How to clip Sentinel 2 bands to an area of interest with Python, Rasterio and Fiona - Tutorial

Under the concept of “spatial Python” we have developed a tutorial for the spatial processing of multiple bands from a Sentinel 2 image. The tutorial shows the procedure to read the set of bands, import a shapefile, clip each band and export the clipped version in another folder. The spatial process is independent from raster resolution and can be easily modified for Landsat images.

Tutorial

Input data

You can download the input data from this link.

Code

This is the complete Python code:

#import required libraries
import os
import fiona
import rasterio
from rasterio.mask import mask
from rasterio.plot import show
import matplotlib.pyplot as plt

#get band names
bandPath = '../S2B_MSIL1C_20200917T151709_N0209_R125_T18LUM_20200917T203629.SAFE/GRANULE/L1C_T18LUM_A018455_20200917T151745/IMG_DATA/'
bandNames = os.listdir(bandPath)
bandNames

['T18LUM_20200917T151709_B01.jp2',
 'T18LUM_20200917T151709_B02.jp2',
 'T18LUM_20200917T151709_B03.jp2',
 'T18LUM_20200917T151709_B04.jp2',
 'T18LUM_20200917T151709_B05.jp2',
 'T18LUM_20200917T151709_B06.jp2',
 'T18LUM_20200917T151709_B07.jp2',
 'T18LUM_20200917T151709_B08.jp2',
 'T18LUM_20200917T151709_B09.jp2',
 'T18LUM_20200917T151709_B10.jp2',
 'T18LUM_20200917T151709_B11.jp2',
 'T18LUM_20200917T151709_B12.jp2',
 'T18LUM_20200917T151709_B8A.jp2',
 'T18LUM_20200917T151709_TCI.jp2']

#import area of interest as Fiona geometry
aoiFile = fiona.open('../shp/AOI.shp')
aoiGeom = [aoiFile[0]['geometry']]

#clip one raster B01
bandName = bandNames[0]
rasterPath = os.path.join(bandPath,bandName)
rasterBand = rasterio.open(rasterPath)
outImage, outTransform = mask(rasterBand, aoiGeom, crop=True)
outMeta = rasterBand.meta
outMeta.update({"driver": 'JP2OpenJPEG',
                 "height": outImage.shape[1],
                 "width": outImage.shape[2],
                 "transform": outTransform})
outRaster = rasterio.open("../rst/"+bandName, "w", **outMeta) 
outRaster.write(outImage)
outRaster.close()

#plot original and clipped rasters
bandZero = rasterio.open("../rst/"+bandName,'r')
fig, ax = plt.subplots(figsize=(16,16))
show(rasterBand, cmap='Blues', ax=ax)
show(bandZero, cmap='viridis', ax=ax)
ax.set_ylim(rasterBand.bounds.bottom,rasterBand.bounds.top)
ax.set_xlim(rasterBand.bounds.left,rasterBand.bounds.right)
plt.show()
bandZero.close()

#clip all rasters
for band in bandNames:
    rasterPath = os.path.join(bandPath,band)
    rasterBand = rasterio.open(rasterPath)
    outImage, outTransform = mask(rasterBand, aoiGeom, crop=True)
    outMeta = rasterBand.meta
    outMeta.update({"driver": 'JP2OpenJPEG',
                 "height": outImage.shape[1],
                 "width": outImage.shape[2],
                 "transform": outTransform})
    outPath = os.path.join('../rst',band)
    outRaster = rasterio.open(outPath, "w", **outMeta) 
    outRaster.write(outImage)
    outRaster.close()