Title Wind speed prediction using the analog method over complex topography
Title (croatian) Prognoza brzine vjetra upotrebom metode analogona nad složenom topografijom
Author Iris Odak Plenković
Mentor Kristian Horvath (mentor)
Committee member Maja Telišman-Prtenjak (predsjednik povjerenstva)
Committee member Kristian Horvath (član povjerenstva)
Committee member Željko Večenaj (član povjerenstva)
Granter University of Zagreb Faculty of Science (Department of Geophysics) Zagreb
Defense date and country 2020-07-09, Croatia
Scientific / art field, discipline and subdiscipline NATURAL SCIENCES Geophysics Meteorology and Climatology
Universal decimal classification (UDC ) 53 - Physics
Abstract The performance of the analog-based post-processing method is tested in climatologically and topographically different regions, for point-based wind speed predictions at 10 m above the ground, and compared to the baseline Kalman filter (KF) model. This research shows that the deterministic analog-based predictions produced using deterministic numerical weather prediction (NWP) model output improve the correlation between predictions and measurements while reducing the forecast error compared to the starting model predictions regardless of the terrain complexity. While the KF based approach generally outperforms the analog-based predictions in the bias reduction, the combination of the KF and analog approach can be similarly successful. In the coastal complex area, characterized by a larger frequency of high wind speed, the analog-based predictions are more successful in reducing the dispersion error than the KF. The application of the KF algorithm to the analogs in the so-called analog space (KFAS) is the least prone to the standard deviation underestimation among the analog-based predictions. All analog-based predictions improve prediction of larger than diurnal motions while the KFAS is superior among all analog-based predictions in predicting alternating wind regimes on the time scales shorter than a day. The analog-based predictions better distinguish different wind speed categories in the coastal complex topography by using a higher-resolution model input. The analog method is also applied to the ensemble NWP. Evaluation of several configurations using various predictor variables is conducted through a set of sensitivity experiments. The results are compared to the ensemble model output statistic (EMOS) baseline model. Results show that both analog-based and EMOS experiments considerably improve the raw model forecast. The analog-based predictions are overall comparable to or even outperform the EMOS. Assessing the post-processing performance for high wind speeds, it is shown that the analog experiments can improve the raw forecast, exhibiting significantly higher skill than the EMOS. The processes at lower altitude stations seem to be better represented by the raw model, which leads to better input forecast to the post-processing and better overall result than for the mountain stations. Generally, the difference between several analog-based experiments is less pronounced. Furthermore, it is demonstrated that the usage of summarized ensemble measures is an optimal way to improve the forecast skill, compared to the other analog-based experiments.
Abstract (english) Metoda analogona, koja se koristi za naknadnu obradu produkata numeričkog modela, testirana je za prognoze vjetra na 10 m iznad tla na lokacijama koje pripadaju topografski i klimatološki različitim područjima te uspoređena s metodom koja koristi Kalmanov filtar (KF). Deterministički produkt metode analogona ima veću koreliranost prognoze i mjerenja te manju pogrešku u odnosu na numerički model koji metoda koristi kao ulazni podatak, neovisno o složenosti topografije. Metoda naknadne obrade KF iznimno je uspješna u uklanjanju pristranosti prognoze. Kombinacija metode analogona i KF gotovo je jednako uspješna u uklanjanju pristranosti, pri čemu pokazuje i dodatne prednosti svojstvene metodi analogona. U obalnom području, karakteriziranom kompleksnom topografijom i učestalim jakim vjetrom, metoda analogona uspješnija je od KF u uklanjanju pogreške disperzije. Dodatno, primjena Kalmanovog filtra u takozvanom prostoru analogona (KFAS) je eksperiment koji je najmanje podložan podcjenjivanju prirodne varijabilnosti vjetra, mjereno standardnom devijacijom. Svi eksperimenti koji koriste analogije poboljšavaju prognoze na vremenskim skalama duljima od jednog dana. Međutim, na skalama kraćima od jednog dana je KFAS najuspješniji eksperiment. Korištenje modela veće rezolucije kao ulazni podatak za metodu analogona doprinosi da prognoza lakše razlikuje kategorije vjetra. Metoda analogona primijenjena je i na ansambl prognozu numeričkog modela. Pritom je testirano nekoliko različitih konfiguracija metode kroz testove osjetljivosti. Eksperimenti se prvenstveno razlikuju po ulaznim parametrima, tj. po načinu korištenja informacija iz početne ansambl prognoze modela. Rezultati metode analogona uspoređeni su s metodom naknadne obrade koja je bazirana na statistici simuliranih podataka za ansambl prognoze (EMOS). Obje testirane metode naknadne obrade vidno poboljšavaju prognozu ulaznog modela. Pritom je metoda analogona usporediva s metodom EMOS, ili čak i bolja. Dodatno, metoda analogona ostvaruje signifikantno bolji rezultat za prognozu jakog vjetra od početnog modela te metode EMOS. U numeričkom modelu procesi su bolje razlučeni za lokacije smještene na nižoj nadmorskoj visini nego za planinske lokacije. Posljedično, to znači i bolji rezultat nakon naknadne obrade produkata modela te bolji ukupan rezultat za lokacije nižih nadmorskih visina. Općenito, razlika među eksperimentima s različitim konfiguracijama metode analogona manje je izražena. Štoviše, pokazano je da je upravo korištenje sažetih informacija o prognozi ulaznog modela optimalan način da se poboljša točnost prognoze.
Keywords
analog-ensemble forecast
complex topography
ensemble model output statistics
Kalman-filter
mesoscale model
statistical post-processing
wind ensemble forecast
Keywords (english)
EMOS
Kalmanov filtar
kompleksna topografija
mezoskalni model
metoda analogona
statističke metode naknadne obrade
ansambl prognoza vjetra
Language english
URN:NBN urn:nbn:hr:217:103543
Promotion 2020
Study programme Title: Physics Study programme type: university Study level: postgraduate Academic / professional title: doktor/doktorica znanosti, područje prirodnih znanosti, polje fizika (doktor/doktorica znanosti, područje prirodnih znanosti, polje fizika)
Type of resource Text
File origin Born digital
Access conditions Open access
Terms of use
Created on 2020-07-27 13:25:59