Kaleidoscope

Dale Lehman writes:

This one’s on a topic you have blogged about often and one that I still think is under-appreciated: measurement. The Economist recently reported on this fascinating article about lightning strikes and their apparent sensitivity to shipping lanes and the associated pollution.

I [Lehman] immediately wondered about whether there is a bias in the measurement of lightning – are they more frequently measured in shipping lanes? Not knowing anything about how lightning strikes are measured, I went to the data source.

According to the article:

WWLLN detects the majority of all lightning-producing storms, even in regions with no stations within 2000 km

and

Because of this trend in the detection efficiency, we compare lightning statistics over the shipping lanes with those for adjacent regions having comparable WWLLN coverage.

The network site says

Typically only about 15 to 30% of strokes detected by one sensor are detected by 5 or more. These strokes are usually the stronger ones. Recent research indicates our detection efficiency for strokes about 30 kA is approximately 30% globally.

It makes me wonder whether there is a systematic difference in the ability to measure lightning strikes within major shipping lanes and outside them. Since I have absolutely no background in this subject, I may well be completely off-base. On the other hand, it would seem that with measurement accuracy so low, that there could well be measurement biases in this data. Somebody (actually many somebodies) undoubtedly knows more than I do about this. I thought you might be able to reach a wide enough audience to find out.

I don’t know either. But maybe some of you could shed some light on the topic?

P.S. Above foto courtesy of Diana Senechal.

5 thoughts on “Kaleidoscope

  1. From the data source link that was provided, it seems like the placement of sensors and shipping lanes were independent – so no bias introduced there. However, there may be bias by chance based on the proximity or density of sensors towards shipping lanes or outside areas, which could be easily checked and fixed during analysis stage. This question seems to inquire about the sensors’ behaving differently in shipping lanes than they do outside, which is a technical issue and the engineers/scientist in the field might know.

  2. This one is easy – like homework for multivariable calculus or electricity and magnetism, you just ask your Dad:

    “I know this paper, and there is little chance of any spatially-relevant bias in the data because the sensors are very far away, and any geometrical aberrations due to specific sensor locations would not “curve” the way the shipping lanes and lightning trends do. Also, I have seen these same data using a more sensitive network, and the pattern is there (but faint). We are playing with the possibility that the “pollution” has a bigger impact on the high-current lightning discharges…”

  3. Most likely sensors. As I understand it these detectors are serious business, and businesses use them for critical purposes, i.e. either processes that are sensitive, or for insurance purposes. They pay to have access to this data! I’ve heard a person working for NZ MetService say they put a lightning strike within 50ms on their site, and customers complain if it isn’t there in 1 second or so.

  4. There are a bunch of local systems run for commercial purposes: more accurate and sensitive than WWLLN, which is a global network that makes its data available. You could look at the cumulative distribution of distances to sensors
    for points in shipping lanes vs points nearby but not in shipping lanes. I’ll bet there isn’t much of a difference.

Leave a Reply to Ayse Tezcan Cancel reply

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