Memorandum #2

This paper discussed a new wind study about the winds Amelia Earhart would have experienced. Scroll down to see additional details.

Additional Details

Below you will find the above memorandum restated with additional details and charts.

Overview

A four-year study of actual winds during June and July at 10,000’ from Amelia Earhart’s last known position (0718z) to Howland Island, reveals the average headwind component to be just 5.1 knots and never more than 13 knots. Crosswind component for the route was never more than 4 knots from the left (north) and never more than 5 knots from the right (south). These findings should come as no surprise since this part of the world near the equator is, after all, the doldrums, where winds are notoriously calm.

Consequences

Light wind conditions undermine all three major Earhart disappearance hypotheses. 1) Blown off course to the south and accidentally found Gardner Island and arrived there in time to land on a reef before the tide covered it, 2) Ran out of fuel due to strong headwinds and crash-and-sank near Howland Island, and 3) Blown off course to the north and captured by the Japanese.

Likewise, derivative theories like the Taraia Object—which depend on a reef landing (#1 above) actually occurring—are now flawed and need another explanation for how the plane got into the lagoon.

Methodology

It is accepted that Earhart encountered strong headwinds early in her flight. This study examines the winds she faced during the final 11.9 hours—from her undisputed 0718z position report to Howland Island—where the competing theories diverge in their assumptions about wind effects

0718z Position

At 7.3 hours since takeoff (0718z) Earhart reported by radio her longitude (159.7) and latitude (4.33 south) and 23 knots of wind. This position is near several easily identifiable islands during daylight, so there is no reason to doubt it. And the time-distance-speed numbers to that position all work out almost perfectly.

That position is 772 nautical miles from takeoff at Lae. So 772/7.3 = 106 knots average groundspeed. 106+23=129 knots true airspeed, which is very close to the expected true airspeed of 130 knots (150 MPH).

Since this position appears reliable and consistent in every way, we will consider it as the jumping off point for the remainder of the flight, of which little is known. Therefore, this study concerns itself only with the remaining 11.9 hours of flight until at 1912z Earhart radioed “We must be on you but cannot see you.”, clearly indicating she thought herself to be near Howland Island.

Wind Data Source

This study uses wind data from the National Oceanic and Atmospheric Administration (NOAA) GFS model as distributed in GRIB2 format, the standard format for distributing operational weather data. GRIB files represent the fundamental source of gridded wind observations and forecasts used across atmospheric research and operational meteorology. This model is run four times per day and includes global gridded winds at 0.5 degree intervals, for many altitudes and with incremented forecasts for days into the future.

Data covering June and July of 2020, 2024, 2025, and 2026 were downloaded and archived by the author. The author began this study in 2024 but 2020 was also included because the author already had that data archived.

Various studies¹ indicate that tropical winds have been trending stronger, so we may assume that in 1937 the winds were probably not stronger than modern day data.

The study uses initialization data (analysis time) rather than forecast data, therefore it can be considered real-time data and highly accurate based on all available data sources such as satellites, etc.

Only data for sea level and 10,000' (700 hPa) were archived due to the large size of the original GRIB data files. 10,000' is presumed to be the altitude Earhart spent most of the flight at.

To account for daily variability, the average wind components were computed daily for 15 days +/- the date of Earhart’s flight, July 2, 1937. (June 15 to July 15)

Software

A software program was written that extracted GRIB wind data in the form of UGRD (east-west) and VGRD (north-south) wind components and applied these to the route from the 0718z position to Howland Island at intervals of 0.5 degrees of longitude to obtain the headwind component and crosswind component for each point. Then all points (47) were averaged to obtain the average headwind and crosswind for the overall route.

This results in 47 waypoints where local winds are obtained from the GRIB files and headwind and crosswind computed, as shown in the waypoints on this Google Earth screenshot.

Click here to download the above Google Earth placemarks as a KML file.

Wind components relative to the aircraft's true course are calculated by rotating the meteorological wind vector to align with the flight path. Given a true course θ and wind velocity components u (eastward/zonal) and v (northward/meridional), the components aligned with the flight track are:

Headwind component: u sin(θ) + v cos(θ)
Crosswind component: u cos(θ) − v sin(θ)

Positive headwind values represent tailwinds; negative values represent headwinds. Crosswind values are signed such that negative values indicate winds from the left (port) side and positive values indicate winds from the right (starboard) side of the flight course.

Headwinds

Below the graph shows the resulting headwinds for each year of the study. From left to right is June 15 to July 15, centered on Earhart's flight date of July 2. Headwinds were never more than 13 knots and on average for all years 5.1 knots.

Note that in 2026 headwinds were noticably less than other years and sometimes even a tailwind. During 2026 a strong El Niño condition was in progress, which probably explains this.

Crosswinds

Below the graph shows the resulting crosswinds for each year of the study.

Sea Level vs. 10,000'

Since the only known wind reports from the day of Earhart's flight were from the Itasca ship, comparing GRIB winds at sea level gives us something we can use to infer what Earhart was encountering at her cruise altitude of 10,000', where you would expect stronger wind. On average the wind at 10,000' was only 2.75 knots stronger headwind. So if the Itasca was reporting ground winds of 7-11 knots, we could infer that at 10,000' the wind was 9.75 - 13.75 knots. This is closely agrees with the results of the study of maximum headwind of 13 knots. The chart below is only for the year 2020 which was typical of other years.

Validation

Unfortinuately, due to the proprietary nature of some of the software modules used, the author cannot provide source code. Anyone wishing to duplicate this study on their own may download the GRIB files by clicking here.

A quick-and-dirty way to see actual real time winds is to go to the website Windy.com and examine the winds at sea level and 10,000' for real time conditions. If you are doing this in other than June or July, what you see may not be representative. Mostly November through April sometimes has atypical weather.

Below follows a discussion of Pilot Charts and actual weather observations on the day of Earhart's flight, which are consistent with these results and historical record.

Finally as already pointed out, as this portion of the flight progressed it was entering deeper into the equatorial doldrums which have been known for hundreds of years to have very light winds. If you stop and think about that alone, it might be all the validation necessary and a complicated study like this just confirms what we already knew.

Pilot Charts

Pilot charts are statistical maps of ocean wind patterns originally developed in the 1840s by U.S. Navy officer Matthew Fontaine Maury, who compiled centuries of observations from ship captains' logbooks. Organized by ocean region and month, these charts display prevailing winds through "wind roses"—visual representations showing wind direction and frequency. For nearly 150 years, pilot charts have continued to be updated and maintained by the National Geospatial-Intelligence Agency, ensuring they remain a current resource for modern navigation and route planning. For the purpose of this project, they are a good resource to validate the study results.

Here is one such chart for July in the south Pacific that includes the region of Earhart's last flight.

Here is an enlargement of the upper left corner, with the route from Lae to Howland Island shown in magenta.

Here is an enlargement of the wind rose nearest Howland Island, with annotations explaining the meaning of the wind rose presentation.

As you can see, it shows that the wind almost always comes from either the northeast, east or southeast. Most often from the east. The three barbs indicate force 3, which is shorthand for 7-10 knots. That correlates very closely with the findings of this study.

Here is the explaination of how wind rose's are constructed from the pilot chart itself.

By the way, Fred Noonan was previously a mariner and would have been familiar with pilot charts, and may have even carried them on the flight. Or at least consulted them in the planning stages.

Reported Winds

Itasca

Due to the remote part of the world this is in, there are few reported winds. The best would be from the Coast Guard ship Itasca stationed at Howland Island. Captain Thompson of the Itasca, in his formal report to his superiors, states the wind at Howland was "east 8 to 13 miles" (MPH) which converts to 7-11 knots. Exactly in the range found by this study.

Nauru

On the morning of Earhart's arrival near Howland, Nauru, an island 130 NM north of her midway point filed the following wind report using morse code;

"Nauru 8 a.m. upper air observation 2000 feet ninety degrees 14 mph 4000 feet ninety degrees 12 mph 7500 feet ninety degrees 24 mph."

This report has been used by some as evidence of strong headwinds and bears closer examination.

Given the report of 24 mph (21K) at 7500', this is an outlier that is inconsistent with this study in several ways.

1. The velocity of 24 MPH (21K), is much stronger than seen in any of the GRIB data.
2. It shows a very strong wind gradient from 4000' to 7000' which was never seen in GRIB data and would be highly unusual.

The author (who actually knows morse code as an Amateur Radio operator) thinks that maybe 24 mph is a morse code error and should be 14 mph. In morse code "1" is . - - - - (one dot, four dashes) and "2" is . . - - - (two dots, three dashes). These kinds of errors are common. In such case, it would make a lot more sense and change this from being an outlier to consistent with all other data.

You can verify yourself that these kinds of strong wind gradients are rare or nonexistent. Open the Windy.com website, click anywhere in the central Pacific Ocean, then examine the winds from sea level to say 10,000' using the altitude slider in the lower right screen. See how much winds change over ~ 3000' of altitude change.

Forecast Winds

Perl Harbor, which is over 2000 miles from the route of flight, was asked to provide a weather forecast and they did so, but clearly stated "Accurate forecast difficult account lack of reports your vicinity", and went on to state a wind forecast which was 20-25 knots from the east. Keep in mind, in 1937 weather forecasting was at a primative stage. The barometer was the main tool and useless from so far away. Often forecasters were not forecasting, but acting as a clearing agency for local weather observations, as infered by their disclaimer above. So in the authors opinion, this forecast is at best an educated guess and being at odds with the reliable observation by the Itasca, should be given little weight.