Zur Seitenansicht


Probabilistic spatial forecasting of daily precipitation sums over complex terrain / by Reto Stauffer
VerfasserStauffer, Reto
Begutachter / BegutachterinnenRotach, Mathias
Betreuer / BetreuerinnenMayr, Georg J. ; Zeileis, Achim
ErschienenInnsbruck, July 2016
Umfangvi, 109 Seiten : Illustrationen
HochschulschriftUniversität Innsbruck, Dissertation, 2016
Datum der AbgabeJuli 2016
Schlagwörter (DE)Niederschlag / Niederschlagsklimatologie / Numerische Wettervorhersagen / Statistisches Post-Processing / Zensiert / Zensierte Verteilung / Generalisierte Additive Modelle / Farbskalen / Farbe / komplexe Topographie / Tirol
Schlagwörter (EN)Precipitation / Precipitation Climatology / Statistical Post-Processing / Censoring / Censored Distribution / Generalized Additive Models / Color Maps / Color / Complex Topography / Tyrol
Schlagwörter (GND)Tirol / Niederschlag / Numerische Wettervorhersage / Stochastisches Modell
URNurn:nbn:at:at-ubi:1-5281 Persistent Identifier (URN)
 Das Werk ist frei verfügbar
Probabilistic spatial forecasting of daily precipitation sums over complex terrain [8.76 mb]
Zusammenfassung (Deutsch)

The demand of accurate and reliable spatial probabilistic precipitation forecasts rapidly grew over the past decade. Weather forecasts are typically provided by numerical ensemble forecasting systems which are often underdispersive and biased due to necessary simplifications and imperfect model assumptions. This is especially true over complex terrain as small-scale features cannot yet be resolved by the numerical models. To alleviate the errors and to account for the large spatial variability of precipitation, the standardized anomaly model output statistics (SAMOS) approach is extended to account for the properties of daily precipitation sums. SAMOS uses high-resolution spatial climatology as background information to preserve the local characteristics. In the first part, a novel spatio-temporal precipitation climatology is proposed using a generalized additive regression model for location, scale, and shape (GAMLSS) with a left-censored response distribution. These climatological estimates are used to compute the standardized anomalies for the full-distributional spatial SAMOS post-processing method presented in the second part of this thesis. The new approach directly provides full-distributional corrected ensemble forecasts and allows to derive all kind of quantities, such as the probability of precipitation, probabilities of exceeding a certain threshold, quantiles, or expectations on a very high spatial resolution. To communicate such information, colorized visualizations are used frequently. However, applying inappropriate color maps can rapidly distort the information and create misleading artifacts. The third part introduces a perception-based color space called huechromaluminance (HCL) and shows guidelines and examples on how to effectively use color maps for meteorological visualizations.

Das PDF-Dokument wurde 71 mal heruntergeladen.