Labeling data as having an error?

I am curating a large quantity of data from different sensors. If I know that a particular sensor was broken or poorly calibrated for a particular time range, what would be a useful way of annotating the data to make it clear that the data are of poor quality and / or have known errors?

I am thinking a set of key:value pairs (like quality:error, description:'sensor was broken') that I can store in json, yaml, image header (e.g. exif) metadata or similar. If there are any standards or conventions for this type of annotation I would like to adopt them.