‘The Eclipse wind’: real or imagined?

Meteosat satellite image showing the moon's lunar shadow on 11 August 1999. Copyright © 2001 EUMETSAT

I’m no doubt showing my age here but I have very clear memories of standing out on the Department’s external testing area (a.k.a. the coffee balcony) and observing, with many others, the 97% partial solar eclipse over Reading on the 11th August 1999.

The eclipse path. From Espenak and Anderson (1997).

Eyes are good at adapting to diminished light so the partial eclipse we experienced in Reading didn’t seem so very dark, but there was a strange feeling in the air and, even as a scientist, I had a slight niggling worry that perhaps this time the sun wouldn’t re-appear from behind the moon. Fortunately it did though and it turned out that we were actually more fortunate in our eclipse viewing in Reading than were the many people who went down to Cornwall to experience the total eclipse there. Cornwall was cloudy due to an occluded front close to the west of Ireland whereas we only had partial cloud over southeast England. This post is not about how weather can interfere with our eclipse viewing though, instead it’s about how eclipses can interfere with our weather.

Measurements of insolation and 1m screen temperature at Reading (5 minute averages) .

We’ve known for a long time that an eclipse causes a reduction in solar insolation and an associated reduction in surface temperature. Winds also tend to slacken and reduce in variability during an eclipse due to the stabilization of the boundary layer and associated reduction in turbulent motions. Karen Aplin and Giles Harrison made some measurements during the 1999 eclipse at both Camborne in Cornwall and at our Atmospheric Observatory here in Reading. Here they observed a maximum reduction of temperature of about 2C occurring about 15 mins after the time of maximum eclipse and the expected pronounced drop in windspeed and reduction in variability. They also found some evidence of a more controversial impact of an eclipse on weather, wind direction changes at both Reading and Camborne at the onset and end of the eclipse.

Composite of wind and temperature measurements from the 28 May 1900 eclipse over the US obtained by combining the results for stations similarly situated in regard to the path of the eclipse. The dashed lines enclose radii of about 1500 and 2500 miles where 2500 miles is the edge of the penumbra. From Clayton (1901).

To understand why these wind direction changes are controversial we need to study some history. Back in 1901 H. Helm Clayton postulated the existence of a cold-cored eclipse cyclone based on his analysis of measurements of the US eclipse of 28 May 1900 made in several locations. This eclipse cyclone is a cold outflow of air from the umbra (the region totally obscured by an eclipse) leading to negative surface pressure anomaly with outer ring of positive pressure anomaly. This leads to an anticyclonic circulation associated with the cold core extending out to a distance of about 1500 miles. Whilst a review of Clayton’s work in Science at the time proclaimed ‘Clayton has gone far ahead of all previous investigations of the phenomena of eclipse meteorology…’, Frank Bigelow wrote a rather damming letter to the same journal refuting Clayton’s ideas. This controversy still exists today with Founda et al. (2007) declaring that the alleged ‘eclipse wind’ associated with the eclipse cyclone is no doubt rather an enhanced wind chill effect and Anderson (1999) writing ‘Subjective impressions in the highly emotional moments leading up to and through totality are also likely to encourage the persistence of this story….’. There’s nothing better than a century of debate to pique the interest of meteorologists so Giles and I decided to see if the latest high resolution Met Office weather forecast model could help us solve the mystery of the eclipse wind.

We’re very fortunate in the UK to have a high density of meteorological stations reporting hourly (synoptic) weather measurements. The difficulty in using these measurements to determine the effect of an eclipse is that, of course, weather is always evolving and we don’t have measurements from the exact same time ….. but without the eclipse. In the past researchers have got around this problem by interpolating from measurements before and after the eclipse or even comparing against measurements from a different (but somehow equivalent) day. We decided a new way was needed – we compared our synoptic weather measurements against a high resolution (1.5 km gridspacing in the horizontal) simulation of the Met Office weather forecast model. Now, a thorough search of the Met Office model code revealed a distinct lack of an ‘eclipse parameterization scheme’ so our forecast was completely ignorant of this spectacular phenomenon. Hence, we could use the model to tell us what would have happened without the eclipse.

Temporal evolution of MIDAS surface meteorological observations (left column, plotted at the observation hour) and UKV model output interpolated to station locations at times when MIDAS surface station observations are available (right column) for (a-b) temperature, (c-d) wind speed, and (d-e) wind direction. Thin lines are for individual stations with, for a given variable, a given colour corresponding to data from the same station. Thick black line is the mean value with error bars given by ±1.96 standard error (yielding the 95% confidence interval). From Gray and Harrison (2012).

So, what did we find? We concentrated our analysis on an inland, relatively cloud-free region, including Reading and found differences occurred between the evolution of the station measurements and forecast predictions at the same stations after the onset of the eclipse. The winds decreased by an average of 0.7 ms-1 and turned anticlockwise by an average of 17 degrees in the station measurements but not in the forecasts. Is this finally conclusive proof of the eclipse wind?

Eclipse lovers might like to know that the next total Eclipse over the UK will be on 23 September 2090. It’s unlikely I’ll be running the Met Office model then but maybe the partial eclipse of 20 March 2015 will provide the opportunity to further explore the existence of the ‘eclipse wind’.

1 thought on “‘The Eclipse wind’: real or imagined?

  1. More eclipse modelling and measurements in Meteorological Zeitschrift:

    Here are some slides on an eclipse in 2006 (5MB .wmv file which should work in Windows) including photos from Turkey in 2006 by Stu Webster:

    I guess the eclipse must make the atmosphere more stable and reduce cumulus (see before and after) although they could have just blown away!

Leave a Reply

Your email address will not be published. Required fields are marked *