An Effective Gap Filtering Method for Landsat ETM+ SLC-Off Data

Abstract

The Landsat 7 Enhanced Thematic Mapper Plus (ETM+) scan line corrector (SLC) failed on 31 May 2003, causing the SLC to turn off. Many gap-filled products were developed and deployed to combat this situation. The majority of these products used a primary image taken by the SLC when functioning properly in an attempt to correct SLC-off images. However, temporal atmospheric elements could not be reliably reflected using a primary image, and therefore the corrected image was not viable for use by monitoring systems. To bypass this limitation, this study has developed the Gap Interpolation and Filtering (GIF) method that relies on one-dimensional interpolation filtering to conveniently recover pixels within a single image at a high level of accuracy without borrowing from images acquired at a different time or by another sensor. The GIF method was compared to two other methods-Global Linear Histogram Match (GLHM), and the Local Linear Histogram Match (LLHM)-both developed by National Aeronautics and Space Administration (NASA) and United States Geological Survey (USGS) to determine its accuracy. The GIF method accuracy was found superior in land, sea, and cloud imaging. In particular, its sea and cloud images returned Root Mean Square Error (RMSE) values close to or less than 1. We expect the GIF method developed in this research to be of invaluable aid to monitoring systems that depend heavily on Landsat imagery.

Mapping mineralogy in evaporite basins through time using multispectral Landsat data: Examples from the Bonneville basin, Utah, USA

“…Optical sensors utilized are the Landsat‐8 OLI and Landsat‐5 Thematic Mapper (TM) sensors, which share adequately similar sensor spectral resolution (VNIR‐SWIR), 30 m spatial resolution, and 185 km × 180 km scene sizes. Landsat satellites 6 and 7 were not used as Landsat‐6 had a transient life‐span and Landsat‐7 has corrupted scenes due to a permanent scan line corrector failure which occurred in 2003 (Lee et al ., 2017). Additionally, the choice to use Landsat TM and OLI instead of ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer), or other similar multispectral satellites, is based around Landsat's ability to fully capture the extent of the study area(s) in a single day.…”

“…recognizes halite to be the dominant mineral in the sample, when the sample is observed through VNIR spectroscopy the spectral features due to gypsum (water absorption) are very strong, exemplifying non-linear mixing. Thus, there is gypsum in the halite and halite in the gypsum present within the Bonneville basin, and the spectral mixing of these endmembers is non-linear resulting in strong water features even for predominantly pure halite Lee et al, 2017)…”

The role of random forest and Markov chain models in understanding metropolitan urban growth trajectory

“…Firstly, we opted for images with fewer scanline errors, especially in locations with human settlements. Secondly, we employed sophisticated gap-filling techniques, including the Local Linear Histogram Matching (LLHM) method (Chen et al, 2015;Lee et al, 2016). This method utilizes statistical properties of neighboring pixels to estimate missing pixel values, thereby minimizing the effects of data gaps on our analysis (Asare et al, 2020).…”

Urban Features Identification from Dual-Pol SAR Images with Filter Properties

“…Different data were used here one is JERS-1 SAR data and SIR-C/X SAR data with different polarization. And the results proposed that in context to preserve the edges LEE, frost and Lee sigma & Gamma filters are best because they can maintain a balance between speckle reduction and preservation of detail in much application areas (Jian et al, 2020;Lee et al, 2016). The paper discusses the smoothing algorithm, the filter is based on the sigma of Gaussian distribution which helps to smooth the speckle noise by averaging the neighbourhood pixels to the centre pixel.…”