# Why do certain times of day have more launches than others?

(The question Day versus night launches is similar but the answers there, that time of day is irrelevant, seem to conflict with my data which shows clear biases towards certain times)

I feel like most of the interesting NASA launches happen before I get up in the morning: I live out West and like to sleep in, and it seems that whenever I start seeing excitement on NASA's Twitter accounts or on NASA TV for an upcoming launch and I look up details, I find that the launch is going to happen before my alarm clock goes off. I was curious if this was true so I tried to make a graph of launch-local time:

Here, with data from JSR's launchlog and looking just at launches from Cape Canaveral and Kennedy Space Center, I partly confirm my hunch by seeing that a big chunk of launches happen in the morning with a significant drop-off at 9am Pacific time. However I also see that there's an even bigger component in their evening (yay for me!) and this leads me to ask why these two chunks of time (7am-noon and 6pm-9pm) seem to be the most popular times for launches. And a follow-on, why the evening hours seem to be the most popular with launches from 6-7pm happening at almost double the rate of morning launches and about four times as often as afternoon and night launches.

I won't pretend my data is accurate or complete, but it passes my sniff test: lots of launches in those morning hours, and fewer launches at night than in daylight. But I wouldn't be surprised if someone can get more accurate data that shows something different or finds a mistake in my methodology:

After downloading the 1.6MB (don't be put off by the "LARGE" warning) launchlog.txt, I ran these commands (they're pretty hacked-together so don't judge me too harshly, and among other things they rely on both launch sites having at least 1 launch per hour, which is part of why I didn't include Wallops launches in my graph) to get just the Canaveral & KSC launches, convert UTC to Eastern, and plot them by hour:

sites="CC KSC"; for launchsite in $sites; do echo trying$launchsite; cut -b 14-29,160-163 launchlog.txt | grep -E $launchsite[[:space:]]*$ | awk '{print $2,$3, $1,$4}' | while read utc; do TZ='America/New_York' date -d "$utc UTC" +%H; done | sort -n | uniq -c > site$launchsite.out; done; gnuplot -e "set terminal pngcairo; set xrange[-1:24]; set boxwidth 0.75; set style data histograms; set style fill solid; set style histogram rowstacked; set multiplot; set xlabel 'hour of day'; set ylabel '# of launches'; plot 'siteCC.out' using 1 , 'siteKSC.out' using 1" > launch_tod.png


There was a very good point made that this data goes back to the very earliest rocket launches, and launch operations have matured significantly over that time. I would have liked to include only more-recent launches, but there just aren't that many. Here is a harder-to-read graph that shows the top 7 launch sites from Jan 1st 2000 through the first launch of 2015: Baikonur (304 launches), Canaveral (148), French Guiana (109), Plesetsk (96), Vandenberg (63), Xichang (58), and Jiuquan (48). If I did everything right, times are for launchsite-local time (the extra launch at '0' is a CSG launch with a date but no hour). A couple sites appear to have some trends towards preferred times but I think there's just too little data to make any real conclusions:

data since 2000: launches for all launch sites and top-7 hourly numbers on pastebin

• I wonder if there's a correlation with local weather at the launch site. E.g. lower windspeeds in the evening. Jan 30 '15 at 21:36
• It could be that dawn and dusk offer the quietest wind conditions, although I don't know for sure. Another point: Very nice graph but could you label the axes? It took me a second to work out what you where plotting against which. Jan 30 '15 at 21:36
• looks like they prefer to do it over dinner together. Thats the obvious answer ;) Jan 30 '15 at 22:06
• Your weather-related premise doesn't ring true. I live ~40 miles from the Kennedy Space Center, and those early evening hours are when the almost-daily thunderstorms tend to roll through, particularly in the summer months. Jan 31 '15 at 2:20

I don't have the reputation to comment, but it seems like you're using data that stretches back to the late 50's. I'd say their concept of operations is so much different from now that you shouldn't be including that data in any discussion of modern-day practices. What does your plot look like if you limit yourself to the last decade or so?

I also suspect that orbital mechanics would play only a minor role in the larger statistics of when launches occur. Orbital mechanics definitely determine your launch azimuth and launch timing for trying to rendezvous with the ISS or to target some interplanetary orbit or a formation / constellation flight...but across a wide array of missions, I don't see how that timing would possibly correlate to the local time of day. Also, if you just want your satellite ground tracks to stay centered over your area of operations (which is the case for telecom satellites), you can do that launch at any time of day to make the orbit. I suspect that any deviations from a uniform time-of-launch distribution are much more mundane.

My personal intuition is that launches from the Cape or from Vandenberg should tend to be in the early morning or in the late evening. This is because very large chunks of the national airspace have to be closed off from aircraft traffic and ships-at-sea must also be cleared out of the downrange areas. Logistically speaking, this is a nightmare. If you launch during the middle of the day, that's when most of the aircraft are in the air and so you will cause a very large disruption to the national airspace. Without going into the specifics of proprietary data, the analyses that I've seen for historical launches to orbit from launch locations in Florida estimate that the aggregate cost to the airlines is around a few hundred thousand dollars per launch. The cost comes from having many aircraft being rerouted around the launch danger zone, which translates into paying for more fuel burned and having to compensate your flight crew for working longer hours during that longer flight time. There is a strong economic incentive from the airlines to have space vehicles operate at times that minimize the disruption to air traffic.

• When I have more time in a few days and if there's interest, I can upload a picture or two of the size of the no-fly zones aircraft have to avoid during historical launches. Mar 6 '15 at 20:31

How much do you want to learn about orbital dynamics, orbital plane changes, and synergistic launch directions?

Part 1: Launch

It takes a change in velocity (here on called delta V) of ~7.8 km/s to go from stationary to orbital speed. It takes another 1.5-2.0 km/s of delta V losses (gravitational and fluid dynamic drag) to actually achieve orbit.

The Earth's rotation can provide ~0.5 km/s for free to your velocity provided you launch directly East from the equator. From anywhere else it works as 0.5km/s * cos(latitude). But the key is to launch directly East. From Cape Kennedy, this gives you ~0.43 km/s in free velocity.

Part 2: Destination

Most launches require the satellite to get to a specific orbit. That isn't just height, that includes a specific plane relative to the Earth, Sun, Moon, other planets, or Space Station.

In the example of the Space Station, its orbit precesses about the Earth. Launching directly East from Cape Kennedy doesn't always point you at where the Space Station will be. So part of launching timing is picking a Space Station orbit that will allow for the most benefit from Earth's rotation.

ISS ground track from Wikipedia

Secondly in cases of a direct launch to the destination, the launcher must time their launch such that it will reach the correct location at the same time as the target (e.g. a Shuttle trying to reach the Space Station).

Part 3: Orbital Plane Changes

Are extremely costly in fuel. It is much more efficient to achieve the correct orbit from launch than to correct it after launch.

Part 4: Location, Location, Location

I haven't looked over the launch list but what I'm betting your initial bar chart is showing is the desirability of some orbits over others. The timing must be based upon the objects destination orbital inclination and timing in the day. This could suggest sun synchronous orbits (spy & weather satellites) but I doubt they can entirely explain the blip.

• I've got to prepare for my daughter's choir concert so I'll have to finish later. Mar 5 '15 at 22:54
• None of the things you've said appear to have an obvious link to the sun or time of day... Mar 6 '15 at 22:29
• I actually thought about that all the way through the concert. There are orbits requiring launches at specific inclinations (sun synchronous, geostationary, etc.). However, they'd need multiple different inclinations of orbits to get proper coverage for sun synchronous. Geostationary is always in plane. So this might influence the statistical skew but isn't likely to be the main cause. I think the discussion of weather is more likely to be accurate. Morning has less atmospheric energy than evening so fewer storms. Mar 7 '15 at 13:36
• I'll adjust my answer accordingly. Keep the original info but explain why it's likely to be a minor effect. Perhaps make another answer about weather and why that's more likely to be important. Mar 7 '15 at 13:37
• This "However, they'd need multiple different inclinations of orbits to get proper coverage for sun synchronous." should read "However, they'd need multiple different times of day to get proper coverage for sun synchronous." Mar 7 '15 at 13:41

For Florida, surrounded by water and with a reputation for electric storms, weather is a big factor. Early morning and evening is preferred for stable atmospheric conditions.

The following is surmise: Whatever the launch site, the economic factors mentionned in the other answers would favorize a night-time or early morning launch. Human factors such as the sleep cycle would tend to an early morning launch. Any launch emergency would be better dealt with in daytime.

There would be a compromise among these preferences which would be followed subject to compatiblity with the principal constraint of orbital criteria