Abstract
A key challenge in developing models for the fusion of surface reflectance data across multiple satellite sensors is ensuring that they apply to both gradual vegetation phenological dynamics and abrupt land surface changes. To better model land cover spatial and temporal changes, we proposed previously a Prediction Smooth Reflectance Fusion Model (PSRFM) that combines a dynamic prediction model based on the linear spectral mixing model with a smoothing filter corresponding to the weighted average of forward and backward temporal predictions. One of the significant advantages of PSRFM is that PSRFM can model abrupt land surface changes either through optimized clusters or the residuals of the predicted gradual changes. In this paper, we expanded our approach and developed more efficient methods for clustering. We applied the new methods for dramatic land surface changes caused by a flood and a forest fire. Comparison of the model outputs showed that the new methods can capture the land surface changes more effectively. We also compared the improved PSRFM to two most popular reflectance fusion algorithms: Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM) and Enhanced version of STARFM (ESTARFM). The results showed that the improved PSRFM is more effective and outperforms STARFM and ESTARFM both visually and quantitatively.
Highlights
The demands for satellite images with higher spatial, temporal and/or spectral resolution from different application developments such as regional land-cover mapping, environmental monitoring, crop mapping and yield estimation, and climate change studies etc. are rapidly growing
The satellite imagery data collected systematically by various sensors such as NASA’s Moderate resolution Imaging Spectroradiometer (MODIS), Landsat Operational Land Imagers and Sentinels Multispectral Imagers developed by the European Space Agency (ESA) enable us to produce some images for such demands
The problem to be solved by the image fusion model is to predict a synthetic fine-resolution image on date t1 given two pairs of co-incident clear-sky fine-resolution (e.g., Landsat Operational Land Imager or Sentinel-2 Multispectral Imager) and coarse-resolution (e.g., MODIS) surface reflectance images on a start date t0 and an end date t2 and a coarse-resolution image at t1 within this period
Summary
The demands for satellite images with higher spatial, temporal and/or spectral resolution from different application developments such as regional land-cover mapping, environmental monitoring, crop mapping and yield estimation, and climate change studies etc. are rapidly growing. Examples include the Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM) [1], the Enhanced version of STARFM (ESTARFM) [2], the Spatial Temporal Adaptive Algorithm for Mapping Reflectance Change (STAARCH) [3], the Unmixing-based Spatial-Temporal Reflectance Fusion Model (U-STFM) [4,5], the Flexible Spatiotemporal Data Fusion (FSDAF) model [6], the Spatiotemporal image-fusion model (STI-FM) for enhancing the temporal resolution [7], the Hybrid Color Mapping (HCM) approach [8], the Rigorously-Weighted Spatiotemporal Data Fusion Model (RWSTFM) [9], the Prediction Smooth Reflectance Fusion Model (PSRFM) [10] These studies have demonstrated that the fusion methods are mostly limited to predicting spectral changes in reflectance due to gradual vegetation phenological changes [6,7]. There is a clear need for better fusion model to predict both gradual and abrupt land surface changes [10]
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
