Prognoza vremenskih serija klimatskih elemenata i pokazatelja

master's thesis

• Title (english): Time Series Forecasting of Climate Elements and Indicators
• Author: Branimir Škrlec
• Mentor: Krešimir Pripužić (mentor)
• Committee member: Krešimir Pripužić (predsjednik povjerenstva)
• Committee member: Jurica Babić (član povjerenstva)
• Committee member: Marin Vuković (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2019-07-12, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering

Abstract

U sklopu ovog rada ukratko je objašnjeno što je točno vremenska serija te što su neuronske mreže. Pokazano je kakav utjecaj ima sjeverno atlantska oscilacija na vrijeme u Republici Hrvatskoj, odnosno postoji li korelacija između NAO indeksa i vrijednosti temperature, tlaka i padalina. Kreiran je model za predviđanje sutrašnjih vrijednosti pomoću LSTM neuronske mreže za 4 mjerne postaje u Republici Hrvatskoj: Bjelovar, Ogulin, Varaždin i Zagreb Maksimir. Pokazano je da kreirani model ima dobre karakteristike kod predviđanja dnevnih vrijednosti temperature i tlaka, dok model koji predviđa dnevnu količinu padalina ne daje dobre rezultate, te je zbog toga kreiran model koji koristi prosječne mjesečne vrijednosti.

Abstract (english)

Within this paper, it is briefly explained what are time series and neural networks. The influence of North Atlantic Oscillation on weather in Republic of Croatia has been shown, ie whether there is a correlation between the NAO index and temperature, pressure and precipitation. A model for prediction of next day's values was created with LSTM neural network for 4 measuring stations in the Republic of Croatia: Bjelovar, Ogulin, Varaždin and Zagreb Maksimir. It has been shown that the model created has good characteristics when predicting daily values of temperature and pressure, while a model predicting daily precipitation amounts does not produce good results and therefore a model that uses average monthly values has been created.

• Keywords: NAO; LSTM; klimatski elementi; vremenska serija
• Keywords (english): NAO; LSTM; climate elements; time series
• Language: croatian
• URN:NBN: urn:nbn:hr:168:079992
• Study programme: Title: Information and Communication Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije (magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2020-01-07 22:26:16