the ioos QA manuals are a trove of recommendations for a number of quality checks on in-situ data. For example, the wind manual contains the following recommended tests: (pg 12):
Timing/Gap Test
Syntax Test
Location Test
Gross Range Test
Climatology Test
Spike Test
Rate of Change Test
Flat Line Test
Multi-Variate Test
Attenuated Signal Test
Neighbor Test
That they describe and organize into whether the test is required, strongly recommended, or suggested. They also give codable instructions, eg : If |WSn-1 - SPK_REF| > THRSHLD_HIGH, flag = 4 for the spike test.
The challenge though, is that for most of the tests there are operator specified variables. In the above example, the THRSHLD_HIGH is operator specified and might be site-specific.
We need to consider how/whether we can implement these tests given that many (most?) include some variables like this which may require too much domain or site specific knowledge for us to implement with confidence.
Thoughts @mbjones?
the ioos QA manuals are a trove of recommendations for a number of quality checks on in-situ data. For example, the wind manual contains the following recommended tests: (pg 12):
Timing/Gap Test
Syntax Test
Location Test
Gross Range Test
Climatology Test
Spike Test
Rate of Change Test
Flat Line Test
Multi-Variate Test
Attenuated Signal Test
Neighbor Test
That they describe and organize into whether the test is required, strongly recommended, or suggested. They also give codable instructions, eg :
If |WSn-1 - SPK_REF| > THRSHLD_HIGH, flag = 4for the spike test.The challenge though, is that for most of the tests there are operator specified variables. In the above example, the
THRSHLD_HIGHis operator specified and might be site-specific.We need to consider how/whether we can implement these tests given that many (most?) include some variables like this which may require too much domain or site specific knowledge for us to implement with confidence.
Thoughts @mbjones?