Wind energy requires accurate forecasts for adequate integration into the electric grid system. In addition, global atmospheric models are not able to simulate local winds in complex terrain, where wind farms are sometimes placed. For this reason, the use of mesoscale models is vital for estimating wind speed at wind turbine hub height. In this regard, the Weather Research and Forecasting (WRF) Model allows a user to apply different initial and boundary conditions as well as physical parameterizations. In this research, a sensitivity analysis of several physical schemes and initial and boundary conditions was performed for the Alaiz mountain range in the northern Iberian Peninsula, where several wind farms are located. Model performance was evaluated under various atmospheric stabilities and wind speeds. For validation purposes, a mast with anemometers installed at 40, 78, 90, and 118 m above ground level was used. The results indicate that performance of the Global Forecast System analysis and European Centre for Medium-Range Weather Forecasts interim reanalysis (ERA-Interim) as initial and boundary conditions was similar, although each performed better under certain meteorological conditions. With regard to physical schemes, there is no single combination of parameterizations that performs best during all weather conditions. Nevertheless, some combinations have been identified as inefficient, and therefore their use is discouraged. As a result, the validation of an ensemble prediction system composed of the best 12 deterministic simulations shows the most accurate results, obtaining relative errors in wind speed forecasts that are <15%.
18 October 2011 in International Journal of Climatology
Surface icing can cause dramatic consequences on human activities. What is more, numerical weather prediction models are not very accurate in determining freezing drizzle, which creates uncertainty when forecasting this type of weather phenomenon. Therefore, it is essential to improve the forecast accuracy of these models for such phenomena to mitigate risks caused by unforeseen freezing drizzle events. On 5 February 2012, an episode of freezing drizzle took place in the Guadarrama Mountains, at the center of the Iberian Peninsula. This episode was preceded by weak snowfall. After the freezing drizzle, moderate snowfall was recorded in the study area. This event was simulated using the Weather Research and Forecasting model. Through this analysis, we identified the meteorological factors at both synoptic scale and mesoscale that caused this episode. Wind perpendicular to an orographic barrier‐generated updrafts and retention of moisture upwind, which caused orographic clouds to appear on the north side of the Guadarrama Mountains. Atmospheric stability prevented cloud formation at midlevels at the time of the freezing drizzle, which maintained cloud top temperatures warmer than −15°C during the episode. The entrance of moisture and instability at midlevels caused cloud top temperatures substantially colder than −15°C, which coincided with snow in the mountain range. Cloud top temperature and thickness control the efficiency of the glaciation process, thereby determining the type of precipitation at the surface. Freezing drizzle risk and in‐cloud icing algorithms were developed with the aim of predicting similar events in the study area, which could mitigate impacts on human activities.
Accurate estimation of snowfall episodes several hours or even days in advance is essential to minimize risks to transport and other human activities. Every year, these episodes cause severe traffic problems on the northwestern Iberian Peninsula. In order to analyze the influence of different parameterization schemes, 15 snowfall days were analyzed with the Weather Research and Forecasting (WRF) model, defining three nested domains with resolutions of 27, 9, and 3 km. We implemented four microphysical parameterizations (WRF Single‐Moment 6‐class scheme, Goddard, Thompson, and Morrison) and two planetary boundary layer schemes (Yonsei University and Mellor‐Yamada‐Janjic), yielding eight distinct combinations. To validate model estimates, a network of 97 precipitation gauges was used, together with dichotomous data of snowfall presence/absence from snowplow requests to the emergency service of Spain and observatories of the Spanish Meteorological Agency. The results indicate that the most accurate setting of WRF for the study area was that using the Thompson microphysical parameterization and Mellor‐Yamada‐Janjic scheme, although the Thompson and Yonsei University combination had greater accuracy in determining the temporal distribution of precipitation over 1 day. Combining the eight deterministic members in an ensemble average improved results considerably. Further, the root mean square difference decreased markedly using a multiple linear regression as postprocessing. In addition, our method was able to provide mean ensemble precipitation and maximum expected precipitation,which can be very useful in the management of water resources. Finally, we developed an application that allows determination of the risk of snowfall above a certain threshold.
21 January 2014 in International Journal of Climatology
Hailstorms have become a serious meteorological risk in mid‐latitude countries because of their small‐scale development and short time span, and their forecasting remains problematic. Moreover, the lack of reliable hailstorm databases is an important handicap for Numerical Weather Prediction (NWP) model validation, time trend studies and establishment of hailstorm relationships with global warming and climate change. However, this can be overcome by ascertaining physical relationships between favourable synoptic patterns and mesoscale trigger factors for hailstorm development in a local study area.North‐eastern Spain is one of the European regions with largest number of hailstorm days (HD) observed in the summer. Since 2001, from May through September, reliable databases of hailstorm occurrence in the middle Ebro Valley have been constructed, which include intensity, temporal frequency and spatial coordinate data. June and July 2006 had the largest number of HD in 2001–2010. With the objective of analysing meteorological factors responsible for this anomaly, atmospheric patterns at low and mid‐tropospheric levels were studied. We determined a set of special synoptic configurations, with an evident deviation from climatic values generating anomalies of 850 hPa temperature relative to the characteristic values of 2001–2010 and 1950–2010. An analysis of these anomalies detected a positive trend of 850 hPa temperature and geopotential height in the western Mediterranean area. As a consequence, the characteristic synoptic circulation has changed since 1950, towards low‐level patterns favouring adequate thermodynamic environments for hailstorm development in north‐eastern Spain. Finally, after a cluster analysis, we studied periodicities in the monthly 850 hPa temperature field to improve knowledge for seasonal forecasting of hailstorms.