BIBLIOS

   Ciências References Management System

Visitor Mode (Login)
Need help?


Back

Publication details

Document type
Book chapters


Title
Machine Learn Estimates of Downward Surface Long-wave Fluxes (DSLF) Based on Reanalysis and Satellite Observations.

Participants in the publication
Lopes, F.M. (Author)
Dutra, E. (Author)
Trigo, I. (Author)

Summary
A machine learning approach based on multivariate adaptive regression splines (MARS) is explored to integrate reanalysis data, satellite cloud information and ground observations of Downward Surface Long-wave Radiation Fluxes (DSLF), to estimate hourly DSLF for all-sky conditions. The MARS estimates are shown to have lower errors than other models when tested against 23 stations (BSRN/ARM), outperforming other DSLF estimates, including the current LSA-SAF operational roduct. In this work, the proposed methodology is shown to be consistent when new validation is performed, particularly with an independent network of 52 stations (FLUXNET2015). Further assessment of MARS estimates for the whole MSG disk is carried out, showing the potential of the model for operational purposes.

Date of Submisson/Request
2022-04-18
Date of Acceptance
2022-05-10
Date of Publication
2022-06-27

Institution
FACULDADE DE CIÊNCIAS DA UNIVERSIDADE DE LISBOA

Where published
CITAB

Publication Identifiers
ISBN - 9789897045288

Edition
1

Document Identifiers
ISBN - 978-989-704-528-8

Download

Export

APA
Lopes, F.M., Dutra, E., Trigo, I., (2022). Machine Learn Estimates of Downward Surface Long-wave Fluxes (DSLF) Based on Reanalysis and Satellite Observations.. CITAB, -

IEEE
Lopes, F.M., Dutra, E., Trigo, I., "Machine Learn Estimates of Downward Surface Long-wave Fluxes (DSLF) Based on Reanalysis and Satellite Observations." in CITAB, 2022, pp. -

BIBTEX
@incollection{59459, author = {Lopes, F.M. and Dutra, E. and Trigo, I.}, title = {Machine Learn Estimates of Downward Surface Long-wave Fluxes (DSLF) Based on Reanalysis and Satellite Observations.}, booktitle = {CITAB}, year = 2022, pages = {-}, address = {}, publisher = {} }