Preliminary assessment of the average temperature on the Ecuador surface for 2010, obtained by the Weather Research Forecasting (WRF) model

Abstract

Weather information is critical to diverse fields of research and environmental management. Meteorological observations are typically recorded in weather stations located at a few points within the area of interest. Through eulerian meteorological models it is possible to generate information for places where direct monitoring is not available. The Advanced Research WRF model was used to simulate the meteorological fields for the year 2010 for continental Ecuador. One master domain (cells of 36 km) and 2 nested subdomains (cells of 12 and 4 km) were used in the model. The modeled monthly mean variation for the temperature captured the tendency of the meteorological records for the 10 stations used in this study. Temperature monthly means for 3 stations were well reproduced, with differences between 0-1 °C. At 7 stations the model yielded higher or lower values in comparison with the records, with differences between 0.1 - 2.7 °C. The best correlation between modeled results and observational records happened for the 5 stations located on the coast side (Costa) (R2 between 0.90 - 0.97) whose topography is flat. The correlation coefficient was lower for the stations in the highlands (Sierra) (R2 between 0.69 - 0.89), possibly due to the influence of its complex topography. At 7 stations percentages higher than 90% were obtained of modeled values that comply with the desired accuracy, according to European criteria. At all stations the average absolute error values were lower than the upper limit that considers the USEPA. At 3 stations values of bias were within the range recommended by the USEPA. Model maps provide an integrated and coherent description of the spatial and temporal variation of temperature throughout the territory. These model products can be very useful for applications in which marginal differences in temperature between model results and observations are not relevant. Assessments with more stations and meteorological parameters are required for a more accurate study. Likewise, it is necessary to evaluate the performance of the model, considering separately the Costa, Sierra and Amazonia; with a higher spatial resolution for the Sierra’s simulations.

Keywords

• Modelo Euleriano
• meteorología
• INAMHI
• precisión deseada
• error medio absoluto