Quality control is an important aspect of any automated weather observation system. Whether it be via human or automatic system.

Here I’ve put together an automated quality control system for data read and saved to final storage to be viewed at a later date.

How it works: When data is read, it’s compared to a preset value. For the first part of this system, the value is compared to the low end reading of the sensor. For the second part of the system, the value is compared to the high end reading of the sensor. This preset number can be set at the discretion of the user. For our example, I am utilizing an R. M. Young 61302L barometric pressure sensor.

The reporting range of this sensor is 500-1100 hPa.

For our low preset, this will be 500 hPa. If the reading of the sensor is less than 500 hPa, the reading within intermediate storage will be sent to an unused INLOC (intermediate location.) If the reading is above 500 hPa, the reading within intermediate storage will be sent to a preliminary holding location for further processing.

Further processing will utilize the same logic. Only this time with 1100 hPa. This portion of the system will be our high end filter. To begin, the program will compare the current value with the maximum. If it reads below this preset it will save the value to final storage. If the value read is above the preset, it will be sent to the same unused INLOC where it’s discarded.

The system checks each time a reading is made. If the value is within the specified range, it will be allocated to final storage. If the value is not within the specific range, it will be allocated to the unused INLOC and discarded.

An example of this system implemented within Edlog can be found below. Here I’ve put together an auto QC checking the raw input of the barometric pressure sensor using the method described above:

The system implemented is completely automated negating any need for human intervention. If the sensor reports an erroneous reading, the QC system will only report the last valid measurement. One can also implement a flagging system to notify the individual of a sensor fault.