Extracting the central strip from LANDSAT 7 imagery

Extracting the central strip from LANDSAT 7 imagery

February 8, 2016

Here is a simple Python code to extract the central strip from Landsat 7 imagery (SLC-off),  that is not affected by the SLC failure. The algorithm shrinks the striping zones through a morphological filter (erosion) and creates a new shapefile AOI that extracts the desired raster extent without striping effects. The code is based on Python for ArcGIS (arcpy) – so you require a ArcGIS license.

General steps:

  1. Loop through all Landsat 7 data folders
  2. Stack bands for each image
  3. Create a mask
  4. Erode the mask by 20 pixels
  5. Convert the mask to polygon
  6. Create a minimum bounding box
  7. Clip the original raster through the bbox

 

import arcpy
from arcpy.sa import *

import sys,os

#  Environment settings (Activate Spatial Analyst, Overwrite Outputs allowed and TIFF compression is LZW)
arcpy.CheckOutExtension("spatial")
arcpy.env.overwriteOutput = True
arcpy.env.compression = 'LZW'

# this is your main directory with all unzipped Landsat datasets
 rootdir = "D:\\DATA\\Landsat7\\"

# create scratch folder "temp" 
temp = "D:\\DATA\\temp\\"

# loop through directory with all unzipped Landsat 7 folders
 for files in os.listdir(rootdir):   
    files = os.path.join(rootdir, files)   
    
    # for each loop the subdir "files" is now the current workspace 
    # (e.g. LE71520322015157-SC20160224113319) that contains the Landsat bands
    arcpy.env.workspace = files  
    rasters = arcpy.ListRasters("*", "TIF")  
    
    # create empty array
    stack_liste = []  
    # loop through all rasters in subdir
    for raster in rasters:   

        image = arcpy.Raster(raster) 
        name  = image.name 
        index = name.split("_")[0]  

        # fill up the array only with the actual spectral bands        
        sr = "_sr_band"  
        if sr in raster:   
            stack_liste.append(raster)             

    # now stack all bands within the array
    stack_name = files + "\\" + index + "_stack.tif"    
    arcpy.CompositeBands_management(stack_liste, stack_name)  

    # convert the image stack to a mask by logical operation with an absurd value that will result in an output "0"
    con = EqualTo(stack_name, 123456789)  

    # now shrink the raster mask with value "0" by 20 pixels
    shrink = temp + "shrink"  
    shrinking = Shrink(con, 20, 0) 
    shrinking.save(shrink)  

    zone = temp + "zone.shp" 
    bbox = temp + "bbox.shp"  

    # conver the shrunk mask to polygon and create a minimum bounding box
    arcpy.RasterToPolygon_conversion(shrink, zone, "NO_SIMPLIFY", "VALUE") 
    arcpy.MinimumBoundingGeometry_management(zone, bbox, "RECTANGLE_BY_WIDTH", "NONE")  

    # now use that bounding box as a mask to cut out the central nadir strip from the original stack
    # Final result 
    extract = files + "\\" + index + "_aoi.tif"  
    ExtractByMask = arcpy.sa.ExtractByMask(stack_name, bbox) 
    ExtractByMask.save(extract)

 

you may also like:

R package for time-series animation of rasters

R package for time-series animation of rasters

One of remote sensing's greatest strengths is its capability to deliver consistent and highly resolved time series. These have a wonderful potential for visualizations and analysis.Creating a fluid and pretty animation from a collection of tifs can be complicated, as...

Who can live without chocolates?

Who can live without chocolates?

Itohan-Osa Abu and Dr Michael Thiel, have published a paper in Ecological Indicators in cooperation with the partners at the Joint Research Centre (JRC) titled "Detecting cocoa plantations in Côte d’Ivoire and Ghana and their implications on protected areas". From the...

Our Virtual teaching setup

Our Virtual teaching setup

We posted a while ago some images about our virtual teaching room and mentioned that we will provide briefly the general technical background setup: we have four lecture rooms each with a broadcast camera (that can stream for 6 hours and more), microphones and green...

most recent news:

R package for time-series animation of rasters

R package for time-series animation of rasters

One of remote sensing's greatest strengths is its capability to deliver consistent and highly resolved time series. These have a wonderful potential for visualizations and analysis.Creating a fluid and pretty animation from a collection of tifs can be complicated, as...

Who can live without chocolates?

Who can live without chocolates?

Itohan-Osa Abu and Dr Michael Thiel, have published a paper in Ecological Indicators in cooperation with the partners at the Joint Research Centre (JRC) titled "Detecting cocoa plantations in Côte d’Ivoire and Ghana and their implications on protected areas". From the...

Our Virtual teaching setup

Our Virtual teaching setup

We posted a while ago some images about our virtual teaching room and mentioned that we will provide briefly the general technical background setup: we have four lecture rooms each with a broadcast camera (that can stream for 6 hours and more), microphones and green...