Quantifying temporal observations uncertainty for forecast verification

This work aims to quantify the sub-hourly variability of synoptic observations such as temperature, cloud, visibility, wind, using 1-minute observations. How temporally representative are hourly synoptic observations?

A range of sites around the United Kingdom with varying geographical locations (coastal, inland, upland) were selected to analyse sub-hourly variability as a function of time of day, on daily, monthly and seasonal time scales. We were also interested in establishing to what extent the variability between parameters is linked in a conditional way.

The aim is to understand to what extent results from these different locations can be generalised in a quasi-climatological sense. The application would be to "dress" hourly synoptic observations with an estimated total observations error for the verification of both deterministic and ensemble km-scale NWP forecasts.