Station 535880 Problem Report

Issues with station 535880, Wutai Shan, China - but no influence on the gridbox time series found.

Station 535880 (38.950N, 113.517E, 2210m, WUTAI SHAN, China) has a rather large inhomogeneity towards the end of the record that is apparent in both the mean and the variance. It is clearest for temperature (Figure 1) but also apparent for specific humidity (Figure 2) which is based on dewpoint temperature (it is also clear in the dewpoint temperature):

Figure 1: Monthly mean anomalies and actual hourly values of temperature for station 535880 before homogenisation.

Figure 2: Monthly mean anomalies and actual hourly values of specific humidity for station 535880 before homogenisation.

The PHA algorithm was applied to this station as part of the HadISDHlandq.1.0.0.2012p processing and resulted in removing most of the mean change (Figure 3). However, it is clear that the change in variance remains (Figure 4) and could result in incorrect analyses being drawn from this station.

Figure 3: Annual means of specific humidity for station 535880 before (red) and after (blue) PHA homogenisation. All neighbour stations used during PHA are shown in grey prior to any homogenisation adjustments being applied.

Figure 4: Monthly mean anomalies of specific humidity for station 535880 after PHA homogenisation (top), adjustments applied (2nd panel) and uncertainty estimates (3rd to 6th panels).

On analysis of the gridbox time series both before and after homogenisation, there is no obvious influence from inclusion of this station. Therefore at present this station remains a part of the HadISDHlandq.1.0.0.2012p dataset. Users of individual stations should be aware of the problems with this station.

There are doubtless other stations included within HadISDHlandq.1.0.0.2012p that have issues. If you see anything of interest - please do comment on the blog, tweet to @metofficeHadOBS or email me at kate dot willett at metoffice dot gov dot uk.