In order to assess the PROBA-V data products’ quality, comparisons with various reference datasets are performed. The validations are performed using similar or comparable land surface variables (e.g. surface reflectance and NDVI) that are derived from other satellite platforms, such as from the MODIS satellites. The product validation follows a standardised protocol, in which the following statistical metrics are calculated:
- Geometric mean (GM) regression: the geometric mean regression uses an orthogonal model to calculate the slope and intercept values and includes the errors of both satellite datasets.
- Accuracy: is calculated as the Mean Bias Error (MBE), and is a measure of the overall average difference. It preserves the sign of the difference.
Precision: is calculated as the standard deviation of the bias (MBE) and represents the dispersion of the product retrievals around their expected value.
Uncertainty: is calculated as the Root Mean Squared Difference (RMSD) and measures how far the difference between datasets deviates from 0. It is an expression of overall difference, including random and systematic differences.
In the product evaluations, the following aspects are examined:
The completeness of a dataset is assessed in a spatial and temporal extent. Data can be incomplete due to for example bad radiometric quality in one of the four PROBA-V bands.
The difference magnitude is presented as difference histograms between the examined and reference dataset, as well as global and/or regional scatterplots.
Spatial consistency refers to the realism and repeatability of the spatial distribution of retrievals over the globe, including the absence of artefacts (e.g., missing data, stripes, unrealistic values, etc.), based on expert knowledge.
The realism of temporal variations of the product are qualitatively assessed for point locations, well distributed over the globe. Spatio-temporal evolution of validation metrics is assessed through Hovmöller plots.
Reports Collection 2
- PROBA-V Collection 2 Change Document: In 2022, the entire PROBA-V archive (spanning from October 2013 – present) was re-processed to Collection 2. This document summarizes the major and minor modifications to algorithms, data, and metadata.
- PROBA-V Collection 2 Evaluation: This report provides a comparison of the reprocessed PROBA-V data (Collection 2) with the previous version of the data archive (Collection 1). The comparison was carried out on S1, S5 and S10 surface reflectance and NDVI data over the entire Collection 2 and Collection 1 data archives for the three resolutions. The evaluation focuses on; (i) quantification of the effect of the reprocessing by comparing C2 to C1; (ii) evaluation of the statistical, spatial and spatio-temporal consistency with SPOT-VGT C3 and Sentinel-3 VGT-SYN products, and (iii) evaluation of the statistical, spatial and spatio-temporal consistency of the PROBA-V archive through comparison to METOP/AVHRR and MODIS. The NDVI profiles for all LANDVAL sites of all datasets used in the evaluation can be found in the digital annex (NDVI_profiles.001.zip - NDVI_profiles.002.zip).
- PROBA-V Collection 2 Atmospheric Correction Validation: This report provides the validation of the new atmospheric correction for PROBA-V Collection 2 using the ACIX approach for validation.
- PROBA-V Collection 2 Cloud Masking Validation Report for 1 km, 300 m and 100 m data: These reports provide the validation results of the new cloud detection method for the different resolutions.
Reports Collection 1
- PROBA-V Collection 1 Change Summary: 2016, the entire PROBA-V archive (spanning from October 2013 - present) was re-processed to Collection 1. This document summarizes the major and minor modifications to algorithms, data and metadata.
- PROBA-V Collection 1 Cloud Mask Evaluation: This report documents the validation of the cloud mask method for PROBA-V Collection 1.
- Quality Note - Extending the SPOT/VEGETATION-PROBA-V Collection 1 archive with Sentinel-3 SYN-VGT products: The Sentinel-3 Ocean and Land Colour Instrument (OLCI), in synergy with the Sea and Land Surface Temperature Radiometer (SLSTR) instrument, is foreseen to provide continuity to the SPOT/VGT capability (Donlon et al., 2012). The ESA Sentinel-3 synergy (SYN_VGT) products replicate the attributes and quality of standard 1 km SPOT/VGT products through innovative spectral remapping and co-location techniques (North and Heckel, 2010).
- Comparison SPOT-VGT C3 - PROBA-V C1: A short note reporting on the comparison between SPOT-VGT and PROBA-V surface reflectances and NDVI for data before and after the re-processings. The comparison was performed on S10 composites for the overlapping observational period (November 2013 - May 2014).
- SPOT-VGT Collection 3 Evaluation: The entire SPOT-VGT data archive (21 April 1998 - 31 May 2014) was reprocessed in 2015 - 2016. The paper describes the comparison S10 surface reflectance and NDVI from the re-processed Collection 3 with the previous Collection 2, as well as a consistency analysis between Collection 3 VGT1 and VGT2 data. Further, the Collection 3 data were compared with external satellite reference data (MODIS and METOP-AVHRR).