Preliminary Evaluation of the Consistency of Landsat 8 and Sentinel-2 Time Series Products in An Urban Area—An Example in Beijing, China

Zhen Nie,Bing Xu,Karen Kie Yan Chan

doi:10.3390/rs11242957

Abstract

Global urbanization is occurring rapidly, and numerous moderate resolution remote sensing data are being used to monitor this process. Landsat 8 OLI and Sentinel-2 MSI data are combined in many applications but few studies haves focused on either urban change or consistency between these two data in time series. To evaluate the varying correlation between the two sensors in a time series, the correlation coefficient (R) and root-mean-square deviation (RMSD) of seven band pairs and three indices (NDVI, NDBI, and MNDWI) were calculated in this study and the results of the built-up area identified by IBI derived from the above three indices were compared. It was found that the correlation between the two sensors (R > 0.8534, p < 0.0001) was good in most bands but not as good for indices (in half of the results, R < 0.9). Meanwhile, the correlation of the two sensors of both bands and indices fluctuated between seasons and the comparative results of built-up area identification between the two data are relative to this variation. Therefore, when the OLI and MSI data are used in future collaboration applications, the data and threshold selection should consider the consistency and the fluctuation between the two data, especially in both time series studies and urban detection.

Highlights

Urban areas, though occupying a small percentage of the Earth’s land surface, have an important impact on economic contribution, population services, energy consumption, and global environmental change [1]
In the comparison between the various indices, the consistency characteristics of Landsat 8 (L8) and S2 are obvious and diverse: a) most value of OLI bands are smaller than the value of corresponding bands of MSI, b) except for individual bands, R ranged from 0.9133 to 0.9606 and c) the distribution diagram between OLI and MSI are all consistent in NDVI, normalized difference built-up index (NDBI), and MNDWI
Since this study focused on a preliminary comparison between Landsat 8 OLI data and Sentinel-2 MSI data, the extraction accuracy was not verified and these built-up extraction results were only used to test the similar results of the two data when applied

Summary

Introduction

Though occupying a small percentage of the Earth’s land surface, have an important impact on economic contribution, population services, energy consumption, and global environmental change [1]. Natural vegetation inside and around cities has been replaced by buildings and impervious surfaces [2]. This unprecedented progress has brought many challenges in many fields, such as public health [3], the urban heat island effect and climate change contribution [4,5], and phenology impact and ecological services [6,7]. Landsat 8 (L8) and Sentinel-2 (S2) are the successors of these two import series [16,17], and an average revisit time of 2.9 days may be achieved through the combination of Landsat 8 detection [13,14]. Landsat 8 (L8) and Sentinel-2 (S2) are the successors of these two import series R[1em6,o1te7S],enasn. 2d01a9n, 1a1v, 2e9r5a7ge revisit time of 2.9 days may be achieved through the combination of Lan2dofsa15t

Methods

Discussion

Conclusion