Home

Data Quality Control of in-situ observations. TITAN has been tested or it is currently under test for:

variable name	abbreviation	command-line argument	comment
air temperature	T	--variable T	default
total precipitation amount	RR	--variable RR
relative humidity	RH	--variable RR
surface snow thickness	SD	--variable SD

Several tests are applied sequentially to detect those observations that are most likely affected by: gross measurement errors, large systematic errors and large representativeness errors.

• An observation affected by gross measurement error does not contain any information on the true atmospheric state.
• An observation affected by large systematic error contains inaccurate information about the true atmospheric state, such that it cannot be trusted as a correct observation.
• An observation is said to have a large representativeness error if its value cannot be accurately predicted by using the neighbouring observations. The observation is significantly influenced by local atmospheric conditions, where local meant having spatial scales smaller than the average distance between the observation and its neighbours.

Each test gets its own data quality code, which is assigned to the "suspect" observations.

The strategy we follow is to divide the observations depending on the observation providers. The TITAN tests can use different thresholds and parameters that are also specified as a function of the observation provider. As a result, the user is free to use the same TITAN tests with different settings depending on the observation provider.

Command line

$>./titan.R input_file output_file [options]

The detailed description of the input/output is reported here