Comparison with Sentinel-2, PlanetScope, SPOT, and Pleiades - a preliminary quality assessment
This report addresses SPOT TrueSharp processed imagery and aims to assess whether such imagery can be considered another regulatory option (Article 40a of the implementing regulation (EU) 746/2018 of 18 May 2018 amending the Implementing Regulation (EU) No. 809/2014) for data requiring qualities "equivalent or better than Sentinel". It is of interest to conclude whether SPOT TrueSharp imagery is comparable and can be used equally as the native HHR data with an inherent better Ground Sample Distance (range between 2.5-5.0 m).
The analysis includes visual and radiometric quality assessment of the SPOT TrueSharp product by comparison with available Sentinel-2, PlanetScope Classic Dove, SPOT 6/7, and Pleiades 1A/1B data sets. The test was limited to a single acquisition date (or the closest possible date) for two restricted geographic areas (Italy and Belgium). The study sites were selected to have “small” or elongated agricultural parcels where SPOT and Pleiades imagery were made available by Airbus, and where JRC has ground truth data.
Visual comparison of the extracts of true colour rendering or nIR false colour composites does not show the radiometric difference between TrueSharp and the standard SPOT product, or the difference is very small. TrueSharp also shows comparable or better radiometry cf. Planet for visual interpretation. The exercise performed by the JRC, also shows that there is clear visible enhancement of the spatial resolution in the TrueSharp 4m product compared to Planet (and obviously S2). This brings on the conclusion from the visual comparison that the product (TrueSharp) makes the visual interpretation just as feasible as with the other products.
Comparison of the radiometric Bottom of Atmosphere (or surface reflectance) values of TrueSharp with the other products (S2 and PlanetScope) shows that the NDVI values of S2, and TrueSharp 4m are very much aligned, while Planet most often was found lower (‘damped’). Also, in the resulting linear regression the best NDVI correlation was found in the TrueSharp vs. S2 components, which confirms the visual result. However, all regressions have high R2 values which brings the conclusion that all three products can supplement each other. This is confirmed also in the application of Pearson’s correlation coefficient, when used as an estimator of similarity, where results tell us that the NDVI profiles for the three sensor products are comparable. Of interest, in the last JRC test, is also the heterogeneity test which tells that heterogeneity is not more evident in either of the HHR data sets (Planet, TrueSharp) compared to S2. But such tests need to be done on a larger population of parcels, and smaller ones.
Regards to determining whether the HHR Time Stacks specification can be revised permitting the use of SPOT TrueSharp equally as the source HHR data with an inherent better Ground Sample Distance (GSD), the answer is as regards spatial resolution, and radiometric BOA reflectance yes, however to conclude on this and the real complementarity to the others, of the TrueSharp product, a test of the time series frequency should be made.
The report concludes discussing the need of HHR imagery within the frame of the new CAP. It suggests that the Time Stacks could be of use in regions with almost exclusively small arable land parcels. Other hypothetical use cases could lie in the potential for the discrimination of landscape elements not discernible on Sentinel data, or to use, even if our small test showed that the HHR products did not make evident more heterogeneity, the intra-parcel heterogeneity of these HHR products to conclude on parcel maintenance activities or abandonment. In both these latter cases, however, there is probably competition from more cost-effective and accurate solutions. Other fields of research may arise depending on the detailed settings currently under discussion for the new CAP.
AASTRAND Paer;
LEMAJIC Slavko;
WIRNHARDT Csaba;
2021-06-17
Publications Office of the European Union
JRC124413
978-92-76-38677-3 (online),
1831-9424 (online),
EUR 30737 EN,
OP KJ-NA-30737-EN-N (online),
https://publications.jrc.ec.europa.eu/repository/handle/JRC124413,
10.2760/73137 (online),
