2017 Madras

Over the last 18 years eclipses have taken me to places that I likely would never have gone on my own. Who would have thought I would be standing on a frozen river in Svalbard, a lake in Novosibirsk, a farm in Zambia, or riding a ferry in Sierra Leone. Yes there have been other places that could easily satisfy a bucket list (e.g. Pitcairn Island and Easter Island). And some of the other places such as Cairns Australia or Shanghai China are places that I have visited before.

This time was very different. It was a trip north to Oregon (another place I had been before). In fact, I suspect that at one point we drove through Madras as a part of a family vacation.

The lack of long international travel and the way the trip was organized allowed me to bring first rate gear including a heavy mount, computer, and an observing table. My goal was to exceed the quality of the work I did in Libya in 2006. A goal I believe I met.

2017 Madras

The Journey

Our group based itself in Bend which, without traffic, is about 1 hour south of Madras. We spend Saturday night in Portland and traveled through Madras on the way south. We had been monitoring the preparations in Madras. Still I was amazed by what we saw as we approached the town

The plan was to leave Bend at 4 AM on Monday. Needless to say this seemed risky, but seeing the number of people who heeded the advice to arrive early gave us some confidence.

When the appointed time came our buses rolled out of their hotels and headed north. Rick Fienburg traveled ahead and reported no delays. That was our experience as well.

Traffic to Madras at 4 AM

Shortly after 5 AM we arrived at our viewing site comfortably before dawn. Our trip leaders advised us to delay the setup until dawn as the field was rather treacherous in the dark.

Once the sun rose we could see just how beautiful the site was. There were clouds of smoke to the south, which at time, did not seem worrisome.

The Site

Our group arraigned for us to view the eclipse from the soccer field at Jefferson County Middle School

This was a perfect site!!! Our group had exclusive access to the site. Once it was light we moved our equipment onto the field. The grass was rather long, but it felt like walking on a pillow. No problems setting up equipment.

Across the field with my equipment in the foreground	The traditional pinhole sign Photo courtesy of Dave Darrah	Chris from FPOA and myself
My equipment	Another of my setup

We had plenty of time to get setup. I did have a couple of emergencies. Once the mount stopped tracking. After C1 I noticed the focus was soft. So it was a rather start and stop morning with periods of big smiles and times of near panic.

Smoke was also an issue. The cloud we saw at sunrise became thicker and headed over us. Fortunately by C2 it had largely past. I think some of what I see in the diamond ring pictures is due to smoke (or that's my story).

The exact location was 44° 38' 5.020" -121° 6' 30.530" 731 meters

Movie of Eclipse Experience

Movie of Totality

As I say in the movie I have seen several montages of the eclipse. While an effective way of presenting the event it lacks the ability to convey the rapidly changing views at C2 and C3. I think this movie makes the point more effectively. It can also serves to document all of the various exposures I used. The base for all exposures is f 6.3 ISO 200 and then varying the shutter as indicated.

Images of the Eclipse

Here are the best of the best. The Corona images were used in the HDR.

Partial Eclipse

2017 Total Solar Eclipse Madras Sunspots

'2017-08-21T16:20:44' 1/350 ISO 100

2nd Contact Diamond Ring

1/250

There were no prominences visible at C2.

2nd Contact Bailey's Beads

In 2010 I made capturing the beads a goal. Unfortunately without full computer control I did not do a very good job. This time I nailed it. .... and they were very good beads.

1/8000

3rd Contact Prominence

While C2 was a disappointment in the Prominence department C3 more than made up for it.

1/4000

C3 Bailey's Beads

Again the beads did not disappoint

1/8000

2017 Solar Eclipse Madras beads and prom

High Dynamic Range Pictures of Corona

High Dynamic Range is a processing technique to overcome the limitations of a camera's limited dynamic range (the ability to represent different levels of brightness). While cameras are fine capturing uniformly lit outdoor scenes, they fail badly when the image contains very bright and very dim portions. Since electronic displays only permit 256 levels of brightness, HDR will sort through the image brightening dim sections and dimming bright sections to form a compromise image. HDR can work very well as shown in the example in the Wikipedia pages. Eclipses are notoriously difficult since the brightest and dimmest portions are many orders of magnitude apart. Buried in that range of brightness is the detail our eyes see.

Some astronomers such as Druckmüller use special tools and techniques. For the rest of us we try to get something that approaches what our eyes can see.

The basic technique is to capture the image at a variety of exposures. Software will then select what portions of which image to use in the combined image and how bright those components will be.

I got a good image of the inner corona the first night. The outer corona was more difficult. Also even Druckmüller's images do not adequately display the vast difference in brightness between the prominences and even the brightest portions of the corona. Others have produced good results on this eclipse. I finally got a result I am willing to share.

Inner Corona

I present my HDR of the inner corona two ways. The first is as produced by the tools. The section is with the prominences Photoshopped (although I really used Pixelmator) in. This is closer to what we saw even though it looks fake.

2017 madras HDR of inner corona with prom

Outer Corona

The Outer Corona presented more of a challenge. After some initial failures I finally got the entire image into PixInsight. That is the tool I use for my nighttime images. It took almost 2 1/2 weeks after the eclipse before I finally hit on the right recipe to get a good result.

After about two weeks of false starts I finally hit on the way to bring out the outer corona and preserve most of the inner. Above is a slight modification of the former HDR3 image I posted on 9/8. I felt the colors were too saturated in the HDR3 so I used Luminance (detail) from that image and combined it with the Chrominance (colors) from the result of applying the ln equation as described below. The result is closer to my memory of the actual corona color.

The moon and prominences are from an a different version chain so it is not "Photoshopped".

Note the triple helmet streamer at 12 o'clock.

Showing the moon is an aesthetic decision. I did versions with a prominent moon. In the end I decided that providing a little moon was better.

This HDR created by Photomatix highlights the structure in the outer corona even though it is not itself an aesthetic picture.

Compare these to the prediction by Predictive Science, Inc

Eclipse from Space

I always enjoy adding a satellite picture to give the eclipse context. In this case this is a portion of the GOES-16 Weather Satellite.

How Did This One Compare?

Every Eclipse is unique. Still one is left with an impression afterwards of what was special about this one. So here it goes

The eclipse was more than 40º above the horizon. In past experience that diminishes the effect of the approaching shadow. Compare the shadow in this movie verses 2015 or 2008 where the sun was lower in the sky. It was more like the more diffuse shadow of 2016. Smoke also played a role in making the shadow less distinct, but it could have added to the effects of the sunset
The prominences were mostly underwhelming. The large one at 12 o'clock being the exception. The eclipse in 2006 was also in a period of low activity but had better prominences. Four years later in 2010 there was only one significant prominence as this year. I cannot compare 2008 and 2009 since I did not do magnified images either time. Contrast this to 2013 where they ringed the sun.
Best for last. We come to eclipses to see the corona. This was the best yet! I thought the 2016 corona was excellent due to its curving structures. The 2006 corona is now on my business cards. This year there were significant helmet streamers at 7 and 2 o'clock (visually) and another at midnight. Very distinct solar brushes as I commented in my training videos. This is the first eclipse I viewed through binoculars (before I used the camera viewfinder). The images above capture the detail that I saw.

The Trip

The trip this time was an easy flight from home. It started in Seattle where my nephew lives.

We then traveled to Portland.

From there it was the setup for the eclipse. Overnight in Bend. An early departure for Madras😴 . A 5 hour return to Bend 😩 .

We then returned to San Francisco via the coast.

Just a limo at the end of the trip instead of a 24 hour flight (and then a limo).

Technical Stuff

Warning - Propeller Head Stuff follows

As a scientist and engineer I feel it is important to document my process. The sections below can be safely skipped by the casual reader. On the other hand if you are interested in how I produced the images then read on.

Equipment and Process

My Optics were the same as I have used in previous eclipses. A TV-76 and a Canon 60Da. This combination produces a FOV of 160' x 106'.

For this eclipse I used an Orion Skyview Pro mount. This is much, much heavier than anything I had used before on a group trip. The tour company allowed extra baggage this time. I packed it in a specially built cardboard box for the plane flight to Seattle. The head rode in my 20" cube that has been a frequent companion.

I controlled my imaging with Solar Eclipse Maestro as I said I would in my movies. The laptop was synced with time right before I left Bend. My experience is that it was never more than 0.5 seconds off anyway.

Processing was done with Canon Digital Professional tools, Pixelmator, Photomatix Pro, and PixInsight. The movie was created with Final Cut Pro. I used a MacBook Air and Mac Pro 2013 for processing.

The HDR images were created by Photomatix and PixInsight directly reading the raw images. The pictures were first converted by the Canon tools and then touched up with PixInsight.

How I processed the HDR

In my movie Why is this Hard? I describe the problems of imaging an eclipse. The range of brightness in the image was well above anything I had deal with in night time images. In the unmodified linear HDR the prominence at 12 o'clock had a brightness value of 0.5 (in range from 0.0 to 1.0). The dimmer parts of the corona were values closer to .0001. The normal Pixinsight tools that compress dynamic ranges were just not able to produce an image that showed both. So I had to get creative. I produced the image with a couple of tricks

First instead of using HistogramTransformation I flattened the image with the equation

.1*(10+ln($T))

Why ln? I tried with a more conventional log₁₀ but that did not produce the correct range of final brightnesses. Natural log did.

The image had significant background that was hiding the features of the corona. I used MLT to extract the residual layer from a copy of the image after removing the first 7 layers of data. The result was a good model for the background. I then subtracted some of that from the image. The result left the structures in place with more contrast.
I added in the prominences from a linear (non-stretched) image

Everything you see is from the original image. No Photoshopping fake stuff.

On consumer displays you can only distinguish 256 levels of brightness.

This is what the image looked like after all of the images making up the HDR are combined. The model represents all of the brightness values as a 64 bit floating point number ( range 1.0 to 0.0). It can easily encode vast differences in brightness. As it processes the images it figures which image should contribute which pixel and how that pixel should be scaled so it has the correct brightness value relative to the brightest pixels.

The corona is there (mathematically), but the values are all small enough that they are all represented as near black on your display.

How difficult is it to display all of the values at once? If we try to just display the 64 bit brightnesses on the screen we are, in effect, multiplying by 256 and then displaying the result. Thus we can only resolve differences that are at least .004 in brightness or about a order of magnitude more than the smallest value. In fact, most of the corona is dim enough that on this scale it resolves to the smallest brightness values.

To show this I have coded brightness values in gray. Note the largest area would be less than 1 and thus would appear black. Most of the corona would be too dim to display. The inner circles are for counts of 1, 2, and 3.

2017 Solar Eclipse brightness values in first 3 linear pixels

Clearly one has to be more creative to see what is in the image.

2017 Solar Eclipse Madras Stretched using ln

click for a larger image

To display the data we have to re assign the brightness values to larger numbers so more of the 256 values are used.

I did this by applying the formula

.1*(10+ln(image))

to the image. Now both the brightest and the dimmest pixels are visible. I continued processing to produce Outer Corona above. Much of this was further reassignment of brightness levels to provide additional contrast for details within the corona.

Solar Eclipse Maestro Predications

I made a point of not looking at the SEM predictions before processing my images to prevent prejudicing my processing. Here is what SEM predicted.

	17:19:31
	17:21:40

Copyright

The 2017 Madras pages by Robert J. Hawley are licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 United States License. This permits the non commercial use of the material on this site, either in whole or in part, in other works provided that I am credited for the work and you apply this or similar license to the result. Some of the included works bear a similar copyright.

rjh 9/15/17