• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 1
  • 1
  • Tagged with
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 1
  • 1
  • 1
  • 1
  • 1
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

Jämförelse av korta temperaturprognoser från SMHI och Meteorologisk institutt med fokus på post-processingmetodikens betydelse för prognoskvaliteten / Comparison of Short-Range Temperature Forecasts from SMHI and the Norwegian Meteorological Institute - Focus on the Importance of Post-Processing Methods for the Quality of the Forecasts

Petersson, Sofie January 2019 (has links)
Temperaturprognoser är av stor betydelse för många i dagens samhälle, både privatpersoner och diverse olika sektorer. Förväntan på att prognoserna håller hög träffsäkerhet är stor och god kvalitet på dessa är viktigt av många olika aspekter. De numeriska vädermodellerna, som används för att göra väderprognoser, har brister som i stort sätt alltid leder till systematiska fel i prognoserna. Bristerna beror exempelvis på dålig representation av atmosfärens fysikaliska processer och för att korrigera och reducera dessa fel efterbehandlas prognoserna med olika metoder, så kallad post-processing. För att minimera de systematiska felen och öka träffsäkerheten för prognoserna pågår ständigt en utveckling och förbättring av både modellerna och post-processingmetodiken. Uppföljning och utvärdering av prognoser är av stor nytta för denna utveckling som ska leda till minimering av prognosfel och optimering av modell och metodik. I denna studie har temperaturprognosdata, med prognoslängd 0-12 timmar, från Sveriges Meteorologiska och Hydrologiska Institut (SMHI) och norska Meteorologisk institutt (met.no) jämförts med uppmätta värden för 2 m-temperatur. Observerad temperaturdata från 22 olika synoptiska väderstationer på platser utspridda över hela Sverige har använts i studien och perioden som studien är baserad på är 20 februari till 31 maj 2018. Statistiska mått, med mest fokus på korrelationskoefficient och bias, har analyserats och jämförts för att undersöka likheter och skillnader i temperaturprognoserna från de två olika väderinstituten. Resultaten av studien visar att temperaturprognoserna från met.no generellt sett har något högre träffsäkerhet än SMHI:s för de allra flesta av de 22 geografiska platserna. Båda institutens prognoser har för flertalet av stationerna i fjällen samt norra Sverige generellt sett lägre träffsäkerhet för februari än för mars, april och maj. / Temperature forecasts are of great importance for many different reasons in today's society, both for private individuals and various sectors. The expectations that the forecasts maintain high accuracy and good quality is important in many different aspects. The weather models, which are used to make the forecasts, have deficiencies which in large part always lead to systematic errors in the forecasts. The deficiencies are for example, due to poor representation of the physical processes of the atmosphere and to correct and reduce these errors, the forecasts are post-processed by various methods. To minimize the systematic errors and increase the accuracy of the forecasts, there is an ongoing development and improvement of both the models and the post-processing methods. Evaluation of forecasts is of great benefit to this development, which will lead to minimization of forecast errors and optimization of the model and methodology. In this study, temperature forecast data, with a forecast length of 0-12 hours, from the Swedish Meteorological and Hydrological Institute (SMHI) and the Norwegian Meteorological Institute (met.no) were compared with measured 2 m-temperature values. Observed temperature data from 22 different weather stations in locations scattered all over Sweden have been used in the study and the period on which the study is based is from the 20th of February to 31st of May, 2018. Different statistical measures have been analyzed and compared to examine similarities and differences in temperature forecasts from the two different weather institutes. The results of the study show that met.no's temperature forecasts generally have slightly higher accuracy than SMHI's for most of the 22 locations. For any of the stations in the mountains and northern Sweden forecasts from both institutes generally have lower accuracy for February than March, April and May.
2

Reduction of Temperature Forecast Errors with Deep Neural Networks / Reducering av temperaturprognosfel med djupa neuronnätverk

Isaksson, Robin January 2018 (has links)
Deep artificial neural networks is a type of machine learning which can be used to find and utilize patterns in data. One of their many applications is as method for regression analysis. In this thesis deep artificial neural networks were implemented in the application of estimating the error of surface temperature forecasts as produced by a numerical weather prediction model. An ability to estimate the error of forecasts is synonymous with the ability to reduce forecast errors as the estimated error can be offset from the actual forecast. Six years of forecast data from the period 2010--2015 produced by the European Centre for Medium-Range Weather Forecasts' (ECMWF) numerical weather prediction model together with data from fourteen meteorological observational stations were used to train and evaluate error-predicting deep neural networks. The neural networks were able to reduce the forecast errors for all the locations that were tested to a varying extent. The largest reduction in error was by 83.0\% of the original error or a 16.7\degcs decrease in the mean-square error. The performance of the neural networks' error reduction ability was compared with that of a contemporary Kalman filter as implemented by the Swedish Meteorological and Hydrological Institute (SMHI). It was shown that the neural network implementation had superior performance for six out of seven of the evaluated stations where the Kalman filter had marginally better performance at one station.

Page generated in 0.0888 seconds