An audit was done of the historical climate elements in the computer database to determine the data availability and correctness. Historical data before April 1998 for 20 randomly selected weather observation stations in South Africa were used. The following elements were tested: temperature, rainfall, pressure, wind, sunshine, clouds, humidity, global radiation, diffuse radiation and upper air. Availability was determined from the data extraction procedure and correctness by comparing weather elements with each other and using pattern recognizing for the upper air data.
It was found that the averages for availability and correctness is 76.3 % and 98.8 % respectively for an average station with climate (at least temperature and rainfall) data for 60 years and rainfall data for 89.4 years.
The performance of different weather stations (Weather Office, 1st Order, 2nd Order, Very old and Rainfall) was also determined. Wilcoxon's sum of Ranks test was used to determine statistical significant differences between the AWS(s) and 1st / 2nd Order stations with regards to availability and correctness. No statistical significant difference could be found between them as well as between AWS(s) and 1st and 2nd Order stations grouped together.
A breakdown of data losses was also done. Data was lost due to the following: computer change, climate data not yet computerised and instrument failure or data capturing faults.
This auditing system can be used to monitor the status of the computer database and the recapturing of historical data.