Statistical model for short-term prediction of heavy rainfalls locally -- Basque Ph.D. thesis

April 28, 2010

Storm water tanks are an essential element in the sanitary water networks of cities like Bilbao, as they can avoid floods in heavy rain situations. They carry out storage functions: principally, when it starts to rain heavily, they hold the initial volume rush of rainfall, preventing the sudden arrival of a large amount of water at the treatment plant. The problem is that, in order to have them work efficiently, it is necessary to know beforehand that, effectively, it is about to rain intensely; it is estimated that a margin of six hours notice is needed in order for everything to be prepared. This is why the physicist Alejandro Fernández has undertaken trials with an innovative model for statistical prediction, applying it to the public water supply network run by the Aguas Bilbao Bizkaia Water Consortium.

The PhD thesis presented at the University of the Basque Country (UPV/EHU) is entitled, Predictive precipitation model applied to the management of the public water supply network in the metropolitan area of Bilbao. Mr Fernández studied the behaviour of the variants corresponding to precipitations within a small perimeter (10 x 15 km, approximately) and with an interval of time of between 3 and 24 hours, and investigated up to what point rainfall can be predicted under these conditions.

Downscaling technique

He used the technique of downscaling, i.e. he combined various methods of obtaining useful data on predictions at a European and world scale, and he reduced this scale to the point of adapting it to the dimensions of Bilbao. For example, he took the ERA-40 re-analysis of the European Centre for Medium-Range Weather Forecast (ECMWF) as a working basis; a study of atmospheric data gathered between 1957 and 2002 and which took into account, amongst other considerations, temperature, pressure, humidity and wind. Mr Fernández also used the dynamic MM5 method; it is used in order to obtain meteorological data in real time, as well as to carry out simulations. He started at a worldwide level, gradually reducing the net until centring on Bilbao. Finally, the statistical methods were included, together with the re-analysis of the data and the dynamic methods. In this case, the researcher looked for antecedents, comparing the day of prediction with rainy days that previously had had similar characteristics.

Mr Fernández undertook trials in order to validate the model and, to this end, he calculated the number of correct hits and the number of false alarms. He compared these with prediction methods for non-local rainfall and, as he concluded, the results proved to be superior to methods such as pure climatology or certain dynamic methods. The results, likewise, are on a par with more complex models which use Earth and satellite radar; for example, Nimrod. The thesis concluded that the proposed model could improve the management of the public water supply network in Greater Bilbao. However, he also made it clear that greater development of this model is required for the management to be totally automatic, and that for the moment human supervision is essential. Mr Fernández proposed providing more data for the model for its enhancement; such as, for example, that from local pluviometry or that gathered in real time from the radar at Punta Galea, a cape located in the Basque Country.

Extreme episodes every 250 years

When a water sanitation network is built, the frequency of extreme rainfall in this metropolitan zone has also to be taken into account. In his thesis Mr Fernández investigated how to predict this frequency.

The researcher contrasted Gumbel and Pareto statistical methods to calculate this frequency. His aim was to find out the maximum possible thresholds, with a degree of certainty of at least 95 % regarding the intensity of possible precipitation in the region of Bilbao every certain number of years. In fact, in order to verify that the mentioned calculation complied with a minimum level of reliability, he used a computer simulation technique called Bootstrap Resampling.

According to the conclusions of the PhD, rainfalls such as that which devastated Bilbao in 1983 can happen with greater frequency than has been thought. To date it has been believed that these phenomena could only occur every 500 years at the most but, in fact, they can happen, on average, once every 250 years.
-end-
About the author

Mr Alejandro Fernández Ferrero (Barakaldo, 1968) has a degree in Physical Sciences and is a doctor in Physical Engineering. He undertook his thesis under the direction of Gabriel Ibarra Berastegi and Jon Saenz Agirre; lecturers respectively at the Department of Fluid Mechanics at the Higher Technical School of Engineers in Bilbao and the Department of Applied Physics II of the Faculty of Science and Technology at the Leioa campus of the UPV/EHU. Mr Fernández is currently a senior technical officer for the Technical Services of the Aguas Bilbao Bizkaia Consortium. He undertook his thesis at the UPV/EHU and at the Consortium.

Elhuyar Fundazioa

Related Rainfall Articles from Brightsurf:

Study projects more rainfall in Florida during flooding season
A new study by researchers at the University of Miami (UM) Rosenstiel School of Marine and Atmospheric Science projects an increase in Florida's late summertime rainfall with rising Atlantic Ocean temperatures.

Importance of rainfall highlighted for tropical animals
Imagine a tropical forest, and you might conjure up tall trees hung with vines, brightly colored birds, howling monkeys, and ... rain.

New study could help better predict rainfall during El Niño
Researchers at the University of Miami (UM) Rosenstiel School of Marine and Atmospheric Science have uncovered a new connection between tropical weather events and US rainfall during El NiƱo years.

Mediterranean rainfall immediately affected by greenhouse gas changes
Mediterranean-type climates face immediate drops in rainfall when greenhouse gases rise, but this could be interrupted quickly if emissions are cut.

Future rainfall could far outweigh current climate predictions
Scientists from the University of Plymouth analysed rainfall records from the 1870s to the present day with their findings showing there could be large divergence in projected rainfall by the mid to late 21st century.

NASA estimates Imelda's extreme rainfall
NASA estimated extreme rainfall over eastern Texas from the remnants of Tropical Depression Imelda using a NASA satellite rainfall product that incorporates data from satellites and observations.

NASA estimates heavy rainfall in Hurricane Dorian
Hurricane Dorian is packing heavy rain as it moves toward the Bahamas as predicted by NOAA's NHC or National Hurricane Center.

NASA looks at Barry's rainfall rates
After Barry made landfall as a Category 1 hurricane, NASA's GPM core satellite analyzed the rate in which rain was falling throughout the storm.

NASA looks at Tropical Storm Barbara's heavy rainfall
Tropical Storm Barbara formed on Sunday, June 30 in the Eastern Pacific Ocean over 800 miles from the coast of western Mexico.

NASA looks at Tropical Storm Fani's rainfall rates
Tropical Storm Fani formed in the Northern Indian Ocean over the weekend of April 27 and 28, 2019.

Read More: Rainfall News and Rainfall Current Events
Brightsurf.com is a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to Amazon.com.