Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Standard view
Full view
of .
Look up keyword
Like this
0 of .
Results for:
No results containing your search query
P. 1
Torrential Rains on the Spanish Mediterranean Coast: Modeling the Effects of the Sea Surface Temperature

Torrential Rains on the Spanish Mediterranean Coast: Modeling the Effects of the Sea Surface Temperature

Ratings: (0)|Views: 28 |Likes:
Published by Paco Pastor Guzman

More info:

Categories:Types, Research, Science
Published by: Paco Pastor Guzman on Nov 30, 2011
Copyright:Attribution Non-commercial


Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less





40J O U R N A L O F A P P L I E D M E T E O R O L O G Y
2001 American Meteorological Society
Torrential Rains on the Spanish Mediterranean Coast: Modeling the Effects of the SeaSurface Temperature
, M
J. E
, D
M. M
Centro de Estudios Ambientales del Mediterra´neo, Paterna, Valencia, Spain
(Manuscript received 18 January 2000, in final form 27 November 2000)ABSTRACTTorrential rains are a frequent meteorological risk in the Mediterranean Basin, and the work reported here ispart of a long-term study that includes the analysis of the synoptic conditions involved in their genesis. This paperstudies the role of SST in torrential rain development. Two episodes were selected for simulation with the RegionalAtmospheric Modeling System. Three runs of each were performed by progressively improving the SST input datasources: from monthly climatological averages, to data from the International Satellite Land Surface ClimatologyProject, to near-to-real-time data derived from National Oceanic and Atmospheric Administration (NOAA) satelliteimages. For the first episode, the maximum total precipitations calculated were 242, 301, and 496 mm, respectively,versus more than 550 mm measured. For the second event, the maxima were 316, 349, and 378 mm, respectively,versus more than 450 mm measured. The conclusion is that significant improvements in the modeling of peak precipitation can be expected when using SST derived from NOAA satellite data.
1. Introduction and background
Intense rain events in the western MediterraneanBasincan occur from late summer to winter, and in the lastfew decades they have caused numerous catastrophicfloods in several countries, with human casualties. Onthe east coast of the Iberian Peninsula, the Valencia re-gion, shown in Fig. 1, is one of the most affected areas.Figure 2 shows the frequency and annual distribution of torrential rain events in this region during the last 27years, including 48 severe episodes. Damages could bereduced by the proper forecasting of the meteorologicalsituations leading to these events; some efforts have al-ready been made to this effect (Codina et al. 1997).Our framework for the study of these events has beenthe back door (cold) front concept (Huschke 1959;Blue-stein 1993), in which the cold front is generated by thenortherly, northeasterly, or easterly flow of cold(er)con-tinental air over the warmer Mediterranean. Most of theprecipitation events have been observed to occur di-rectly over the sea, and the torrential rains over easternSpain appear to be the result of a two-step process. Theformation of a potentially unstable air mass through theadvection of the cold(er) air over the warmer sea is thefirst step; here the temperature difference between theadvected air and the sea surface temperature(SST)playsthe key role in the recharge of moisture. The second
Corresponding author address:
Prof. Milla´n M. Milla´n, Centro deEstudios Ambientales del Mediterra´neo (CEAM), Charles Darwin 14,Parque Tecnolo´gico, 46980 Paterna, Valencia, Spain.E-mail: millan@ceam.es
step is the easterly transport of the potentially unstableair mass toward the coastal regions where orographiclift can trigger the precipitations.We studied 32 torrential rain events for whichthewest-ern Mediterranean SST could be retrieved from NationalOceanic and Atmospheric Administration (NOAA) sat-ellite images. These images were processed with an al-gorithm developed by Ba´denas et al. (1997), which ob-tains the SST with an error of less than 0.5
C. Our anal-ysis of the NOAA images and the trajectories of thesurface air masses indicated that the Mediterranean SSTalong their path drops by 3
C with respect to valuesprevious to the events. Because the SST, as a long-term-varying parameter, did not show significant changes dur-ing the days previous to the event, our assumptions were1) that the drop in the observed SST after the event wasmainly the result of vigorous evaporative cooling alongthe back door (cold) front and thus 2) that the warmerareas of the Mediterranean acted as a source of moisturefor the torrential episodes (Milla´n et al. 1995). Otherworks have also confirmed the relevance of SST on thedevelopment of torrential rains on the Spanish Mediter-ranean coast (Ferna´ndez et al. 1995, 1997).The objective of this work was to study further theimportance of SST data quality in the quantitative sim-ulation of precipitation amounts. For this purpose, thetool used was the Regional Atmospheric Modeling Sys-tem (RAMS), which was adapted to the specific con-ditions in the western Mediterranean by including moreaccurate regional topography, as well as better land use
P A S T O R E T A L .F
. 1. Location and orography of the Valencia region in the western Mediterranean Basin, andmodeling grids used for this work.
data and grid-size selection, and by testing the outputsagainst field data from the study of forest fires (Go´mez-Tejedor et al. 1999) and air pollution dynamics in theregion (Salvador et al. 1999, Milla´n et al. 2000).Through this process, results from previous studieswereincorporated to improve the model.
2. Modeling
a. The model
The numerical mesoscale model used for this work (RAMS) was developed at Colorado State Universityby merging different mesoscale and cloud models (Piel-
40J O U R N A L O F A P P L I E D M E T E O R O L O G YF
. 2. Frequency and annual distribution of torrential rain eventson the Spanish east coast, Valencia region (1971–95). The criterionused for selecting a rain event is a rainfall equal to, or higher than,125 mm of precipitation in 24 h.T
1. RAMS grid settings.GridNo.Grid dimension(mesh units)Mesh size(
 x, y
; km)Time step(s)12322
ke et al. 1992; Walko et al. 1995a). The RAMS modelis built around the full set of primitive dynamical equa-tions, which describe atmospheric motions, supple-mented with optional parameterizations for differentmeteorological magnitudes. It was fundamentally de-veloped as a limited-area model for regional mesoscalestudies.RAMS uses the Arakawa C grid (Messinger and Ar-akawa 1976), in which all the thermodynamic and mois-ture variables are defined in the center of a grid volumewith the velocity components staggered by one-half of a grid space in their normal direction. This procedurehas several advantages, including the isotropy of thevelocity component locations relative to the thermo-dynamic variables, which is very important for massand flux conservation.Detailed descriptions of the microphysics and mois-ture schemes used by RAMS can be found in Walko etal. (1995b) and Meyers et al. (1996). In the case of convection, the RAMS model uses a simplified Kuo(1974) parameterization as described by Molinari(1985). For this work, the explicit microphysics optionsof RAMS—cumulus parameterization and cloud hailand rain microphysics—were activated (for all threegrids). These options facilitate the diagnosis of the num-ber concentration of different species of liquid or icefrom a mean diameter specified from a default value inthe code and the forecasted mixing ratio.Properties of bare soil are calculated in RAMS witha multilayer soil model described by Tremback and Kes-sler (1985). This model calculates prognostic equationsfor the soil surface temperature and water content. Soilmoisture content is initialized in the model simulationsby a parameter specified in the model before the exe-cution of the simulations. The values of this parameterrange from 0.0 (totally dry soil) to 1.0 (fully saturatedsoil) and represent the percentage of total water capacitythat the soil can hold; in our case we have chosen avalue of 0.3 for each soil level.Our simulation uses three interactive nested gridsshown in Fig. 1. Grid 1, with a 1600 km by 960 kmdomain, links with the synoptic systems. The Valenciaregion is covered by grid 2, with a 400 km by 480 kmdomain; grid 3 with a 120 km by 120 km domain, isintended to resolve the small-scale phenomena. Themesh resolutions for each grid are shown in Table 1.The ratio between the second and third mesh sizes is 8:1, which is greater than normally used. It represents acompromise between our computing capabilities and thespatial resolution desired, and it was selectedaftertryingdifferent configurations for the grids. All three gridsspan 32 vertical levels, with their tops reaching 15 km.
b. Methodology
Two cases have been selected for which the NOAAdatasets were of excellent quality. The first, from 4 to7 September 1989, was one of the strongest episodesin the last 10 years, and the second, from 28 Septemberto 1 October 1986, can be considered to be a more‘‘average’’ event.The methodology followed was to run three RAMSsimulations of each event to detect model output vari-ations as the SST data input was changed. The firstsimulations were run with the RAMS-distributed SSTfiles and are considered the ‘‘control’’ runs. The secondsimulation used SST data obtained during the Interna-tional Satellite Land Surface Climatology Project(ISLSCP), and the third simulation used the SST dataderived from the closest available NOAA images pre-vious to the event. In each event, therefore, the initialmeteorological parametersremainedunchangedforeachof the three simulations, and the only changes intro-duced in each new initialization were the sea surfacetemperature data.The model has been initialized with a three-dimen-sional grid of atmospheric data provided by EuropeanCentre for Medium-Range Weather Forecasts analysis,using the analysed datasets with 2.5
and 1
horizontal spacing for the 10 standard pressurelevels.These datasets are ingested by the model Data Prepa-ration Package and converted to the RAMS format. TheIsentropic Analysis package is used to create the ini-tialization variable fields, which the model utilizes intwo different ways. The first way is that, when a sim-ulation is begun, the model fields are initialized directlyfrom the first initialization file.

You're Reading a Free Preview

/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->