Missing data and the accuracy of magnetic-observatory hour means
Analysis is made of the accuracy of magnetic-observatory hourly means constructed from definitive minute data having missing values (gaps). Bootstrap sampling from different data-gap distributions is used to estimate average errors on hourly means as a function of the number of missing data. Abs...
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2009-09-01
|
| Series: | Annales Geophysicae |
| Online Access: | https://www.ann-geophys.net/27/3601/2009/angeo-27-3601-2009.pdf |
| Summary: | Analysis is made of the accuracy of magnetic-observatory hourly means
constructed from definitive minute data having missing values (gaps).
Bootstrap sampling from different data-gap distributions is used to estimate
average errors on hourly means as a function of the number of missing data.
Absolute and relative error results are calculated for horizontal-intensity,
declination, and vertical-component data collected at high, medium, and low
magnetic latitudes. For 90% complete coverage (10% missing data), average
(RMS) absolute errors on hourly means are generally less than errors
permitted by Intermagnet for minute data. As a rule of thumb, the average
relative error for hourly means with 10% missing minute data is
approximately equal to 10% of the hourly standard deviation of the source
minute data. |
|---|---|
| ISSN: | 0992-7689 1432-0576 |