Korekcija predikcija meteoroloških parametara metodama strojnog učenja

master's thesis

• Title (english): Correcting the predictions of weather parameters by machine learning techniques
• Author: Andrej Pustak
• Mentor: Tomislav Capuder (mentor)
• Committee member: Tomislav Capuder (predsjednik povjerenstva)
• Committee member: Božidar Filipović-Grčić (član povjerenstva)
• Committee member: Viktor Milardić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-15, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovome diplomskom radu napravljene su korekcije predikcija numeričkog meteorološkog modela metodama učenja. Numeričkom modelu je potrebno nekoliko sati da obavi sve kalkulacije od početka pokretanja te mi u međuvremenu dobijemo prava mjerenja. Tu se javlja mogućnost usporedbe tih dobivenih predikcija i dobivenih mjerenja te korekcije ostatka predikcija. Metode strojnog učenje koje su korištene su regresija potpornim vektorima te duboke neuronske mreže. Korekcija je napravljena na pet vremenskih parametara: temperaturi, brzini i smjeru vjetra, sunčevoj radijaciji te relativnoj vlazi zraka. Za svaki od modela strojnog provedena je evaluacija indikatorima uspješnosti kao što su MSE, RMSE i R2 ocjena te su rezultati evaluacije modela strojnog uspoređeni sa evaluacijom predikcija numeričkog meteorološkog modela.

Abstract (english)

The goal of this thesis was to make and evaluate machine learning models whose purpose was correcting the predictions of numerical weather prediction model. Numerical weather prediction takes several hours to fully finish all calculations during which we get true measurements from meteorological stations. That means that when the NWP model releases the results, we have the measurements for the first several hours which we can use to correct the rest of the predictions if necessary. Machine learning methods that were used were support vector regression and deep neural networks. The correction was made for five weather parameters: temperature, wind speed and direction, radiation, and relative humidity. Each of the machine learning models was evaluated using performance indicators like MSE, RMSE and R2 score.

• Keywords: korekcija; predikcija; numerički; meteorološki; model; strojno; učenje; temperatura; vjetar; radijacija; vlaga
• Keywords (english): correction; prediction; forecast; numerical; weather; prediction; machine; learning
• Language: croatian
• URN:NBN: urn:nbn:hr:168:495969
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-14 09:46:01