Real-time Water Vapour Maps from a GPS Surface Network: Construction, Validation and Applications

In this paper the construction of real-time integrated water vapour
(IWV) maps from a surface network of Global Positioning System
(GPS) receivers is presented. The IWV maps are constructed
using a two-dimensional variational technique with a persistence
background of 15 minutes old. The background error covariances are
determined using a novel two-step method, which is based on the
Hollingsworth-Lonnberg method. The quality of these maps is assessed
by comparison with radiosonde observations and IWV maps from a
numerical weather prediction (NWP) model. The analysed GPS
IWV maps have no bias against radiosonde observations and a small
bias against NWP. The standard deviation with radiosonde
observations is around 2 \kgm, while the standard deviation
increases with increasing forecast length (from 2 {\kgm} for the
NWP analysis to 4 {\kgm} for a forecast length of 48
hours). To illustrate the additional value of these real-time products for
nowcasting, two thunderstorm cases are discussed. The constructed
GPS IWV maps are combined with data from the weather radar, a
lightning detection network and surface wind observations. Both cases
show that the location of developing thunderstorms can be identified a
couple of hours prior to initiation from the convergence of moist air.