IPCC Site
"
The CRU Climate Baseline Quality Issues
Known problems with the CRU 0.5° latitude/longitude 1961-90 terrestrial climatology and 1901-95 grids of monthly surface climate
Problems have come to light through user feedback and routine quality control in the Climatic Research Unit (during work towards the construction of an updated version of the 1961-1990 climatology). These problems range from systematic errors (for example, due to incorrect units being assumed for a whole country) to single-station errors. Errors at single stations have varying importance depending on the density of the surrounding network. Where the network is sparse, the station error will influence a larger area because of its greater spatial influence during the interpolation.
Systematic errors arise because of varying reporting methods and units adopted by different national meteorological agencies (and other data contributors). The methods used to standardise data to a common format are discussed by New et al. (1999). Some problems are due to shortcomings in the standardisation methods used. In addition, misunderstandings (or a lack of information) about the units in which data were supplied means that the incorrect (or no) standardisation was used.
The majority of problems/errors identified to date are due to confusion of units and shortcomings in the standardisation towards the use of common units. The climate variable with the largest potential for confusion is wind speed. Not only do reporting units vary between metres per second, miles per hour and knots; anemometer heights can vary greatly from the desired 10m (e.g. between 2m and 20m).
The following (variable) gridded files are known to have either systematic or significant single station errors which potential users need to be aware of. The errors are present in the mean-monthly 1961-90 terrestrial climatology files and are thus propagated into the gridded 1901-1995 monthly time-series. The affected part of the globe is defined where possible.
| Wet days | |
| Area affected | Reason for error |
| Brazil (Amazonia) | Conversion was required from available data (the threshold was 1.0mm). The method used probably gave a positive bias and thus overestimated the number of wet days. |
| Spain and Spanish stations in N. Africa | Error over definition for approximately 40 stations. Threshold was assumed to be 0.1mm but was in fact 1.0mm. |
| Syria | Error over definition for all stations. Threshold was assumed to be 0.1mm but was in fact 1.0mm. |
| Diurnal temperature range | |
| Area affected | Reason for error |
| Greenland *(see below) | Lack of station data in central Greenland has caused (too) high values to be interpolated to the region. |
| Poland | Some stations were found to have their range-values based on monthly extreme max. and min. temperature values instead of average values - therefore values too large. |
| Wind speed | |
| Area affected | Reason for error |
| Bolivia | Values for several stations found to be in knots – therefore too high |
| Greece | Values for six stations found to be in knots – therefore too high. |
| Honduras | Values for all stations were found to be in knots – therefore too high. |
| Sierra Leone | Values for all stations found to be in knots – therefore too high. |
| Sudan | Values for all stations originally in miles per hour but incorrect conversion to m/s produced values too low |
| Peru (Iquitos & Cajamarca) | Values for both stations found to be in error (much higher than is feasible in the area). |
Relative humidity
Station data relating to humidity was split roughly half-and-half between pressure (VP) or relative humidity (RH). The conversion of VP to RH, or vice versa, (see New et al.) does pose problems in some parts of the world – notably in the coldest areas in winter months. This is due to a loss of instrumental precision at very low temperatures. Small errors are magnified when conversion (using mean temperature) is undertaken. For this reason, systematic differences in winter RH are apparent according to political divisions in areas like the northern Russia and northern Canada.
In addition, mean monthly RH is affected by the timing of daily readings. For this reason, mean RH may be biased if mean monthly values are not based on true daily mean values, or if the time of measurement of daily temperatures and RH do not coincide.
* Greenland – all variables
The interior of Greenland is poorly covered by meteorological observation. This coupled with the presence of the high elevation ice cap makes interpolation of climate normals very difficult due to the potential for unusual lapse rates. It is likely that significant bias may be present with all variables for the interior of the landmass (e.g. diurnal temperature range and precipitation too high).
References
New, M., Hulme, M. and Jones, P.D., 1999: Representing twentieth century space-time climate variability. Part 1: development of a 1961-90 mean monthly terrestrial climatology. Journal of Climate 12, 829-856.
New, M. G., M. Hulme and P. D. Jones, 2000: Representing 20th century space-time climate variability. II: Development of 1901-1996 monthly terrestrial climate fields. J. Climate, 13, 2217-2238.