Relative Humidity at Flight Altitudes. Resources and Contrail prediction.

Mick West

Staff member
Some "ground truth" here, the most current RAP 0h forecast, with my location:

Panorama view outside my window:

A mixture of contrails and natural high cirrus.
[INFORMATION=Useful RH and Water Vapor Links]

The use of often highly inaccurate radiosonde (weather balloon) relative humidity (RH) readings for contrail prediction is a common problem in the "chemtrail" conspiracy community. Exhaust contrails need RH level above around 60%, and temperatures below about -40°C/-40°F. Unfortunately many of the radiosondes simply don't work at that temperature (and are inaccurate at warmer temperatures). They are also very widely spaced (hundreds of miles apart) and are generally only used every 12 hours, during which time the local RH at altitude can vary as much as 80%, for the simple reason that the air at that altitude is moving at up to 100 miles per hour.

So I think it would really help the conversation if we could not only explain this, but give a variety of resources to allow people to more accurately determine the RH at flight altitudes (around 30,000 to 45,000 feet). This will help people understand why there are contrails one day, but not the next)

I'd like to use this thread initially for discussion of this, and eventually add a useful list of resources to this post for easy reference later.

An initial point of confusion is that in most of these resources, the altitude is measured in mb (millibars of pressure), and not feet of meters. (You will also see hPa, which is exactly the same unit as mb)) This actually makes sense for planes, as they fly (and report their altitude) as pressure altitude, and not the actual altitude (which they can now get from GPS). There's a full discussion of the relationship between mb and altitude here, but for out purposes there's generally four mb values that cover the range we'd use for exhaust contrails:

300 mb -> 30,000 feet
250 mb -> 34,000 feet
200 mb -> 38,600 feet
150 mb -> 44,300 feet

If you don't know the altitude of the plane you are observing, then ideally use the 250mb/34000feet observation. Some sites don't have this, so 300mb/30,000 feet is a good option - as well as being easy to remember.

Note this covers the normal altitudes of exhaust contrails - i.e. those formed by the addition of water to the atmosphere from the engine exhaust. It does not cover persistent aerodynamic contrails, which more typically occur below 25,000 feet (and almost always below 34,000 feet). For those we can add:

450 mb -> 20800 feet
400 mb -> 23500 feet
350 mb -> 26600 feet

Here I'll focus on the more common exhaust contrails, but the contrail observer should always be aware of the possibility that they are looking at aerodynamic contrails, as they can look quite similar from a distance.

GFS and RAP Model Forecast RH

All RH readings you get are some kind of forecast. Unless you are actually in the plane, directly sampling the air outside, there is really no way of knowing exactly what the RH level is. Somewhat circularly the best indication of current upper air high RH is the presence of contrails. The radiosonde data is theoretically a good measurement, but as noted it's highly inaccurate, and highly limited in terms of time and location.

So what we use instead are very short range weather forecasts. Weather forecasts are computer models that take an incredibly large number of readings of many different types (ground readings, satellite readings, radiosondes, and aircraft readings), and then simulate the weather of the entire planet in order to predict what the weather will be in the future. Long range forecasts of several days are only roughly accurate, however the forecasts can be made for just a few hours, or even minutes, in the future. A forecast model can also give a "zero hour" (0h) forecast, which is what it calculates the weather to be right now (although typically it only creates these every few hours).

The most common one I've been using is the GFS model. GFS is the Global Forecast System run by the United States National Weather Service (NWS). Being government run the data from the GFS is free, and hence lots of different sites use it to create weather forecasts for different purposes. While not the most accurate for long term forecasting, it's still a reasonably accurate indicator of current conditions. And by looking at previous 0h forecasts we can use it as a reasonably accurate historical record of past conditions.

So what WE want is the Relative Humidity and temperature at flight altitudes, the simplest way of getting this is to use one of several RH maps for a given pressure altitude, and a good one to pick in the absence of altitude information is the 250mb/34,000 feet map. If that's not available, then 300mb/30,000 feet is good.

The following is basically several different ways of getting the same GFS data

Instantweathermaps is good, and has a clear color scheme where green is generally good contrail conditions. But limited to 300mb.


Europe version: is imited to CONUS (Conterminous United States - i.e. the lower 48 states) but gives a slightly higher resolution image, overlaid with wind arrows. They also have better time resolution than InstantWeatherMaps. You need to
By default it uses the RAP forecast. RAP is NOAA's North American specific model. For 0h forecasts it should be exactly the same as GFS. However it's updated hourly, so is more accurate for US forecast. It also has a wider range of altitudes on

Here I picked 300 mb to show it's the same as the IWM forecast, above, even though it's using RAP instead of GFS. Because it's a 0h forecast, there's almost no difference between the models.

Here's the same but at 250mb, 34,000 feet:

And the lowest pressure / highest altitude is 200mb/38,600feet.

Note that while there are certainly large differences between the 300mb and 200mb models, the broad strokes are similar. It's typically somewhat drier higher up, but not always. For example, look at the thin brown (dry) area over Northern California, it's green (humid) in the 200mb model

A bigger (but less detailed) picture can be had at, which displays the entire planet in 3D, and allows you to select RH at 250mb. You can also turn off the wind display by pressing the "play" (▷)button.

I've outlined the CONUS region so you can compare it to the previous charts. Here we've basically got blue meaning good contrail humidity conditions. While it's roughly similar, its not as high resolution as the other sites, and a little harder to read. It's a great visualization of the globe conditions though.

All this talk of "models" might make people suspicious. So a good things to check agains is the actual weather, as seen in satellite image. In particular the images taken at wavelengths that emphasize water vapor:

This just shows what the atmosphere looks like from above, so does not tell us much about particular altitudes. But it's a good general approximation a combination of all the levels.

The infrared satellite images are useful as they show areas were cloud is thick enough to obscure the sky, the grey streaky haze areas are more likely good contrail areas.

Now I've not focussed much on temperature here, largely because it's always cold enough at some flight altitude. However this does become a more important factor in the summer. Unfortunately TwisterData does not show high altitude temperatures. But InstantWeatherMaps does. Here's the 300mb map:

Cold enough in California everywhere North of Los Angeles. Now look at 250mb:

Cold enough everywhere in CONUS. And that's just 250mb/34,000 feet. Generally it's pretty safe to say it's cold enough at most flight altitudes above 30,000 feet, certainly above 35,000 feet.

The highest you get non-military planes flying is in the 150mb range, 45,000 feet. It's VERY cold there.

Because above a certain altitude it becomes very fuel-inefficient to fly, you will typically only get private jets at that altitude (as they are more concerned with getting there fast vs. cost). Sometimes you will see a solitary contrail in an otherwise blue sky, and it's generally a private jet above 40,000 feet.

A new way of looking at upper air humidity is to use the GOES-16 Satellite and select the Band-8 or Band 9 products. THen you can see just how quickly it can change over a few hours, or even minutes.
Last edited:
A mixture of contrails and natural high cirrus.

...and after a few days of dry weather and mostly contrail free skies, weather forecast is calling for a pattern change with raining moving into NorCal starting tomorrow and into the week...and lo and behold cirrus clouds and contrails are leading the charge.
Another source of RH plots, from the University of Oklahoma center for analysis and prediction of storms (CAPS) (Upper Aiir -> 250 mb RH). It's useful in that it overlays the temperature isotherms (lines of equal temperature).

Direct link to latest 300, 250 and 200mb:

Not the best color scheme for our purposes, as the 0.60 (60%) band isn't very distinct, and it only gives the most recent reading. You can however click on the "L" button to get a large animation of the map.
Earlier this year NASA updated their contrail forecast page from GFS to RAP, unfortunately they did not tell anyone or update the link on their web page. The new page is:

Which initially seems to be broken, as you need to select a jet engine efficiency factor (0.2, 0.3 to 0.4), and then it plays an 18 hour forecast. To get the 0h forecast you have to stop it and step back to zero

So above is the 13Z/0h forecast.

You can also link directly to the current images, here's the 0h forecast for 13Z at 0.3

The NASA forecast is showing regions suitable for contrail formation and persistence. It combines several altitudes into one chart, and so will result in a larger area than simply looking at one particular altitude.

It's a bit confusing though, as the color scale on the left contains more distinct pressure altitudes than are actually used, the chart only seems to show five altitudes: 125, 175, 250, 300 & 275

I'm not sure how well this chart tallies with the simple RAP RH map. I would presume it's more accurate, but it's unclear why some regions of high humidity are not showing up as contrail-friendly.
Last edited:
External Quote:

Unedited footage of closeups on sprayers over Chiloquin, OR, 11-4-16. I think it’s worth recording just how many planes are spraying, even if I often struggle to hold focus on them.

By the way, the relative humidity at 33,000ft in Medford (40 miles west) was just 40% half an hour before this was filmed.

Here's the GFS Forecast at 00:00 UTC on 11-05 (note it's the next day in UTC), for 250mb

His location is marked in red. You can see it's right on the edge of a very humid region 60% to 100%, perfect contrail conditions. Note also he's looking west, towards that region.

So how does he get: "relative humidity at 33,000ft in Medford (40 miles west) was just 40% half an hour before". He's using the Medford observations at the correct time, (00Z) and day (5 nov, the next day).
External Quote:

hPa m C C % g/kg deg knot K K K

307.0 9313 -40.5 -54.5 21 0.08 238 22 326.0 326.3 326.0
300.0 9470 -41.7 -55.7 20 0.07 240 19 326.5 326.8 326.5
298.0 9515 -42.1 -56.1 20 0.06 242 19 326.5 326.8 326.6
287.6 9754 -44.2 -55.3 28 0.07 250 19 326.8 327.1 326.8
277.0 10006 -46.5 -54.5 40 0.08 258 22 327.1 327.4 327.1
271.0 10152 -47.7 -57.7 31 0.06 263 24 327.4 327.6 327.4
262.4 10363 -49.6 -60.4 27 0.04 270 26 327.6 327.8 327.7
261.0 10400 -49.9 -60.9 26 0.04 269 26 327.7 327.9 327.7
250.0 10680 -52.7 -59.7 42 0.05 265 27 327.6 327.8 327.6
246.0 10783 -53.7 -59.7 48 0.05 265 28 327.6 327.8 327.6
234.0 11101 -56.7 -61.7 53 0.04 265 32 327.8 328.0 327.8
227.5 11278 -57.5 -65.1 37 0.03 265 34 329.3 329.4 329.3
222.0 11432 -58.1 -68.1 27 0.02 257 34 330.6 330.7 330.6
213.0 11691 -59.3 -69.3 26 0.02 245 33 332.7 332.7 332.7
200.0 12080 -62.9 -69.9 38 0.02 225 32 333.0 333.1 333.0
196.4 12192 -63.5 -70.0 41 0.02 225 34 333.8 333.8 333.8
193.0 12299 -64.1 -70.1 44 0.02 230 37 334.5 334.6 334.5
184.0 12591 -64.5 -70.5 43 0.02 245 42 338.4 338.5 338.4
182.0 12659 -60.7 -67.7 39 0.02 248 43 345.7 345.8 345.7
177.8 12802 -61.5 -71.5 26 0.01 255 45 346.6 346.7 346.6

And the RHw at 33,000 feet (10,000m) there IS correct.

So this seems to be a very good example of the problems of just using individual soundings. Here's the GFS forecast with the approximate positions of the sounding stations that UWYO uses.


Notice that in Medford the GFS forcast is indicating around 60-70%at 250mb. However to the west it's 100% and to the east it's 40%. The level at Medford tells you vey little.

In addition the 250mb sounding is not actually above Medford. The balloons move with the wind as they rise and can end up some considerable distance away. Here the winds would have carried the balloon to the east, towards the drier air.
At there are various downloadable programs to calculate contrail formation and persistence conditions from radiosonde sounding data. Unfortunately, they are in Italian, and only Windows executables are provided.
The program named "Contrails Soundings & Forecasting" can download the radiosonde data from the U Wyoming web site, apply corrections to the relative humidity ("Miloshevich correction") for the Vaisala RS92 sensor, taking the solar angle into account. The contrail factor can also be specified. Then it displays a table showing whether contrails can form and whether they persist at each altitude.

I think this could be a useful tool, but would be more useful in English and as a web app rather than a Windows executable.

There is also another program there that expects the user to input the altitude, pressure, contrail factor, RH, and optionally solar angle, and calculates corrected RH and whether contrails form and persist.

Literature references are provided on the calculation method:
Instant Weather Maps has a RAP model output, NAM, and GFS, for the same time period GFS seems to give much higher readings with . Here's the GFS first

Then NAM:


What might account for this? In the GFS output all of California is marked as >90% RH, but the RAP output has it mostly 60-70%

I suspect the RAP numbers are more accurate, but then NAM seems more detailed. All are 12-31-2017 12z 300mb RH%
Last edited by a moderator: