Verification of the BOLAM weather prediction model over the area of Cyprus

Transcription

1 Adv. Geosci., 23, , 2010 doi: /adgeo Author(s) CC Attribution 3.0 License. Advances in Geosciences Verification of BOLAM wear prediction model over area of Cyprus K. Savvidou 1, K. Lagouvardos 2, S. Michaelides 1, V. Kotroni 2, and P. Constantinides 1 1 Meteorological Service, Nicosia, Cyprus 2 National Observatory of Ans, Ans, Greece Received: 24 March 2009 Revised: 26 August 2010 Accepted: 10 September 2010 Published: 4 October 2010 Abstract. The purpose of this study is verification of BOLAM wear prediction model over area of Cyprus. The verification period spans from 1 January 2007 till 31 December 2007 and parameters studied are: temperature, wind and precipitation. The model forecasts are compared to observations recorded at three locations on island, where Automatic Wear Observing Systems are operated by Meteorological Service of Cyprus, namely, those of Larnaka and Paphos Airports and at Athalassa. The statistical analysis includes calculation of mean error, mean absolute error and standard deviation. Based on construction of a contingency table, probability of detection of a precipitation event and false alarm rate are calculated. Finally, an example of BOLAM forecasts for a case study of a low pressure affecting Cyprus during winter is presented and discussed. 1 Introduction In framework of INTERREG RISKMED project (Wear Risk Reduction In The Mediterranean), Cyprus Meteorological Service implemented hydrostatic model BOLAM (Bologna Limited Area Model) as core forecast tool, in order to provide wear risk predictions over area of Cyprus, including its maritime region. BOLAM wear forecast outputs are used in order to provide warnings for high impact wear (high and low temperatures, strong winds, heavy precipitation, etc.) through a dedicated early warning system. In framework of this project, oneyear verification of BOLAM forecasts is performed, in order to quantitatively estimate forecast errors. In this manner, model s forecast skill is tested as a scaled representation of Correspondence to: K. Savvidou (lefele@cytanet.com.cy) forecast accuracy with reference to observed conditions. The verification is performed using ground wear stations across Cyprus, while a specific case-study with heavy precipitation over island is investigated in more detail. The paper is organized as follows: Sect. 2 provides a short description of BOLAM model, as well as a description of verification method and data used. Section 3 provides verification results for temperature, wind and precipitation; Sect. 4 presents case-study analysis. Finally, conclusions are drawn in Sect BOLAM model, verification data and methodology The BOLAM model used in this study is described in Buzzi et al. (1994, 1997, 1998) and Buzzi and Foschini (2000). It is a hydrostatic model on an Arakawa C grid (rotated lat.-lon. coordinates) and uses σ vertical coordinates. BO- LAM uses an explicit microphysical scheme with two water and three ice species (Schultz, 1995), as well as convective parameterization scheme proposed by Kain and Fritsch (1993), with implementation of modifications suggested by Spencer and Stensrud (1998), regarding delay of downdrafts in newly developed convection. For operational implementation of BOLAM at Cyprus Meteorological Service, two grids are used: The coarse grid consisting of points with a 0.21 horizontal grid interval ( 23 km), covering area of Eastern Mediterranean, The fine grid, consisting of points with 7 horizontal grid interval ( 7 km), covering area of Cyprus and adjacent seas. In vertical, 30 levels are used in coarse grid and 40 levels in fine grid, with model top at about 10 hpa. The nesting of inner grid starts at t + 6, based on boundary conditions provided by BOLAM outer grid at 2 h interval. Published by Copernicus Publications on behalf of European Geosciences Union.

2 Larnaka reveals that maximum values are skill to noted local does again noon not at and decrease lead times values with which are forecast higher correspond during lead time w to local noon and values are higher during period. attributed warm to months, weakness around of 4kt. This model can to be predic attributed 94 to weakness K. Savvidou of et al.: Verification model to of Similar predict BOLAM results wear diurnal prediction concerning cycle modelof over magnitude area sea ofbreeze. Cyprus of wind Similar results concerning magnitude of Lagouvardos wind forecast et al. errors (2003) were for also verification found by of BO Lagouvardos et al. (2003) for verification of BOLAM Wind model Speed over Mean Greece. Error 1.5 The MAE and Standard Deviation (STDEV 0.5 The MAE and Standard Deviation (STDEV) of ME have a similar behavior in all three stations and for all lead times (see Figs all three stations and for all lead times (see Figs. 2a, 2b). The lowest values are found at - Athalassa and highest at Larnaka. Furrmore -1.5 Athalassa and highest at Larnaka. Furrmore, skill does it not should decrease be noted with that forecast forecast lead time w -2.5 skill does not decrease with forecast lead time within 72 hours of simulation period. period. The orography fields are derived from a terrain data file with 30 arc s resolution provided by US Geological Survey (USGS). The initial and boundary data for BOLAM outer domain are provided by operational global model GFS, at 6-h intervals. The duration of simulations is 72 h, initialized every day using 00:00 UTC GFS data. The parameters that are verified are temperature, wind speed and precipitation. The observations are provided by three automatic wear observing stations located at Airports of Larnaka (33.37 E, N) and Paphos (32.49 E, N) which are coastal stations, and at Athalassa (33.4 E, 35.1 Wind N), Speed which Mean is an Error inland station with elevation 162 m a.m.s.l. (above mean sea level). For temperature and wind speed, observed values were compared against fine grid forecast values at 11 lead times: T + 12, T , T + 24,..., T + 72, for entire The forecast values are selected from nearest to stations land grid box of model. The (ME), Mean Absolute Error (MAE) and ir respective standard deviations were calculated for whole year, but also separately for cold and warm months of year. For precipitation, comparison was performed through use of daily recorded precipitation amounts at three stations and of BOLAM fine grid forecasts, averaged over 5 grid boxes in cross arrangement Wind Speed around ME for each warm station. months The probability of detection (POD) and false alarm rate (FAR) for four precipitation thresholds: 1, 5, 10 and 20 mm for 2007 were calculated. (kts) (kt) - T+12 3 Results -5.0 T+18 T Wind speed T+30 T+36 T+42 The forecast skill ATHALASSA of modellarnaka for wind PAPHOS is good (see Fig. 1a) and especially for Athalassa, where ME is less than 1kt for all lead times (1 kt = m/s). At Paphos and Larnaka, model underestimates wind speed at all lead times with ME exhibiting a diurnal cycle. The highest underestimation is found at lead times which correspond at 12:00 UTC; less than 2 kt for Paphos and around 3 kt for Larnaka. The seasonal verification (Fig. 1b and c) of model reveals that at Athalassa wind speed is overestimated during cold months (with ME values lower than 1 kt) and is underestimated during warm months (when ME reaches its maximum values, around 2.5 kt) at lead times which correspond at 18:00 UTC. The seasonal pattern of ME at Larnaka reveals that maximum values are noted again at lead times which correspond to local noon and values are higher during warm months, around 4 kt. This can be attributed to weakness of model to predict diurnal cycle of sea breeze. Similar results concerning magnitude of wind forecast errors were also found by T+48 T+54 T+60 T+66 (kts) Mean (kt) Error (kt) (kts) -5.0 T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T+66 Wind Wind Speed Speed ME for Mean cold Error months 1.5 Wind Speed ME for warm months T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T ATHALASSA Lead LARNAKA time PAPHOS ( b ) T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T+66 Wind ATHALASSA Speed ME for warm LARNAKA monthspaphos Fig. 1. Wind verification: for th - T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T+66 months, and for warm months. (kt) Fig. 1. Wind verification: mean error for year, mean Fig. error for 1. Wind cold months, verification: and mean error for Mean warm months. Error for th Fig. 1. Wind verification: months, for and year, Mean Error for for warm cold months. months, and for warm months T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T+66 Lagouvardos et al. (2003) for verification of BOLAM model over Greece. The MAE and Standard Deviation (STDEV) of ME have a similar behavior in all three stations and for all lead times (see Fig. 2a and b). The lowest values are found at Athalassa and highest at Larnaka. Furrmore, it should be noted that forecast skill does not decrease with forecast lead time within 72 h of simulation period. 3.2 Temperature The temperature is underestimated by model in all three stations and at all lead times (Fig. 3a). A 24-h cyclic behavior is noted for forecast lead times of second and third day (kt) (kt) Adv. Geosci., 23, , 2010

3 Athalassa and Larnaka is overestimated during night a The 4 largest temperature underestimation is underestimated is found during by warm model mon in The temperature is underestimated by model (Fig. which in 3a). all correspond three A 24-h stations again cyclic at and behavior 0600UTC. at all lead is noted times for forec (Fig. K. Savvidou 3a). A et 24-h al.: Verification cyclic behavior of BOLAMis wear noted prediction for day forecast model of over lead simulation area times of Cyprus of period. second For and first third 95 3 lead tim day of simulation period. For first 3 lead The times, best forecast ME is skill less is than found or equal at Athalassa to 1K. and Wind Speed Mean Absolute Error Wind Temperature Speed STDEV of ME The best forecast skill is found at Athalassa and values worst of at ME Paphos. are found The highest at lead absolute times which co values of ME are found at lead times which absolute correspond values at 0600UTC are found and during lowest nighttime ( absolute values are found during nighttime verification ( T+12 T+18 and reveals T UTC). T+30 T+36 (see T+42Figs. T+48 T+54 The 3b, T+60seasonal 3c) T+66 that during verification T+ reveals (see Figs. 3b, 3c) that during Athalassa cold and Larnaka months, is overestimated temperature during at - nigh Athalassa and Larnaka is overestimated during largest night and underestimation underestimated is found during during day. The warm m largest underestimation is found during warm which months correspond at Larnaka again (4-5K) at 0600UTC. at lead times T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T+66 T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T+66 which 4 correspond again Lead at time 0600UTC. MAE STDEV of ME Temperature Temperature for Warm months 5.0 Fig Wind verification: Mean Absolute Error, - and Standard deviation of T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T+66 Mean error. STDEV of ME 3.2 Temperature The temperature is underestimated by model in all three stations and at all lead times ATHALASSA (Fig. 3a). A 24-h cyclic behavior is noted for forecast lead times LARNAKA PAPHOS of second and third day of simulation period. For first 3 lead times, Temperature ME Mean ( is c) Error less for than Warm or months equal to 1K. Fig. 2. Wind verification: mean absolute error, and standard The deviation best Temperature offorecast mean Mean error. skill Error is for Warm found months at Athalassa and Fig. 3. Temperature worst Paphos. verification: The highest Mean absolute Error for th olute values Error, and of ME Standard are found deviation lead of times which months, correspond and Mean at 0600UTC Error for and warm lowest months. absolute values are found during nighttime -(1800 T+12 T+18and T+24 T+30000UTC). T+36 T+42 T+48 T+54 The T+60 T+66 seasonal of simulation period. For first 3 lead times, ME - verification T+12 T+18 T+24 reveals T+30 T+36(see T+42 T+48 Figs. T+54 T+60 3b, T+66 3c) is less than or equal to 1 K. The best forecast skill is found that during cold months, temperature at at Athalassa and Larnaka worst at Paphos. is overestimated The highest absolute during night -5.0 and underestimated during day. The values of ME are found lead times which correspond -6.0 largest -5.0 underestimation is found during warm months at Larnaka (4-5K) at lead times at 06:00 UTC and lowest absolute values are found during in nighttime correspond all three (18:00stations andagain 00:00 UTC). and at 0600UTC. The at all seasonal lead verifica- times -6.0 he model which tion reveals (see Fig. 3b and c) that during cold months, ted for forecast lead times of second and third temperatureathalassa Athalassa LARNAKA and LarnakaPAPHOS is overestimated st 3 lead during times, night and underestimated ME is less than or equal to 1K. Temperature Mean during Error day. The largest underestimation worst isat found Paphos. ( during c) The warmhighest months at absolute Larnaka Fig. mean 3. Temperature error for cold months, verification: and mean error formean warmerror for Fig. 3. Temperature Temperature verification: ( c) for mean Cold error months for year, assa and (4 5 K) lead times which correspond again at 06:00 UTC. months. es which Fig. 3. correspond Temperature 0600UTC verification: and Mean lowest Error months, for and year, Mean Error for for warm cold months. The temperature MAE (Fig. 4a) has a periodical behavior with (1800 and ighttime months, T+12 peaks T+18 and at T+24 06:00 Mean T UTC). Error T+36 andt+42 lowfor T+48 points The T+54atwarm T+60 18:00 seasonal T+66 UTC. months. - T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T+66 that during The highest values cold aremonths, found at coastal stations. temperature The STDEV time between nocturnal katabatic wind to daily prevailing wind, but this must be examined in a future study. at - of ME (Fig. 4b) follows a similar behavior, but lowest during values night forand coastal underestimated stations are foundduring at 00:00 UTC. day. Once The warm more, months verification Larnaka has shown(4-5k) that forecast at lead skilltimes does 3.3 Precipitation 66 Mean Erron not decrease within 72-h duration of simulation. The underestimation of temperature by model can The results of calculations of POD and FAR for four be attributed to weakness of model to predict daily precipitation thresholds are shown in Table 1. As it was superadiabatic warming Temperature Mean especially during summer and expected, as precipitation threshold Error for Cold months increases, POD radiation cooling during winter, since model produces decreases, except at Larnaka where, an increase in POD is forecasts Temperature for temperature at for 2 m. Warm Also, months largest underestimation which is noted at 06:00 UTC can be associated than or two stations for almost all precipitation noted from 5mm to 10 mm. At Larnaka, POD is higher with time lag of model in predicting transition thresholds. The lowest FAR is found at Athalassa (13% T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T+66 Adv. Geosci., 23, , Mean Erron Temperature Wind Speed STDEV of ME T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T+66 Mean Mean Erron Error Mean Erron Temperature for Cold months T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T+66 - T+12 T+18 T+24 T+30T+36 T+36T+42 T+42 T+48 T+48 T+54 T+54 T+60 T+60 T+66 T ATHALASSA LARNAKA PAPHOS

4 ME (Fig. 4b) follows a similar behavior, but lowest values for coastal stations are found at 0000UTC. Once more, verification has shown that forecast skill does not decrease 96 within 72-h K. Savvidou duration et al.: of Verification simulation. of BOLAM wear prediction model over area of Cyprus Table The 1. POD underestimation and FAR for three stations of and fortemperature four precipitation thresholds. Table by 2. The ME model and MAEcan for hit be events. attributed to 5 weakness of model to predict daily superadiabatic Athalassa warming Larnaka especially Paphos during summer and radiation cooling during winter, since Precipitation Athalassa Larnaka Paphos ME model 1.1 produces forecasts for s a periodical temperature Thresholds behavior at POD FAR with 2m. peaks Also, POD FAR at 0600UTC largest underestimation POD FARand MAEwhich 4.3is noted 3.9 at 0600UTC 4.1 can alues be are associated found (mm) at (%) coastal with (%) stations. time (%) lag (%) The of STDEV (%) model (%) of in predicting transition time between nocturnal 1 katabatic wind 96 to 43 daily 91 prevailing 33 r, but lowest values for coastal stations are wind, but this must be examined in a rification future has 10 study. shown 64 that 13 forecast skill 56does 31 not e simulation. erature by model Temperature can be MAE attributed to e daily superadiabatic warming especially during 5.0 ing winter, since model produces forecasts for Mean Absolute Error 4.0 st underestimation 3.0 which is noted at 0600UTC can odel in predicting transition time between prevailing wind, but this must be examined in a T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T+66 STDEV of ME Temperature STDEV of ME T+12 T+18 T+24 T+30 T+36 T+42 T+48 T+54 T+60 T+66 Temperature STDEV of ME Fig. 4. Temperature verification: MAE, and STDEV of ME. STDEV of ME Precipitation 0.5 Fig hpa geopotential height analysis at 12:00 UTC, The results T+12 T+18of T+24 T+30calculations T+36 T+42 T+48 T+54 of T+60 POD T+66 and FAR for four precipitation thresholds are 3 February Mean sea level pressure analysis at shown in Table 1. As it was expected, as 12:00 precipitation UTC, 3 Februarythreshold increases, POD decreases, except at Larnaka where, an increase in POD is noted from 5mm to 10mm. At 4 A case-study: a depression that affected Cyprus Larnaka, POD is higher than or two stations for almost all precipitation Fig. 4. Temperature verification: MAE, and STDEV of ME. 3 February 2007 AE, and thresholds. STDEV The of lowest ME. FAR is found at Athalassa (13% for 10mm threshold) and at Paphos (19% for 5mm threshold). for 10 mm threshold) and at Paphos (19% for 5 mm threshold). Table It is 1. alsopod interesting and tofar presentfor results three of stations ME and and for four precipitation thresholds. MAE for hit events (forecast and observed) which are shown in Table 2. and FAR The for values offour ME precipitation and MAE are similar thresholds inathalassa all three are Larnaka Paphos stations, indicating that re is not an area-dependent bias, as Precipitation of precipitation model in forecasting threshold precipitation increases, over domain POD of increase simulation. in POD is noted from Thresholds 5mm to 10mm. POD FAR At PODFARPOD FAR or two stations for almost all (mm) precipitation (%) (%) (%) (%) (%) (%) at Athalassa (13% for 10mm threshold) and at ations and Adv. for Geosci., 23, four , precipitation 2010 thresholds halassa Larnaka Paphos On 3 February 2007, a diffluent upper- level trough associated with a deep surface low (998 hpa) centered south of Cyprus (see Fig. 5a and b) affected area. In 12 h, it moved to east, leaving area in a northwest flow (not shown). This well organized frontal depression resulted in high accumulation precipitation amounts, as shown in Fig. 6 (re is no accessibility to measurements in north part of island). In souast part of island, daily recorded precipitation reached 160 mm. At Larnaka, recorded precipitation between 15:00 and 18:00 UTC was 35.2 mm. In order to evaluate model s skill in predicting depressions that affected area of Cyprus, a first comparison is performed between actual surface charts and 500 hpa isobaric surface with model s forecasts.

5 forecast surface low is very close to actual one. It can be stated here that model was able to catch evolution of surface and upper air patterns, two days before synoptic event. K. Savvidou et al.: Verification of BOLAM wear prediction model over area of Cyprus 97 Fig. 6. The area averaged precipitation for 3 February Fig. 6.The area averaged precipitation for 3 February The BOLAM coarse grid forecasts are shown in Fig Conclusions The T + 60 forecast of 500 hpa geopotential height Accumulated precipitation maps, provided by BOLAM fine grid, are shown in Fig. 8 (Fig. 7a) and of mean sea level pressure (Fig. 7b), valid at Verification of model outputs is an important component in (first day of simulation). An examination of maps shows good results as far as 12:00 UTC, 3 February 2007, show trough positioned wear forecasting, as it is through this phase that confidence with toprecipitation an operational amounts model is built. exceeding The Cyprus 40mm Meteo- more to east, with location respect to of verifying precipitation analysis is (shown concerned, in Fig. 5a), whilewithin surface 3 hours lowover is predicted sourn over Cyprus between rological 0900 Service and 1200UTC verifies limited (Fig. 8a) areaand models later that over it runs norastern part of central island, part with of a central island. pressure The of results operationally for (see third Orphanou day of etsimulation al., 2006); this were is quite first time 1002 hpa, 4 hpa shallower satisfactory, than as it real concerns conditions (shown location of that precipitation BOLAMbut model is predicted verified over amount area was of very Cyprus. in Fig. 5b). The T + low; 36 forecasts 15mm of within same 3 hours. fields (shown For in second In thisday study, of one simulation, year of BOLAM forecast model forecasts output was over Fig. 7c and d), depict totally a small wrong; scale cut-of precipitation low was 500predicted hpa area only of over Cyprus are norrnmost verified by comparing part of island. model output level which is not evident in analysis chart, while low with observations at three main locations over island. pressure at surface 5 Conclusions is to west with respect to its real The verified parameters are temperature, wind speed and precipitation. The verification was performed using data for position, but once more about 2 hpa shallower than actual low. Finally, T + 12 forecast (Fig. 7e and f) reproduces year Verification of model outputs is an important component in wear forecasting, as it is well structure at 500 hpa level, and forecast surface low is very close to actual one. It can be stated here formance is found at Athalassa, inland station, with ME Verification of wind speed has shown that best per- through this phase that confidence to an operational model is built. The Cyprus that model was Meteorological able to catch evolution Service of verifies surface limited area models that it runs operationally (see less than 1 kt. The seasonal verification reveals higher ME and upper air patterns, Orphanou two dayset before al., 2006); synoptic this is event. first time during that warm BOLAM months, model around is 2.5verified kt at leadover times which Accumulated precipitation area of Cyprus. maps, provided In this study, by BOLAM one year correspond of BOLAM at 18:00 model UTC. forecasts Over coastal over stations, area of wind fine grid, are shown Cyprus in Fig. are 8 (first verified day ofby simulation). comparing An model speed is output underestimated, with observations with highest at ME three around main 4 kt at examination of locations maps showsover good results island. as far as The verified location Larnaka parameters during are warm temperature, months and wind at 12:00 speed UTC. and Over- of precipitation. is concerned, The verification with precipitation was performed all, using skill of data for model forecasts year of wind speed is very amounts exceeding 40 mm within 3 h over sourn Cyprus good. between 09:00 and 12:00 Verification UTC (Fig. 8a) of and later wind over speed has shown The temperature that isbest generally performance underestimated is found by at model central part of island. Athalassa, The results inland for third station, day of with sim-mulation were quite satisfactory, except less than during 1kt. nighttime The seasonal and during verification cold months reveals at Athalassa and2.5kt Larnaka. at lead Thetimes absolute which values correspond of ME are at higher higher ME as during it concerns warm location months, of around precipitation but predicted amount was very low; 15 mm during warm months and during daytime. The highest within 3 h. For second day of simulation, forecast absolute values are found at 06:00 UTC at Larnaka, being output was totally wrong; precipitation was predicted only around 4 5 K. The highest absolute values during cold over norrnmost part of island. months are noted at 06:00 UTC at Paphos, being around 3 K. The forecast skill of model in predicting precipitation is better at Larnaka for all precipitation thresholds, as values of POD are higher than values for or two Adv. Geosci., 23, , 2010

6 98 K. Savvidou et al.: Verification of BOLAM wear prediction model over area of Cyprus (d) (e) (f) Fig. 7. T hpa geopotential height, verifying at 12:00 UTC, 3 February T + 60 mean sea level pressure, verifying at 12:00 UTC, 3 February T hpa geopotential height, verifying at 12:00 UTC, 3 February (d) T + 36 mean sea level pressure, verifying at 12:00 UTC, 3 February (e) T hpa geopotential height, verifying at 12:00 UTC, 3 February (f) T + 12 mean sea level pressure, verifying at 12:00 UT, 3 February Adv. Geosci., 23, , 2010

7 K. Savvidou et al.: Verification of BOLAM wear prediction model over area of Cyprus 99 (d) Fig h accumulated precipitation, ending at 12:00 UTC, 3 February h accumulated precipitation, ending at 15:00 UTC, 3 February h accumulated precipitation, ending at 18:00 UTC, 3 February (d) 3-h accumulated precipitation, ending at 21:00 UTC, 3 February stations. Meanwhile, values of FAR for two higher precipitation thresholds are lower at Athalassa than at or two stations. Finally, analysis of a specific case study, showed ability of model to provide accurate forecasts for events producing high-impact wear. It is in authors plans to continue verification of model for a longer period and to extend verification to warnings provided by BOLAM. Acknowledgements. This work was carried out as part of RISKMED project which was co-funded by European Union (INTERREG IIIB) and Government of Cyprus. The authors acknowledge provision of BOLAM model by Andrea Buzzi (ISAC-CNR, Italy) and his help during implementation of model at Cyprus Meteorological Service. The authors are also grateful to National Centers for Environmental Prediction (USA) for providing GFS initial and forecast field data which allow operational use of meteorological models at Cyprus Meteorological Service. Edited by: K. Nicolaides and A. Orphanou Reviewed by: S. Davolio and F. Tymvios Adv. Geosci., 23, , 2010

8 100 K. Savvidou et al.: Verification of BOLAM wear prediction model over area of Cyprus References Buzzi, A., Fantini, M., Malguzzi, P., and Nerozzi, F.: Validation of a limited area model in cases of Mediterranean cyclogenesis: surface fields and precipitation scores, Meteorol. Atmos. Phys., 53, , Buzzi, A., Cadelli, R., and Malguzzi, P.: Low level jet simulation over Antarctic ocean, Tellus A, 49, , Buzzi, A., Tartaglione, N., and Malguzzi, P.: Numerical simulations of 1994 Piedmont flood: role of orography and moist processes, Mon. Wear Rev., 126, , Buzzi, A. and Foschini, L.: Mesoscale meteorological features associated with heavy precipitation in sourn Alpine region, Meteorol. Atmos. Phys., 72, , Kain, J. S. and Fritsch, J. M.: Convective parameterization for mesoscale models: The Kain-Fritsch scheme, The Representation of Cumulus in numerical models, Meteor. Monogr., No 46, Am. Meteororl. Soc., , Lagouvardos, K., Kotroni, V., Koussis, A., Feidas, C., Buzzi, A., and Malguzzi, P.: The meteorological model BOLAM at National Observatory of Ans: assessment of two-year operational use, J. Appl. Meteorol., 42, , Orphanou, A., Michaelides, S., Savvidou, K., Constantinides, P., Schulz, J.-P., and Voigt, U.: Preliminary verification results of DWD limited area model LME and evaluation of its storm forecast skill over area of Cyprus, Adv. Geosci., 7, , 2006, Schultz, P.: An explicit cloud physics parameterization for operational numerical wear prediction, Mon. Wear Rev., 123, , Spencer, P. L. and Stensrud, D. J.: Flash flood events: importance of subgrid representation of convection, Mon. Wear Rev., 126, , Adv. Geosci., 23, , 2010