Frequency: 10 days
Data can be accessed here.
Algorithm description: The algorithm is similar to GEOv2 algorithm – (1) Estimation (neural networks) of instantaneous biophysical variables atmospherically corrected surface reflectance and then (2) smoothing and gap-filling.
The main advantage is finer resolution than GEOv2. Note that the training datasets are still the once developed for SPOT-VGT, hence PROBA-V reflectance is converted to VGT.
Accuracy: Early validation conducted over sites in Europe for year 2014. It resulted in RMSE=0.1, bias=0.043. Actual ground based validation over two sites in Spain showed overall good performance. The recent (2017) products are consistent with earlier version (2014).
- Validation conducted over Europe (not global)
- Large fraction of missing values in northern latitudes during wintertime
- Larger values than earlier versions over broadleaf forest by the end of summertime
- LAI accuracy depends on site
- Positive bias for bare or harvested areas
Baret, F., M. Weiss, A. Verger, and B. Smets. “ImagineS ATBD for LAI, FAPAR and FCOVER from PROBA-V Products at 300m Reolution (GEOV3).” INRA, May 31, 2016.
Camacho, F., J. Sanchez, and C. Latorre. “GIOGL1 Quality Assessment Report LAI, FAPAR, FCover Collection 300m Version 1.” EOLAB, September 20, 2016.
Camacho, Fernando, Jorge Sánchez, Roselyne Lacaze, Marie Weiss, and Frédéric Baret. “Validating GEOV3 LAI, FAPAR and Vegetation Cover Estimates Derived from PROBA-V Observations at 333m over Europe.” Geophysical Research Abstracts 18 (2016).