 |
| Titel |
Real-time adjusting of rainfall estimates from commercial microwave links |
| VerfasserIn |
Martin Fencl, Michal Dohnal, Vojtěch Bareš |
| Konferenz |
EGU General Assembly 2017
|
| Medientyp |
Artikel
|
| Sprache |
en
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 19 (2017) |
| Datensatznummer |
250148268
|
| Publikation (Nr.) |
 EGU/EGU2017-12509.pdf |
|
|
|
|
|
| Zusammenfassung |
| Urban stormwater predictions require reliable rainfall information with space-time
resolution higher than commonly provided by standard rainfall monitoring networks of
national weather services. Rainfall data from commercial microwave links (CMLs)
could fill this gap. CMLs are line-of-sight radio connections widely used by cellular
operators which operate at millimeter bands, where radio waves are attenuated by
raindrops. Attenuation data of each single CML in the cellular network can be
remotely accessed in (near) real-time with virtually arbitrary sampling frequency and
convert to rainfall intensity. Unfortunately, rainfall estimates from CMLs can be
substantially biased. Fencl et al., (2017), therefore, proposed adjusting method which
enables to correct for this bias. They used rain gauge (RG) data from existing rainfall
monitoring networks, which would have otherwise insufficient spatial and temporal
resolution for urban rainfall monitoring when used alone without CMLs. In this
investigation, we further develop the method to improve its performance in a real-time
setting.
First, a shortcoming of the original algorithm which delivers unreliable results at the
beginning of a rainfall event is overcome by introducing model parameter prior distributions
estimated from previous parameter realizations. Second, weights reflecting variance between
RGs are introduced into cost function, which is minimized when optimizing model
parameters. Finally, RG data used for adjusting are preprocessed by moving average filter.
The performance of improved adjusting method is evaluated on four short CMLs (path length
< 2 km) located in the small urban catchment (2.3 km2) in Prague-Letnany (CZ). The
adjusted CMLs are compared to reference rainfall calculated from six RGs in the
catchment.
The suggested improvements of the method lead on average to 10% higher Nash-Sutcliffe
efficiency coefficient (median value 0.85) for CML adjustment to hourly RG data. Reliability
of CML rainfall estimates is especially improved at the beginning of rainfall events and
during strong convective rainfalls, whereas performance during longer frontal rainfalls is
almost unchanged. Our results clearly demonstrate that adjusting of CMLs to existing RGs
represents a viable approach with great potential for real-time applications in stormwater
management.
This work was supported by the project of Czech Science Foundation (GACR)
No.17-16389S.
References:
Fencl, M., Dohnal, M., Rieckermann, J. and Bareš, V.: Gauge-Adjusted Rainfall Estimates
from Commercial Microwave Links, Hydrol Earth Syst. Sci., 2017 (accepted). |
|
|
|
|
|
|