Quality flags are constructed on a philosophy of two qualifiers. The first describes the origin of the data and the second the quality.

Possible origins of data are:

+ Original: This entails the data value is the original value. It has not been amended by DELFT-FEWS
+ Completed: This entails the original value was missing and was replaced by a non-missing value.
+ Corrected: This entails the original value was replaced with another non-missing value.

Possible qualifiers are:

+ Reliable: Data is reliable and valid
+ Doubtful: The validity of the data value is uncertain
+ Unreliable: The data value is unreliable and cannot be used.

Following this specification, the table below gives an overview of quality flag enumerations

Table 1                   Enumeration of quality flags

Enumeration

Description

0

Original/Reliable
The data value is the original value retrieved from an external source and it successfully passes all validation criteria set.

1

Corrected/Reliable
The original value was removed and corrected. Correction may be through interpolation or manual editing.

2

Completed/Reliable
Original value was missing. Value has been filled in through interpolation, transformation (e.g. stage discharge) or a model.

3

Original/Doubtful
Observed value retrieved from external data source. Value is valid, but marked as suspect due to soft validation limits being exceeded.

4

Corrected/Doubtful
The original value was removed and corrected. However, the corrected value is doubtful due to validation limits.

5

Completed/Doubtful
Original value was missing. Value has been filled in as above, but resulting value is doubtful due to limits in transformation/interpolation or input value used for transformation being doubtful.

6

Original/Unreliable
Observed value retrieved from external data source. Value is invalid due to validation limits set. Value is marked as unreliable and will not be used in transformations and model calculations.

7

Corrected/Unreliable
The original value was removed and corrected. However, corrected value is unreliable

8

Completed/Unreliable
Original value was missing. Value has been filled in as above, but resulting value is unreliable,

9

Missing value in originally observed series. Note this is a special form of  both Original/Unreliable and Original/Reliable.

Notes:

  • No difference is made between historic and forecast data. This is not considered a quality flag. The data model of Delft-FEWS is constructed such that this difference is inherent to the time series type definition.
  • No labels

Disclaimer