Use of medium-resolution imagery in the Belgian crop growth monitoring system (B-CGMS).

peer reviewed ; The Belgian Crop Growth Monitoring System (B-CGMS) uses the 1km²-resolution imagery of NOAA-AVHRR and SPOT-VEGETATION to improve its yield estimates. The pre-processed images are converted to fAPAR and combined with meteorological data (irradiance, temperature) to daily growth values by means of the Monteith approach. The regional means of the cumulative monthly growth numbers are calibrated against official harvest statistics by means of crop-specific neural networks. The quality of the yield estimates varies with the importance of the crop: the R²-values for winter wheat, sug... Mehr ...

Verfasser: Eerens, Herman
Wouters, Katy
Dehem, Didier
Tychon, Bernard
Buffet, D.
Oger, Robert
Dokumenttyp: conference paper
Erscheinungsdatum: 2000
Verlag/Hrsg.: Joint research Centre
Schlagwörter: B-CGMS / SPOT-VEGETATION / fAPAR / Crop yield forecast / Life sciences / Agriculture & agronomy / Sciences du vivant / Agriculture & agronomie
Sprache: Englisch
Permalink: https://search.fid-benelux.de/Record/base-26514487
Datenquelle: BASE; Originalkatalog
Powered By: BASE
Link(s) : https://orbi.uliege.be/handle/2268/94321

peer reviewed ; The Belgian Crop Growth Monitoring System (B-CGMS) uses the 1km²-resolution imagery of NOAA-AVHRR and SPOT-VEGETATION to improve its yield estimates. The pre-processed images are converted to fAPAR and combined with meteorological data (irradiance, temperature) to daily growth values by means of the Monteith approach. The regional means of the cumulative monthly growth numbers are calibrated against official harvest statistics by means of crop-specific neural networks. The quality of the yield estimates varies with the importance of the crop: the R²-values for winter wheat, sugar-beets and fodder maize are respectively 60%, 48% and 36%. The final yield forecasts will be better, because the B-CGMS integrates the image-based yield estimates with the assessments of the agromet-model and of the technological trend function. The technique of linear unmixing seems promising but in its actual state it is too unpredictable to be included in an operational scheme.