Suitability Evaluation for Products Generation from Multisource Remote Sensing Data

With the arrival of the big data era in Earth observation, the remote sensing communities have accumulated a large amount of invaluable and irreplaceable data for global monitoring. These massive remote sensing data have enabled large-area and long-term series Earth observation, and have, in particular, made standard, automated product generation more popular. However, there is more than one type of data selection for producing a certain remote sensing product: no single remote sensor can cover such a large area at one time. Therefore, we should automatically select the best data source from redundant multisource remote sensing data, or select substitute data if data is lacking, during the generation of remote sensing products. However, the current data selection strategy mainly adopts the empirical model, and has a lack of theoretical support and quantitative analysis. Hence, comprehensively considering the spectral characteristics of ground objects and spectra differences of each remote sensor, by means of spectrum simulation and correlation analysis, we propose a suitability evaluation model for product generation. The model will enable us to obtain the Production Suitability Index (PSI) of each remote sensing data. In order to validate the proposed model, two typical value-added information products, NDVI and NDWI, and two similar or complementary remote sensors, Landsat-OLI and HJ1A-CCD1, were chosen, and the verification experiments were performed. Through qualitative and quantitative analysis, the experimental results were consistent with our model calculation results, and strongly proved the validity of the suitability evaluation model. The proposed production suitability evaluation model could assist with standard, automated, serialized product generation. It will play an important role in one-station, value-added information services during the big data era of Earth observation.