Evaluation of PROBA-V Collection 1: Refined Radiometry, Geometry, and Cloud Screening

Abstract

PROBA-V (PRoject for On-Board Autonomy-Vegetation) was launched in May-2013 as an operational continuation to the vegetation (VGT) instruments on-board the Système Pour l'Observation de la Terre (SPOT)-4 and -5 satellites. The first reprocessing campaign of the PROBA-V archive from Collection 0 (C0) to Collection 1 (C1) aims at harmonizing the time series, thanks to improved radiometric and geometric calibration and cloud detection. The evaluation of PROBA-V C1 focuses on (i) qualitative and quantitative assessment of the new cloud detection scheme; (ii) quantification of the effect of the reprocessing by comparing C1 to C0; and (iii) evaluation of the spatio-temporal stability of the combined SPOT/VGT and PROBA-V archive through comparison to METOP/advanced very high resolution radiometer (AVHRR). The PROBA-V C1 cloud detection algorithm yields an overall accuracy of 89.0%. Clouds are detected with very few omission errors, but there is an overdetection of clouds over bright surfaces. Stepwise updates to the visible and near infrared (VNIR) absolute calibration in C0 and the application of degradation models to the SWIR calibration in C1 result in sudden changes between C0 and C1 Blue, Red, and NIR TOC reflectance in the first year, and more gradual differences for short-wave infrared (SWIR). Other changes result in some bias between C0 and C1, although the root mean squared difference (RMSD) remains well below 1% for top-of-canopy (TOC) reflectance and below 0.02 for the normalized difference vegetation index (NDVI). Comparison to METOP/AVHRR shows that the recent reprocessing campaigns on SPOT/VGT and PROBA-V have resulted in a more stable combined time series.are widely used to monitor environmental change and the evolution of vegetation cover in different thematic domains [2], hence the relevance to the continuity of the service.On board the PROBA-V, the optical instrument provides data at 1 km, 300 m, and 100 m consistent with the 1 km resolution data products of SPOT/VGT [3]. To support the existing SPOT/VGT user community, the PROBA-V mission continues provision of projected segments (Level 2A, similar to the SPOT/VGT P-products), daily top-of-canopy (TOC) synthesis (S1-TOC) and 10-day synthesis (S10-TOC) products. In addition, top-of-atmosphere (TOA) daily synthesis (S1-TOA) products and radiometrically/geometrically corrected data (level 1C) products in raw resolution (up to 100 m) are provided for scientific use [4,5]. Since PROBA-V has no onboard propellant, the overpass time (10:45 h at launch) will decrease as a result of increasing atmospheric drag [6].In previous years, vegetation monitoring applications were built on PROBA-V data, e.g., on cropland mapping [7], crop identification [8], estimation of biophysical parameters [9,10], and crop yield forecasting [11]. A method was also developed to assimilate data of PROBA-V 100 and 300 m [12]. For the Copernicus Global Land Service, PROBA-V is one of the prime sensors for operational vegetation products [13,14]. Access to and exploitation of the SPOT/...

Radiometric Top-of-Atmosphere Reflectance Consistency Assessment for Landsat 8/OLI, Sentinel-2/MSI, PROBA-V, and DEIMOS-1 over Libya-4 and RadCalNet Calibration Sites

“…Due to the lack of propellant onboard PROBA-V, the overpass time (10:45 at launch) decreases as a result of increasing atmospheric drag [8] and was 09:46 as of 24 September 2019. Due to the absence of on-board calibration devices, the radiometric calibration and stability monitoring of the PROBA-V instrument relies on vicarious and lunar calibration [9,10].…”

Cross-Sensor Adversarial Domain Adaptation of Landsat-8 and Proba-V Images for Cloud Detection

“…Nevertheless, an small gap in performance is observed between results in the source domain (Plan-etScope) and the target domain (Sentinel-2). Finally, in our previous work [5], we showed that transfer learning from Proba-V to Landsat-8 and from Landsat-8 to Proba-V produce accurate results on a par with the FMask [22] model on Landsat-8 and surpassing the operational cloud detection model [27] for Proba-V, respectively. However, a significant gap between transfer learning approaches and state-of-the-art models trained with data from the same domain still exists, which is the focus of the present work.…”

“…We also included the results of the ablation study, where we have set some of the weights of the generator losses to zero and the results using histogram matching [36] for domain adaptation as in [47]. In addition, results are compared with the FCNN trained in original Proba-V images and ground truths (PV-trained), which serves as an upper bound reference, and with the operational Proba-V cloud detection algorithm (v101) [27]. First of all, we see that the proposed DA method increases the mean overall accuracy and reduce the standard deviation of the metrics compared with direct transfer learning (no DA) or with adjusting the reflectance of each band Proba-V (PV) PV → LU L8-Upscaled (LU) with histogram matching.…”

Cloud Detection: An Assessment Study from the ESA Round Robin Exercise for PROBA-V

“…It served as a basis for comparing algorithms participating in the Round Robin exercise [4]. However, this dataset has never been revealed to participants, even after the end of the exercise [10].…”