Reconstruction of High-Temporal- and High-Spatial-Resolution Reflectance Datasets Using Difference Construction and Bayesian Unmixing

Lei Yang,Jing Wang,Lijuan Han,Xin Wang,Jinling Song

doi:10.3390/rs12233952

Abstract

High-temporal- and high-spatial-resolution reflectance datasets play a vital role in monitoring dynamic changes at the Earth’s land surface. So far, many sensors have been designed with a trade-off between swath width and pixel size; thus, it is difficult to obtain reflectance data with both high spatial resolution and frequent coverage from a single sensor. In this study, we propose a new Reflectance Bayesian Spatiotemporal Fusion Model (Ref-BSFM) using Landsat and MODIS (Moderate Resolution Imaging Spectroradiometer) surface reflectance, which is then used to construct reflectance datasets with high spatiotemporal resolution and a long time series. By comparing this model with other popular reconstruction methods (the Flexible Spatiotemporal Data Fusion Model, the Spatial and Temporal Adaptive Reflectance Fusion Model, and the Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model), we demonstrate that our approach has the following advantages: (1) higher prediction accuracy, (2) effective treatment of cloud coverage, (3) insensitivity to the time span of data acquisition, (4) capture of temporal change information, and (5) higher retention of spatial details and inconspicuous MODIS patches. Reflectance time-series datasets generated by Ref-BSFM can be used to calculate a variety of remote-sensing-based vegetation indices, providing an important data source for land surface dynamic monitoring.

Highlights

Dense time-series data from satellites are commonly used to study dynamics of the land surface at large spatial scales, such as monitoring phenological changes of land surface vegetation, evaluating occurrence of natural disasters, mapping distribution of land features, and estimating crop yields [1,2,3]
The quality of Moderate-Resolution Imaging Spectrometer (MODIS) data is more important to the final results; we analyzed the fusion results of the Ref-BSFM model separately for the two cases of MODIS data with good quality and poor quality
We showed that Ref-BSFM has the following advantages: (1) Ref-BSFM maintains high prediction accuracy resulting in a high level of image accuracy and image quality, (2) it generates quality prediction results for cloudy areas, (3) when there are available Landsat data with a large time span, Ref-BSFM achieves result closer to the actual image than the results from other fusion algorithms, (4) the change information captured using MODIS data can be saved through time and fully utilized in Ref-BSFM, and (5) the NDVI-BSFM method inherits the characteristics of clear spatial details and inconspicuous patch effects

Summary

Introduction

Dense time-series data from satellites are commonly used to study dynamics of the land surface at large spatial scales, such as monitoring phenological changes of land surface vegetation, evaluating occurrence of natural disasters, mapping distribution of land features, and estimating crop yields [1,2,3]. In heterogeneous regions with a larger number of land-cover types, change-monitoring studies require a long time series of satellite data with higher spatial resolution to more accurately determine the timing and characteristics of the changes. The Google Earth Engine cloud computing platform, through the use of a combination of user-uploaded algorithms and online data, can and quickly produce time-series data [5]

Methods

Results

Discussion

Conclusion