“Implementation of WRF meteorological model for the : First Step” A. Chacón1, O. Cuevas1, M.Curé1 & H. Beltrami2 1 Astrometeorology Group, Dept. Physics and Astronomy, Universidad de Valparaíso, Chile 2 Environmental Sciences Research Center, St. Francis Xavier University, Canada

INTRODUCTION As for the parameterization were use: Cerro Paranal (24º 40’S & 70º 25’W) is located in Planetary The Yonsei Region, at 2635m of altitude and 12 Km inland Mellor-Yamada-Janic Figure 4: Times series of temperature at 2 m by 1 hour of the weather boundary University from the Pacific coast. Since this place present very dry (MYJ) station (red line), WRF model using MYJ-FNL (green line) and WRF Layer (YSU) model YSU_FNL (black line). conditions and clear skies, it hosts an important Table 2: PBL parameterizations used by the WRF model. astronomical observatory, the (VLT). DATA WRF model were compared with radiosondes launches at Paranal Observatory in 2009 and near-surface-data from the weather station. DATES Paranal 1 August 29th to August 10th, 2009 Paranal 2 November 9th to 19th, 2009

Table 3: Dates used to compare the WRF model with observations. Figure 5: Times series of temperature at 2 m by 3 hours of the weather WRF outputs every 1h (initiated with FNL analyses) and 3h station (red line), WRF model using MYJ-ECHOG (blue line) and WRF

(initiated with ECHO-G) were used in the comparison. model YSU_FNL (magenta line). Figure 1: Location of Paranal Observatory in in METODOLOGY To compare this vertical profile was calculated the average of Chile. each variable (from the radiosonde and the data from the WRF •Selection of meteorological variables and vertical profiles. One of the main aspects to consider in astronomical model). •Statistic analysis (RMSE, BIAS & Corr). activities is the behavior of the atmosphere because depending of his state the observatory can select the RESULTS appropriate instrument to be used during the observational The table 4 and 5 show the statistical summary in the comparison schedule. The European Southern Observatory (ESO) uses between WRF model and radiosondes in Paranal. global weather forecasts for Paranal, but they have low temporal and horizontal resolution. For this reason, the Paranal 1 Astormeteorology group initialed a study to implement the S. P. MYJ-FNL YSU-FNL MYJ-ECHOG YSU-ECHOG Weather Research Forecasting (WRF) model over Paranal, Corr 0.9 0.6 0.5 0.4 with the aim to find the model configuration that best TEMP BIAS (ºC) -1.1 -3.5 -3.7 -4.3 reproduces the atmospheric conditions using high resolution RMSE (ºC) 2.6 6.3 5.4 6.5 Corr 0.6 0.4 0.2 0.1 simulations. A parallel study in conjunction with the HR BIAS (%) 27.8 30.6 24.1 24.9 Environmental Sciences Research Center (ESRC) is in RMSE (%) 32.2 35.8 31.8 34.9 development to understand the long-term evolution of the Corr 0.8 0.7 0.5 0.5 atmosphere over Paranal using the WRF model and PRES BIAS (hPa) 34.7 34.6 34.5 34.5 millennial simulations of the Earth’s climate from the general RMSE (hPa) 34.7 34.6 34.6 34.6 circulation model ECHO-G. Corr 0.5 0.6 0.3 0.4 WS BIAS (m/s) 0 -0.9 0.1 -0.5 RMSE (m/s) 2.6 2.4 2.9 2.9 MODEL Table 4: Statistical parameters for Paranal Site between July 29th and th WRF Model August 10 , 2009. Paranal 2 S. P. MYJ-FNL YSU-FNL MYJ-ECHOG YSU-ECHOG Corr 0.8 0.7 0.2 0.3 TEMP BIAS (ºC) -1.6 -4.1 0.7 0.6 RMSE (ºC) 2.8 6.4 5.1 5.3 Corr 0.4 -0.1 0.1 -0.1 HR BIAS (%) 30.4 37.3 32.2 28.8 RMSE (%) 31.1 41.4 36.1 33.8 Corr 0.6 0.6 0 0 Figure 6:Vertical profile of Temperature for both radiosondes campaigns PRES BIAS (hPa) 34.3 34.3 34 34 over Paranal. RMSE (hPa) 34.3 34.3 34.1 34.1 CONCLUTION Corr 0.4 0.4 0.2 0.2 WS BIAS (m/s) -0.4 1.3 0.4 0 The data simulated using FNL analysis as boundary condition Figure 2: Domains used by the WRF model to simulated the RMSE (m/s) 2.1 2.5 2.8 2.7 atmosphere over Paranal. present best results than the data using the ECHOG data, Table 5: Statistical parameters for Paranal Site between November 9th to because the FNL present the same tendency for the maximum The resolution of these domains are: 19th , 2009. temperatures for each days but a clear underestimation on the Table 4 and 5 shows that the best results are obtained with minimum values of temperature. As for vertical profiles of Domain Resoluon temperature at 2 m. Figure 4 and 5 show the comparison of WRF temperature the FNL analysis present lowest errors than ECHOG 1 27 Km model with weather station observations in Paranal 1. data. 2 9 Km REFERENCE 3 3 Km [1] http://dss.ucar.edu/datasets/ds083.2/ [2] http://www.cru.uea.ac.uk/cru/projects/soap/data/model/ 4 1 Km echog.htm

Table 1: Domains resolution used by the model in this study. The WRF model uses a terrain-following sigma vertical coordinate with initial and boundary conditions are from FNL analyses [1] and ECHOG data [2].