Table 2: Description of the datasets for multispectral images. It includes selected product, reflectance type, platform used to analyze the data (GEE is Google Earth Engine, SH is Sentinel-HUB, EC is Earth Console), period of data availability, spatial resolution and revisit time. Some datasets share data from one or more sensors (e.g. S Sentinel- 2 in Google Earth Engine (GEE) and Sentinel-2 in Sentinel Hub (SH)). To avoid using more than one time the same data, three different combinations were considered for the merging:
Landsat 5-7-8-9; Sentinel 2 (GEE) – 3; MODIS TERRA and AQUA; MERIS Landsat 5-7-8-9; Sentinel 2 (SH) – 3; MODIS TERRA and AQUA; MERIS Landsat 5-7-8-9; Sentinel 2 (GEE) – 3; MODIS TERRA and AQUA; MERIS It should be also noticed that some sensors may be not able to obtain reliable river discharge information over specific rivers due to insufficient spatial resolution (when river width fail of the uncalibrated procedure (impossibility to obtain C or W due to sensor’s spatial resolution or specific characteristic of the river) or other reasons. Therefore, it is important to exclude bad performing data from the merging to avoid worsening the performance.
For the calibrated stations, all the datasets with Spearman correlation with observations uncalibrated procedure, where also stations with non-concurrent data were considered, this strategy is not applicable. Therefore, first the CM index will be averaged to obtain a monthly timeseries, then the CM of each single month (January, February…) will be averaged together to obtain a monthly standard-year CM index. The Spearman correlation between this index and the corresponding monthly standard-year observed data ( considering also non-concurrent periods) will be calculated and considered as a good estimator of the actual Spearman correlation of the calibrated timeseries ().
A categorical score analysis was then carried out to obtain the best monthly Spearman correlation threshold () able to reproduce the condition in which the calibrated Spearman correlation in less than 0.4: for each potential threshold it was calculated the Probability of Detection (POD), False Alarm Ratio (FAR) and Threat Score (TS), considering: Eq.12a Eq.12b D.3 River Discharge ATBD. Reference: RD_CCI-0006-ATBD - Issue 2.0 – 31/07/2025 Open/Public © 2019 CLS. All rights reserved. Proprietary and Confidential. 25/31