the reference frame, or not?  Similarly, the terminator line is just a theoretical line dependent only on the sub-solar point, but the "real" terminator that divides lunar day from night is dependent on topography and is not straight.  So, where you do place the reference line?  Also, if you have an image that just represents a small region of the Moon, how do you align the reference?  You can use a resized NASA simulated image and then place your image over it and load that into WinJUPOS, but this has two problems.  The first is that it adds the extra step of having to align your image to the NASA simulation, and this has its own amount of inaccuracy.  The second is that you are limited to time periods in which you can generate a suitable simulated image, and even then you can only get within an hour of time accuracy.  Needless to say, doing this repeatedly with many images quickly becomes tedious.  

 

There is a common solution to all of the above issues, and that simply involves selecting the option of "Adjustment according to two known points".  Below I will provide a few examples of what I am talking about.  Because each post can only contain 500kb of images, I will have to post this in succession.  Also, this entire post is based on the assumption that one is interested in creating a map that has very accurate latitude and longitude coordinates.  Why would you want to do this?  Mostly just for fun and the challenge, much like anything else in solar system imaging.  If your only goal is to create a projection that looks nice, you don't need a very accurate reference frame.  

 

The starting point for this example is the cropped image below.  I wanted to use an example that did not contain the lunar limb.  This image does contain the terminator, which illustrates my earlier point of how you would be hard pressed to decide exactly where to place the terminator reference line based on the image.  But the method shown below doesn't require any terminator regions either.  You could use an image of a single crater from anywhere on the lunar surface.  

 

Ptolemaeus-09-21-19-TG.jpg


  • R Botero, james7ca, roelb and 5 others like this

#2 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • topic starter
  • Posts: 2,114
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 15 February 2020 - 01:41 AM

Shown below is a screen grab of the menu bar on the left that you want to select, and then choose the option highlighted. 

 

Winjupos-menu.jpg

 

You can then pick any two points in your image.  If you click on the points, a reticle will be placed on your image.  You can use the arrows to make fine adjustments. You then need to enter the selenographic coordinates for that point.  You can get those coordinates from the following link.  

 

https://quickmap.lroc.asu.edu/

 

If you navigate the LRO quick map, you can zoom to anywhere on the lunar surface and get a latitude and longitude readout.  Enter those numbers into WinJUPOS, as shown below.  You can see that I did so using a small crater in the floor of Ptolemaeus, and in Bullialdus.  You'll have to look closely to see the tiny white reticle. 

 

Ptolemaeus-Winjupos-alignment-TG.jpg


Edited by Tom Glenn, 15 February 2020 - 01:42 AM.

  • Euripides likes this

#3 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • topic starter
  • Posts: 2,114
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 15 February 2020 - 01:43 AM

Once you click "OK", the reference frame will be aligned.  If your image is free from distortion, and you chose your points carefully and entered the numbers correctly, it should look good.  Shown below is an example of what happened in this case.  

 

LTVT-reference-alignment-TG.jpg


  • DRohrman and Euripides like this

#4 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • topic starter
  • Posts: 2,114
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 15 February 2020 - 01:50 AM

Given what I said before about the terminator, you may wonder about the accuracy of the placement.  There are programs which will draw the mean line of the terminator for you on the lunar surface.  One such program is LTVT (lunar terminator visualization tool), of which there are threads about on the lunar observing forum.  I took the LTVT simulated image and overlaid my image on top, and shown below is the WinJUPOS terminator reference line in white, with the LTVT terminator lines in red and blue.  There are two lines, because the red line represents the points at which the Sun is completely above the horizon, and the blue line for where the Sun is completely below the horizon.  Again, this only pertains to a point on the mean lunar sphere, because as you can see from the image, it has no relevance to which locations are actually receiving sunlight (because of local topography).  What is relevant is that the white line (from WinJUPOS) lies between the red and blue lines.  There is a slight asymmetry to the arc, as it deviates slightly from dead center between the red and blue lines, but overall this is a remarkably accurate terminator placement.  

 

Ptolemaeus-LTVT-alignment-TG.jpg


  • DRohrman and Euripides like this

#5 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • topic starter
  • Posts: 2,114
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 15 February 2020 - 01:52 AM

Once the reference frame is placed to your liking, you save the image measurement file, and then click on "Analysis" and "Map computation".  You then get a window that looks like that below, and you open your measurement file and select the options you'd like for your map.  

 

WJ-map-computation.jpg


  • Euripides likes this

#6 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • topic starter
  • Posts: 2,114
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 15 February 2020 - 01:58 AM

Then, how to check the accuracy of the map?  That's for you to decide, but one way that I think is fun is to select a point on your map, and get the latitude and longitude readout from WinJUPOS.  Then, load those coordinates into the LRO Quickmap and see how close you were to the real location on the lunar surface.  Shown below is an example.  I selected a small crater, just to the south of the main crater chain of Catena Davy.  You can see from the image that the predicted coordinates for this precise location in my map are -6.91W, -11.51S.

 

Winjupos-map-prediction-TG.jpg


  • DRohrman and Euripides like this

#7 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • topic starter
  • Posts: 2,114
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 15 February 2020 - 02:01 AM

If I locate the coordinates -6.91W, -11.51S in the LRO Quickmap, it reveals the position shown below, which is 500m away from the center of the small crater that I was aiming for.  This value is less than the resolving power of my telescope, and so certainly falls within an acceptable margin of error.  More on accuracy below, however.  

 

LRO-map-error-measurement-TG.jpg


  • DRohrman and Euripides like this

#8 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • topic starter
  • Posts: 2,114
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 15 February 2020 - 02:05 AM

Shown below are two more examples of how the reference frame can be aligned without ever resizing or moving the actual outline of the reference frame.  Even in seemingly trivial cases of images in which you have the entire lunar disk, this method is very easy, and takes all of the guesswork out of it.  An added bonus is that even if you aren't interested in computing maps, aligning the reference frame gives you a very accurate image scale, from which you can calculate what your actual focal length was if you so desire.  

 

Note how in this example below, the terminator line would again be somewhat tricky to determine, as would mountains along the lunar limb.  I prefer just to measure a few points rather than guess.  I don't actually show all the steps this time.....just the final results.  

 

Winjupos-reference-southern-limb-TG.jpg


  • DRohrman and Euripides like this

#9 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • topic starter
  • Posts: 2,114
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 15 February 2020 - 02:06 AM

In this case the entire Moon is included in the image, but again, why mess around with resizing and rotating the reference frame?

 

Winjupos-reference-first-quarter-TG.jpg


Edited by Tom Glenn, 15 February 2020 - 02:06 AM.

  • DRohrman and Euripides like this

#10 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • topic starter
  • Posts: 2,114
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 15 February 2020 - 02:23 AM

That's it for images.  As a general note on accuracy, I see pointing accuracy within 3km in most places in the maps.  If you measure from fairly close to your calibration points, it is often sub-1000m accuracy (as shown in the example above), and this accuracy goes down somewhat the farther you get from those calibration points, but again it is usually just a few km error.  The exception to this is near the limb and poles, in which foreshortening causes a small angular error to translate to a rather large mapping error, sometimes up to around 15km if you are very near the limb (assuming you didn't use a region nearby to calibrate the image).  

 

The accuracy of the reference frame alignment is also an indicator of the integrity of the original data.  If you were imaging with a setup that for some reason introduced perspective distortion, this would make it impossible to create an accurate map.  Also, if you are measuring from a mosaic, there will be some error introduced in the alignment of the panels, and so these map measurements are somewhat of a test of the mosaic projection.  And it's almost always more accurate to align an individual image panel rather than a mosaic.  These map errors also have interesting implications for highlighting the inherent inaccuracy of our Earth based images.  As was discussed in another thread recently, one of the reasons that mosaic panels fail to align is simply that each stacked image represents a set of thousands of APs which are forced to align to a reference image.  This requires local warping of each AP, and this actually changes the relationship between lunar features in a measurable way.  You don't notice these changes when looking at images, but you definitely notice them when measuring against a gold standard (LRO).  Also interesting here is the implication that mapping accuracy is actually somewhat better if you crop a larger image (which will have some distortion from stacking) into smaller ones, and then map the smaller ones independently using calibration points within the small region.  This is much more similar to the way LRO mosaics are actually made, because LRO images are not registered to each other, but rather projected precisely onto the lunar surface, and the overlap in the mosaic comes from the accuracy of the projection, which we will never have from Earth.  That's it for now, just a few interesting things to think about on a Friday night.  


Edited by Tom Glenn, 15 February 2020 - 02:24 AM.

  • R Botero, roelb, DRohrman and 1 other like this

#11 astrolexi

astrolexi

    Explorer 1

  • -----
  • Posts: 81
  • Joined: 30 Apr 2018
  • Loc: Saulheim, Germany

Posted 15 February 2020 - 09:11 AM

Thank you for the incredibly detailed information, Tom!

I will definitely try to apply your Friday night ideas to an appropriate image in the near future.

 

And why?..."just for fun and challenge".

 

Best wishes

Klaus



#12 sunnyday

sunnyday

    Skylab

  • *****
  • Posts: 4,399
  • Joined: 23 Sep 2019
  • Loc: Ottawa,Canada

Posted 15 February 2020 - 03:49 PM

thank you a thousand times.



#13 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • topic starter
  • Posts: 2,114
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 16 February 2020 - 04:13 AM

Thanks Klaus and Sunnyday.  

 

Klaus, your original thread here was possibly the first to get this WinJUPOS lunar projection ball rolling, so kudos to you for that! 


  • astrolexi likes this

#14 james7ca

james7ca

    Cosmos

  • *****
  • Posts: 7,900
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 16 February 2020 - 04:33 AM

This looks to be a useful technique and it certainly is more accurate than aligning to the Moon Visualizations provided by NASA (Dial-A-Moon). Thank you Tom for all of the details.



#15 Luc CATHALA

Luc CATHALA

    Messenger

  • -----
  • Posts: 426
  • Joined: 28 Jul 2013
  • Loc: Mazan 84380 South France

Posted 16 February 2020 - 04:36 AM

Hello Tom

Thank you for these detailed and very interesting explanations.

Clear skies.

Luc



#16 astrolexi

astrolexi

    Explorer 1

  • -----
  • Posts: 81
  • Joined: 30 Apr 2018
  • Loc: Saulheim, Germany

Posted 16 February 2020 - 12:26 PM

Thank you Tom.

I'll pass the kudos on to Nicolas Dupont-Bloch ('Shoot The Moon')...

 

Best wishes

Klaus



#17 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • topic starter
  • Posts: 2,114
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 18 February 2020 - 03:14 AM

Thanks to all for the additional comments.  WinJUPOS provides a very nice tool kit for a free and easy to use program that many of us are already familiar with.  That said, there are better programs out there for mapping, many of which are designed for geologic mapping on Earth, but could be coopted for lunar images.  Also, I'm aware of examples in which people have used NASA elevation data and "colored" that data with their own images.  Luc actually posted an example of that a few years ago.

 

https://www.cloudyni...ver-posidonius/

 

One thing that appears to be lacking on this forum is a discussion about these 3D rendering, other than some very cursory explanations.  I see that some people use QGIS for mapping, but more detailed methods would certainly be widely appreciated.  Although looking at 3D images and maps is very entertaining, it doesn't have much impact without a more detailed description of methods used.  That is partly my motivation for this post, and in the case of WinJUPOS, it is actually very easy to use these techniques.  I'm not sure how easy some of the other mapping programs are, and many require a bit more computer programming skills.  




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics