Simulações numéricas com o modelo WRF para os períodos seco e chuvoso no Maranhão: impacto do uso de diferentes modelos de fechamento de turbulência

Abstract

This work aimed to evaluate the impact of different Planetary Boundary Layer (PBL) parameterization schemes in simulations for of Maranhão (Brazil) state, using the WRF model. Studies on model simulations using PBL parameterizations in tropical regions are generally rare, mainly due to the lack of experimental observations. This work is part of a project developed in partnership between UTE Pecém II, UTE Parnaíba I, Parnaíba II and III generation of Energia S.A. and the Federal University of Santa Maria-UFSM. The Thermoelectric Power Plant (TPP) Parnaíba complex is located in the Santo Antônio dos Lopes city in Maranhão, and is operated by the Brazilian company Eneva. As the production of energy through the combustion of natural gas produces residues that are released into the atmosphere, the environmental regulatory bodies demand that verification and control measures be taken, such as the monitoring of chemical species. Therefore, a good description of the atmosphere is essential for the good performance of the chemical species dispersion models. In this masters dissertation work, data from 16 automatic meteorological stations were used to validate the model simulations, in which 8 simulations were performed using two PBL parameterizations in the dry (September) and rainy (March) periods, of which 4 were performed with activation of cumulus parameterization (Kain-Fritsch) was activated only in the first grid d01 and 4 simulations were carried out with activation of cumulus parameterization (Grell-Freitas) in both grid domains. The PBL parameterizations proposed in this work were Mellor-Yamada Nakanishi e Niino 2.5 (MYNN) and Yonsei Scheme University (YSU); therefore, in the discussion of the results, the statistics of the simulations of temperature at 2 meters and magnitude of the wind at 10 meters from the surface were evaluated. The height of the PBL for the study region and specifically the location of the Thermoelectric Power Plant were also simulated. Although the focus of this work was to study the model’s behavior in simulating the temperature and magnitude of the wind, it was important to understand the precipitation regime in the region, since it was the factor that led us to choose the two periods studied in this work. The model had difficulty in simulating the accumulated precipitation, in the two parameterizations of PBL, when simulated without the cumulus parameterization activated in the d02 grid; however, when the covection (Grell-Freitas) was activated, only the parameterization YSU correctly represented the accumulated precipitation according to the observations in the rainy season, while in the dry period the simulations underestimated the accumulated precipitation. In the temperature evaluation, the model had difficulty to reproduce the temperature values greater than 38 ºC in the dry period. In the rainy season, simulations performed with and without cumulus parameterization enabled showed that the model tended to represent the distribution of temperature values well, even though it underestimated the highest values and overestimated the lowest ones. With the wind simulation, the two parameterizations of PBL overestimated the higher values of V 10m and failed to present the distribution of values of V 10m according to the observations. Considering only the four automatic weather stations around the UTE, the model overestimated the simulation of the values of tm V10m in general, this due to the low capacity of the PBL schemes proposed in this work in reproducing very high wind values low. Therefore, it is concluded that the two parameterizations of PBL were more erroneous when simulating T2m and V 10m in the light wind regime. We can also infer that in general, the parameterization MYNN proved to be more efficient for the rainy season while YSU for the dry period when only the temperature at two meters is analyzed.