Related links: ASTM Standard Practice: Overview of materials  | ASTM notes: Additional information... |  
Revised method compared to that of John Irwin.  |  Model Validation Kit

Kincaid data: outliers

This is the text of the file OUTLIERS.TXT which is part of the package ASTM_SUP.ZIP   -  H.R. Olesen's supplement to the ASTM-related programmes supplied by John Irwin.


Kincaid data: Outliers
Careful inspection of the Kincaid data set leads to the conclusion
that 5 concentration values are outliers and should be disregarded.
These values are far higher than any neighbouring values, both in time
and space.
The Kincaid data have been screened for outliers only in the arc with the
largest concentration during a given hour, so there may be other, less
significant, outliers.
In the data of the Model Validation Kit, the file SF6_KIN.DAT has been
corrected for the existence of the 5 outliers (details below).
SF6_KIN.DAT is a file which contains one record for each arc, in total 1284
arcs.
In the data distributed by John Irwin (file KININPUT.BAT, dated October 23,
1997 and distributed from February 1999), 4 of the 5 outliers have been
marked as NEGATIVE VALUES. This has the implication that John Irwin's
NEWFIT programme disregards them; it may however be tricky if you use your
own software to process the data.
Details on the outliers and their treatment in the Model Validation Kit:
YR MO DY HR  DIST
80  7 20  8  40.0  Outlier with a value of 582 ppt removed. New
                   arc maximum of 32 ppt (0.186 ug/m3)
80  7 20 10  10.0  Outlier with a value of 948 ppt removed. New
                   arc maximum of 36 ppt (0.207 ug/m3)
80  7 24  9   1.0 or 2.0  Arc containing an outlier of 93 ppt removed
                   from the data of the Model Validation Kit, while
                   the value has been left untouched in the current version
                   of KININPUT.BAK.
                   Note:
                   In the data of the Model Validation Kit, the value
                   of 93 ppt is considered to belong to the 1 km arc;
                   thus the 1 km arc has been removed from the data set.
                   On the other hand, in John Irwin's KININPUT.BAK data set
                   the value is found in the 2 km arc, but has not
                   been touched.
                   In reality, the monitor is 1.47 km from the source.
81  5 23 15   7.0  Outlier with a value of 99 ppt removed. New
                   arc maximum of 10 ppt (0.059 ug/m3)
81  5 29 12   3.0  Outlier with a value of 540 ppt removed. New
                   arc maximum of 89 ppt (0.514 ug/m3)
The following values may be questionable but have been kept unchanged in
the data sets:
81  5 22 10  10.0  A value of 208 ppt is NOT considered an outlier.
81  5 29 15  15.0  A value of 125 ppt is NOT considered an outlier

Related links: ASTM Standard Practice: Overview of materials  | ASTM notes: Additional information... |  
Revised method compared to that of John Irwin.  |  Model Validation Kit

Homepage of "Harmonisation..." initiative

This page is maintained by Helge Rørdam Olesen

Document date: March 19, 1999.

[Dep. Homepage] Department of Atmospheric Environment, NERI