Deutsch
 
Datenschutzhinweis Impressum
  DetailsucheBrowse

Datensatz

DATENSATZ AKTIONENEXPORT

Freigegeben

Zeitschriftenartikel

Improving atmospheric angular momentum forecasts by machine learning

Urheber*innen
/persons/resource/dill

Dill,  R.
1.3 Earth System Modelling, 1.0 Geodesy, Departments, GFZ Publication Database, Deutsches GeoForschungsZentrum;

/persons/resource/saynisch

Saynisch-Wagner,  J.
1.3 Earth System Modelling, 1.0 Geodesy, Departments, GFZ Publication Database, Deutsches GeoForschungsZentrum;

/persons/resource/irrgang

Irrgang,  C.
1.3 Earth System Modelling, 1.0 Geodesy, Departments, GFZ Publication Database, Deutsches GeoForschungsZentrum;

/persons/resource/mthomas

Thomas,  M.
1.3 Earth System Modelling, 1.0 Geodesy, Departments, GFZ Publication Database, Deutsches GeoForschungsZentrum;

Externe Ressourcen
Es sind keine externen Ressourcen hinterlegt
Volltexte (frei zugänglich)

5008319.pdf
(Verlagsversion), 2MB

Ergänzendes Material (frei zugänglich)
Es sind keine frei zugänglichen Ergänzenden Materialien verfügbar
Zitation

Dill, R., Saynisch-Wagner, J., Irrgang, C., Thomas, M. (2021): Improving atmospheric angular momentum forecasts by machine learning. - Earth and Space Science, 8, 12, e2021EA002070.
https://doi.org/10.1029/2021EA002070


Zitierlink: https://gfzpublic.gfz-potsdam.de/pubman/item/item_5008319
Zusammenfassung
Earth angular momentum forecasts are naturally accompanied by forecast errors that typically grow with increasing forecast length. In contrast to this behavior, we have detected large quasi-periodic deviations between atmospheric angular momentum wind term forecasts and their subsequently available analysis. The respective errors are not random and have some hard to define yet clearly visible characteristics which may help to separate them from the true forecast information. These kinds of problems, which should be automated but involve some adaptation and decision-making in the process, are most suitable for machine learning methods. Consequently, we propose and apply a neural network to the task of removing the detected artificial forecast errors. We found, that a cascading forward neural network model performed best in this problem. A total error reduction with respect to the unaltered forecasts amounts to about 30% integrated over a 6 day forecast period. Integrated over the initial 3 day forecast period, in which the largest artificial errors are present, the improvements amount to about 50%. After the application of the neural network, the remaining error distribution shows the expected growth with forecast length. However, a 24 hourly modulation and an initial baseline error of 2*10−8 became evident that were hidden before under the larger forecast error.