Answers are grounded in this paper and cite their sources.
Blending of temporal high-frequency-low-spatial resolution MODIS with temporal low-frequency-high-spatial resolution Landsat satellite imagery enhances the frequency and resolution of spatial data, thus enabling continuous monitoring of dynamic and rapidly changing environmental conditions. Blending can be particularly useful in areas with high cloud cover, such as during the monsoon season in the coastal zone of the Ganges Brahmaputra Delta (CZGBD). In this study, MODIS-Landsat blending of reflectance-derived remote sensing indices is trialled and evaluated in the CZGBD. The Sub-pixel class fraction change information Flexible Spatiotemporal DAta Fusion (SFSDAF) algorithm is used to obtain gap free 30 m and 16-day frequency vegetation, salinity and water indices for the entire CZGBD. Pixels obtained through blending were compared to the observed pixels (i.e., not contaminated by clouds to evaluate the accuracy of SFSDAF). Results during the 'dry' months (October to March) had a combined mean coefficient of determination, R2 = 0.65 and mean root squared error, RMSE = 0.09 for vegetation indices, whereas the results during the 'wet' months (April to September) had a combined mean R2 = 0.33 and mean RMSE = 0.12. The reduced accuracy of the blending during the monsoon months showcases the effects of cloudy conditions. Results for salinity and water indices showed similar behaviour as the vegetation indices, influenced by the cloudy monsoon season. The main cause of low accuracy during the 'wet' months is the paucity of data to perform the blending, even at the daily MODIS frequency. In addition, remote sensing indices with equations that normalised the range (generally between -1 and 1) had better results when compared to remote sensing indices that had less constrained ranges.
No takes yet. Share an insight, caveat, or question.
Peña‐Arancibia et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: