|
|
New parameterizations in the RegCM model
Pitaš, Ondřej ; Belda, Michal (advisor) ; Karlický, Jan (referee)
This thesis compares new physical parameterizations in the regional climate model system RegCM with the standard settings used at the Department of Atmospheric Phy- sics and observed climatic conditions. The first part is dedicated to the description of individual parameterizations and to the essential introduction to the RegCM model. In addition, model preprocessing and running simulations are discussed. The simulations were conducted over Europe domain for the ten-year period (1990-1999) with five dif- ferent model settings. Holtslag PBL(Holtslag, 1990) type boundary layer scheme, Grell cumulus convention scheme over land and ocean and Explicit moisture scheme (SUBEX; Pal et al 2000) were used in the reference simulation. In the conducted simulations, the following physical parametrizations were used separately: UW PBL (Bretherton and McCaa, 2004) boundary layer scheme, Explicit moisture Nogherotto/Tompkins mois- ture scheme, Tiedtke (1996) and Kain-Fritsch (1990) & Kain (2004) cumulus convention schemes over land and ocean. None of the new physical parameterizations showed impro- vements over standard setting in all analyzed variables at the same time. Precipitation is best described by reference simulation as well as by simulations with UW PBL and Kain- Fritsch & Kain parametrization type. Kain-Fritsch...
|
|
|
Overview and using of statistical methods for short-term forecasts and nowcasting evaluation
Pitaš, Ondřej ; Žák, Michal (advisor) ; Šaur, David (referee)
Evaluating the success of the weather forecast is a fundamental factor in improving their quality. It is necessary to use statistical methods for this evaluation. However, the existence of a large number of statistical methods complicates their choice and com- paring the results of different studies. To bridge this knowledge problem, we dealt with statistical methods suitable for evaluating accuracy of short-range weather forecasts of deterministic variables. Thesis is dedicated to utilization of statistical methods for veri- fying radar echo nowcasting and for forecasts of numerical weather prediction models. In addition, it also deals with the use of statistical methods for defining loss functions and their use for learning the radar echo nowcasting model. Learned nowcasting models were verified. Numerical weather prediction models ICON and WRF were included to the work. Forecasts were verified with data from synoptic stations and ERA5 reanalysis in the do- main of the Czech Republic. The forecasts of the radar echo of the nowcasting models pySTEPS, rainymotion and MWNet v1.0 were verified with radar images in the domain of the Czech Republic. In most cases, the statistical methods agreed on the results of the evaluation of the weather forecast by the numerical weather prediction models. The differences...
|
|
|
New parameterizations in the RegCM model
Pitaš, Ondřej ; Belda, Michal (advisor) ; Karlický, Jan (referee)
This thesis compares new physical parameterizations in the regional climate model system RegCM with the standard settings used at the Department of Atmospheric Phy- sics and observed climatic conditions. The first part is dedicated to the description of individual parameterizations and to the essential introduction to the RegCM model. In addition, model preprocessing and running simulations are discussed. The simulations were conducted over Europe domain for the ten-year period (1990-1999) with five dif- ferent model settings. Holtslag PBL(Holtslag, 1990) type boundary layer scheme, Grell cumulus convention scheme over land and ocean and Explicit moisture scheme (SUBEX; Pal et al 2000) were used in the reference simulation. In the conducted simulations, the following physical parametrizations were used separately: UW PBL (Bretherton and McCaa, 2004) boundary layer scheme, Explicit moisture Nogherotto/Tompkins mois- ture scheme, Tiedtke (1996) and Kain-Fritsch (1990) & Kain (2004) cumulus convention schemes over land and ocean. None of the new physical parameterizations showed impro- vements over standard setting in all analyzed variables at the same time. Precipitation is best described by reference simulation as well as by simulations with UW PBL and Kain- Fritsch & Kain parametrization type. Kain-Fritsch...
|
guest ::
