A flight report

In this article I will describe a recent navigation flight made in FSX, using classical navigation techniques from celestial observations, that means using sun, moon, stars and planets.

/features/oceanx
The afternoon before departure, the Cat awaits fully prepared.

Introduction

Ocean-crossing flights in aviation history have captured the imagination of the public. There were the early pioneers, the flying boats and the prize money. Names of Lindbergh, Alcock & Brown, Francis Chichester and Kingsford-Smith are etched in the memories of aviation fans. They are inspiring heroes of their time.

But many more didn't make it to fame or glory. In fact, many didn't make it to their last destination. They fell victim to failing engines, bad weather, or .... they lost their way: Amelia Earhart, Nungesser & Coli, and many, many others.

Indeed, one of the major challenges in the 1920's to 1940's was navigation. Without GPS, VOR/NDB, INS and other modern means, it was a true worry to find that tiny island out at sea, and before long a crew included a dedicated navigator, who often was the mathematics wizzkid with spectacles, turned down for a pilot.

Now, for us today, these navigators and flights are long gone. But you can recreate these flights in FSX with the proper aircraft, scenery and weather. However, two problems remain that prevent it from being attractive and realistic.

First, a long ocean-crossing flight in a slow aircraft in FSX is very, very boring. It takes a long time and it's not in any way challenging or difficult. And flying over water is visually unattractive in absence of photoreal terrain. Selecting Time Acceleration then becomes very tempting, but has little to do with flying.

Second, navigation at that time was accomplished by using a sextant to sight the sun, moon, stars and planets. Although complicated, inaccurate and hampered by weather, it generally worked when done properly and with clever techniques. But sextant navigation is not included in FSX. Although one or two gauges are available for download, they seem to focus on the visual aspect of the device, not on the navigation techniques required.

Solving the second problem, achieving sextant navigation in FSX, also nicely solves the first challenge. Because it's guaranteed to keep you very busy and excited. And indeed, it can be done.

In this article I will briefly explain how this is done, by telling you of a recent flight that I made, and that ended somewhat different than I expected.

I will also show that FSX needs a few expansions/add-ons in order to make this work, including a self-made piece of software for the heavenly bodies.

Once accomplished, the result has been very rewarding to me. It feels like I have re-discovered FSX as a valuable practice tool to learn a lost art. For obvious reasons (money, time, an aircraft, courage, a family) I am not able to make real-life ocean crossings as a pilot. But in this way, I have managed to learn this skill and execute it very, very realistically.

The 3D spherical-math of celestial navigation

A word of warning though. Navigation using the stars is complex. Don't expect to be doing the same within a few days. Yet I hope this ahis article inspires you. Basically, You'd need to learn the following in sequence.

1. Celestial observations for non-flying
2. Application to flying
3. Associated special navigation techniques

With skill, patience and dedication, you can do this. I'd be glad to help.

Enroute over the endless ocean, at dawn.

The Flight In Brief

The flight is from Norfolk Island to Lord Howe Island, both in the Tasman Sea between Australia and New Zealand a.k.a. The Ditch. Flight Range is around 480 nm and there is nothing but water in between. The Tasman Sea is also known for its storminess. Both islands are small and mountainous with one airfield each.

If I miss the island, I will bump into Australia. If I have enough fuel.

The aircraft for this flight is Aerosoft's Catalina. Direct flight duration is about 3.5 hours, which I planned to execute in two days with a Save Flight in between. This has some limitations, but can be done with some caution.

Lord Howe Island. How's that for a destination!?

Flight Planning

Timing is crucial. Flying from east to west, we fly the same direction as daylight, but slower. I chose February 2014 as the date (knowing the date is VERY important to celestial navigation), which means that it's autumn on the Southern Hemisphere and the days are shortening but still fairly long. Close to equator at 30 degrees South, it also means dawn and dusk pass quickly.

I chose to depart at very early morning, at 04:20 local time. This is important. It means I could start with observing the stars while darkness reigns. Later, towards my destination, I will observe daybreak and I am able to use the sun. It also helps to have daylight to spot the island and facilitate landing. However, I cannot realiably shoot the sun until it's well up. The twilight period in between is a problem, essentially making me ' blind' for navigation. The moon was low, and therefore also not very suitable for navigation, apart from being a 'difficult' body to shoot from an aircraft.

Cruise altitude is 10,000 feet. With a full load of fuel (although not really needed) this is quite high for a Catalina. But it gives the advantage of climbing over the clouds, otherwise I cannot see the stars through the clouds. And also it improves the visibility range to find Lord Howe Island.

For the route, I used a great-circle route to achieve the shortest connection, but I included a 40 nm deliberate offset towards the end. This is important. If you aim for the island directly, and it's not there when it should....it's on either your left or your right. But which one? By deliberately aiming off to the right, I was sure it's always on my left. This was a very common technique at that time.

Total distance is approximately 480 nm. Speeding along at 115 kts IAS at an altitude of 10,000 feet and at an estimated 10 degrees Celsius, this should give me 140 kts TAS. Flight time then is some three and a half hours.

Magnetic Variation is 15E for most part of the flight, 14E nearer Lord Howe.

The planned route, with pre-computed star sights between Waypoints until WP3. Between WP3 and WP4 to use the sun for the traditional Landfall procedure.

About Weather

Weather means everything. Without wind, ground speed and heading can be set perfectly, and there will be little diversion from the planned route and that would be very unrealistic. In reality, weather would vary significantly along the route across the ocean. FSX default weather does not provide well for over-ocean weather variations. A proper add-on weather module is required. I have used FS Global Real Weather with great satisfaction, as it specifically applies weather variations over the ocean. In reality over-ocean weather forecasting is very poor so I'll do without any forecasting and only use the surface winds at my departure.

A full moon lights the ocean from our 1 o'clock position.

About Time

Time is also worthy of your consideration. In celestial navigation Zulu time is used, more commonly known as GMT or UTC. It means that whereas my flight was on 14 February Zulu time in FSX, it already was 15 February at the departure and destination airfields. Using a wrong day may create a celestial navigation error of many, many miles.

FSX is also known to have a timekeeping bug, both in terms of accuracy and the use of time zones. But with date, the exact time is essential for accurate celestial navigation. A simple 1-minute error either in FSX or by myself in terms of navigation timing produces a 15 nm position error.

I used FS RealTime as a timekeeper add-on to keep FSX tighly synchronised.

Flight Execution

And so the time has come. Preparing this flight and the maps and the star-precomputations have cost me two evenings after job. I made extensive use of the Polhemus Celestial Navigation Sliderule, which is an object worth of studying in itself. I am very impatient to start!

The Polhemus CPU-47/P navigation computer for star shooting.

And then there is the rumble of the Twin Wasp engines coming up to revs and temperature. In the dark I taxi to the runway at Norfolk. Surface winds are 080/9, but I have no information of winds at altitude. 4/8 Cumulus clouds rise up, and I truly feel slight anxiety for this navigation flight. I am not very experienced, and I do want to find that island!

If I don't find it I might fly on to Australia, or land the Cat on the ocean. But that's not the point. I want this to nav thing to work.

On the runway, I wait. It's 16:10Z. Ten more minutes before take-off. Do not mess up the plan now...wait...

But at 16:18 I can wait no more. I tell myself that I will lose the 2 minutes on the initial climbout and turn.

I open the throttle levers, the engines spin up, and above 30" M.A.P. they roar and whine. The Catalina rolls and soon we're at 80 kts.

Rotate! And we're off. Up the gear, throttle down for climb.

I quickly head to my calculated heading, and set for 85 kts climb. The aircraft is heavy, the climb is slow.

After 25 minutes I reach 10,000 feet, and level off. Turn the autopilot heading gain to 3.0, power back and we're in the cruise. Away from Norfolk, away from safety, and into the dark void. Below me the ocean shines in the moonlight, but is mostly black. The dim cockpit light makes it difficult to read the compass easily, but I manage.

Ten minutes later already, it's time for the first series of star shots.

I planned to shoot the stars Rasalhague, Acrux and Regulus at 16:52, 16:56 and 17:00 respectively. Three stars gives a nice triangulation, pretty much as a fix from two NDB bearings. The sextant will run for 2 minutes per shot, so the first shot starts at 16:51, next at 16:55, and finally 16:59. This is in fact very realistic, and allows some time in between the shots.

I am skimming the cloud tops, with a clear view above. Ten seconds to shoot.

...8....6....5..... steady now....2.....1.....and I press the button on my joystick to simulate the sextant. This will run a .lua script through FSUIPC, and invisibly store my exact postion, date and time one minute after the joystick button.

Two minutes later, I execute a small batch file that calculates the true position of Rasalhague at my recorded time and position. The software will apply a random but carefully designed error, and gives me the result as if it were a real sextant shot. The calculated & corrected altitude (angle above horizontal) I determined for Rasalhague is 18°04', and the sextant reading gives 19°21'. Intercept is 43 nm AWAY.

Don't worry if you don't get this...

I do this for three stars, and after some more math what I get is a triangle on the map, and I am supposedly somewhere inside that. And I quickly plot this on the map. For a map, I glued two sheets of A3 millimetre-paper together. This allowed me to plot the full route and navigation fixes with some ease and good accuracy.

Next, I combine the traingle with my dead reckoning (DR): position estimate based on speed, course and stopwatch. At 17:00 hrs I had climbed 25 minutes at 85 kts IAS, and 17 minutes at 115 kts IAS. Without wind, I should be some 67 nm out from Norfolk island. I allow 10% error in range (both ways) and course (both ways).

Now the combined plot of DR and stars looks as follows.

The yellow line is my planned route, the red line my estimated route.

If I am in both areas, it means that I am in the overlap of the two areas. I place my new fix in the central position of that overlap, and work from there. This shows me nicely on track, although slightly south and a bit slower than expected. I quickly plot a new course to my next waypoint WP1, and turn accordingly.

I want to use the driftmeter to check and verify the wind. Obviously the surface wind of 080/9 is not blowing up here. But it's too dark to see the sea surface. Actually, the driftmeter in FSX is a bit like cheating. Using a driftmeter over the sea is in reality difficult because the waves come and go quickly. In FSX the sea is mostly a static texture from that altitude, making it very easy to establish drift within 1 degree. I made this a bit more realistic by limiting myself to a 5-second observation period. Switching to Prepar3D v2 might overcome this with Dynamic 3D waves.

But here I am in the dark. I expected to reach my waypoint WP1 at 1 hour into the flight, at 17:18, but there is no way to check that I am actually there. The air is smooth, and the heading autopilot keeps the flight steady. The Catalina is a stable aircraft on its own, and I can trim it nicely without the pitch autopilot.

Every 15 minutes I note my speed, course, the temperature. Somehow I picked up 1 or 2 knots of speed.

The EA-6B navigation device. Can't live without.

It's still very dark. But no time for boredom. The EA-6B flight computer is busy under my hands, and I am preparing the next star shots at 1800Z. This time it's Arcturus, Nunki and Suhail for shooting, and the time comes up quickly.

...3...2...1...Shooting....calculating.... plotting.

And 5 minutes after the last shot I have plotted the latest triangle. It's surprisingly small, indicating good accuracy. I also plotted my new DR position, taken from the last fix at 1700Z, with assumed compass heading and speed, and apply the same 10% margins.

Again I should be taking the overlap area, but I hesitate. The star fix looks good, and since I do not have a good wind estimtate, the DR position is questionable. I decide to discard the DR box, and fix my new position at 1800Z in the centre-of-gravity of the star triangle. This puts me north of my planned route, but not by much.

The fix at 1800Z. I didn't know is was to be my last and my source of trouble.

I plot for my next waypoint WP2 at 19:18, two hours into the flight. I have one more star shot coming up, but already it's getting light. The sun has been coming up on my 7 o'clock position. I can already tell that my next shot will not happen. And indeed, the stars can no longer be seen at 19:40Z. This is where the shortlived dawn at 30 degrees South shows itself.

Sun is coming up behind me, ruining my sight of the stars earlier than expected.

If only I had taken-off 20 minutes earlier. But then I would have run into trouble towards my destination with a very low sun.

By now I can use the driftmeter! It shows a wind drift angle of -14 degrees!

Wow....and uh-oh. This spells trouble. If the wind is so strong that it puts me off 14 degrees to the south, where I was correcting for only 4 degrees, how can I still have been on my planned route, let alone to the north of it ?!?

The wind may have come up or shifted only recently. I don't know, but I do know that I must alter course immediately. With two more hours of flying ahead without a star fix, I have no room for being sloppy.

I check the drift every 15 minutes and adjust course. Now -12, then -15 degrees. I have no idea about head- or tailwind component. There a methods for this, but I have left them at rest for this flight. One thing at a time...

Time is flying now, and I expect that I am coming up to my turnpoint at WP3 for the offset aim, which is also the start of my sunsights. With the sun straight behind, it is excellent positioned to determine how far I have progressed to Lord Howe island. But it is useless for determining how far I am north or south. Which is what I really need now...

Counting the minutes, and here we go! ....Turn to 272 degrees compass heading, and prepare the sextant shots on the sun.

Presently I run into much thicker and heavier cloud. The sea is often obscured from view. This makes it difficult to see find the island. Fortunately the sun is up above the clouds, and clearly in view.

First shot, sun is still low at 10 degrees above the horizon. Atmospheric refraction will make this an unreliable measurement. It puts me at 224 miles from the island. Can't be right, I am much closer for sure.

10 minutes pass. I keep her steady on course, while my mind races the mathematics and considers options.

Next shot.... at 2010Z. Better, at 68 miles, sun at bearing 98.5 degrees. Quick plot....

5 minutes later, next shot. Puts me at 3 miles overshot! No way, no way a Cat can travel 71 miles in 5 minutes. One or both must be wrong. Accuracy will improve when the sun gets higher, but I don't have so much time.

I prepare the next shot for 2020Z. I have pre-plotted the sun's altitude versus time, so I can quickly look it up without the hard work.

The green line is the 2010Z Sun Line. Lord Howe Island lower left.

I glance forward across the panel, staring. The engines drone as they have for three and a half hours. There is a clearing in the clouds. The sun from behind makes the ocean look great, tranquil.

But what's that, dead ahead. I can't believe it...an island with two steep rocks at the south end.... It's Lord Howe island. I can't believe it!

Good news. I found the island. Bad news, too. I expected to be elsewhere, still a left turn and 15 minutes away.

Lord Howe Island, dead ahead!

Anyway, I executed a quick descent, checked the windsock on the airfield and turned to land. The still-heavy Catalina does not have very strong brakes, and was difficult to stop on the 800m runway. But I was down, and I was in need of a virtual shower.

Epilogue

Something went very wrong here. Apparently I was far more south than I thought. If I had gone further south, I might have missed the island altogether. Where did I go wrong, when my star sights worked out so well?

Luckily I used yet another add-on to keep track of my position, and write it to a text file every ten seconds. It's called Flight Data Recorder. Here's my actual route, compared to my planned route....

Oops...

As you can see, I went off course right from the start, blown far south by a fairly strong WNW wind, which also caused me to be slow. I wasn't aware of the wind beforehand, but I had saved the weather by FS GRW to be reviewed later on. My offset at WP3 turn accidentally put me straight on course for the island, by an extraordinary stroke of luck!

What went wrong? Two things...

1 - I did not apply the compass deviation of the Catalina. The placard in the cockpit says...for 240, steer 243. I didn't, and so went off 3 degrees from the start. 3 degree over a range of 480 nm is approx 25 nm.

2 - I made errors in both star sight computations. For one star at the 17:00Z sight, and one at the 18:00Z sight. I subtracted two values that should have been added. If done properly, the triangle for the second fix would have and extended more south. As a consequence, I would have corrected.

These two errors added up, and did not cancel out. If they had cancelled out, I would never have discovered my mistake. A valuable lesson. Better next time.

The corrected star plot (pink) at 1800Z, much more southerly than the original one (green).

List of Add-ons used:

1. Self-built software for Celestial observations, in SciLab and Lua
2. FSUIPC
3. FS RealTime
4. OPUS FSI
5. FS Global Real Weather
6. Aerosoft Catalina
7. Flight Data Recorder

Eric van der Veen
ericvdveen@tiscali.nl