C-scale appendix

This notebook intends to show couple ideas how to analyse Sentinel-2 satellite images using different spectral bands and visualising the output.

import os
import numpy as np
from matplotlib import pyplot as plt
from cscale_notebooks_functions import load_sentinel_image, display_rgb, image_rgb, normalized_difference, plot_masked_rgb, resampling

# Choose folder where are your downloaded satelite images  
img_folder = "/images/clipped/"
path = os.path.abspath("")
folder_img = path + img_folder

Different channels has different resolution, so it needs to be resampled in order to be able to put them together. You can check the channel resolution here.

In case of L1C data we need to resample channels B05, B06, B11 and B12 with the factor of 2 and B09 with factor 6.

file_temp = "B02.tif"
file_res = ["B05.tif", "B06.tif", "B11.tif", "B12.tif"]

for n in range(len(file_res)):
    resampled = resampling(folder_img,file_temp,file_res[n],2)

NOTE: In case of L1C data the channel B09 is resampled with factor of 6, whereas L2A data needs resampling of ONLY channel B09 with the factor of 3!

file_temp = "B02.tif"
file_res = ["B09.tif", "B10.tif"]

for n in range(len(file_res)):
    resampled_09 = resampling(folder_img,file_temp,file_res[n],6)

NOTE: In case of L1C data we need to enter B05_res, B06_res channels. For L2A just B05 and B06.

We can compute spectral index using different bands. In this notebook we calculate Normalized Difference (ND) index using custom function. ND index is defined as:

ND = (Band1 - Band2)/ (Band1 + Band2)

Here we calculate two indexes, where different areas are emphasized.

img_rgb = load_sentinel_image(folder_img, ["B02", "B03", "B04"])
img_nir = load_sentinel_image(folder_img, ["B03", "B04", "B06_res"])
img_nir2 = load_sentinel_image(folder_img, ["B03", "B04", "B08"])

rgb = image_rgb(img_rgb, 'B04', 'B03', 'B02', alpha=5.)
nir = image_rgb(img_nir, 'B06_res', 'B04', 'B03', alpha=5.)

# Calculating two indices
ndvi = normalized_difference(img_nir2, 'B08', 'B04')
mndwi = normalized_difference(img_nir2, 'B03', 'B08')

from matplotlib import pyplot as plt
# checking the images
fig, ax = plt.subplots(2, 2, figsize=(15, 10))

fontsize=16
ax = ax.flatten()

ax[0].imshow((rgb*255).astype(np.uint8), interpolation='none')
ax[0].axis('off')
ax[0].set_title('Color Image', fontsize=fontsize)

ax[1].imshow((nir*255).astype(np.uint8), interpolation='none')
ax[1].axis('off')
ax[1].set_title('NIR', fontsize=fontsize)

ax[2].imshow(ndvi, cmap='Greens', interpolation='none')
ax[2].axis('off')
ax[2].set_title('NDVI', fontsize=fontsize)

ax[3].imshow(mndwi, cmap='Blues', interpolation='none') # * 255).astype(np.uint8)
ax[3].axis('off')
ax[3].set_title('MNDWI', fontsize=fontsize)

plt.subplots_adjust(wspace=0.1, hspace=0.1)

plt.show()

Colour image is a true colour image, what one can observe for example from a plane. NIR image is more sensitive to different kind of vegetation and water areas. NDVI index gives a better response (higher values = greener because of the colormap we applied) over the vegetation areas and the MNDWI has a higher response (higher values = bluer because of the ‘Blues’ colormap) for water areas like rivers.

We can play more with the bands and their order while displaying.

In case of L1C data we need to enter B12_res and B11_res. L2A just B12 and B11.

img = load_sentinel_image(folder_img, ["B03", "B04", "B12_res"])
rgb = image_rgb(img, 'B12_res', 'B03', 'B04', alpha=5.)
bgr = image_rgb(img, 'B04', 'B12_res', 'B03', alpha=5.)

fig, ax = plt.subplots(1, 2, figsize=(15, 10))
ax[0].imshow((rgb* 255).astype(np.uint8), cmap='Greens')
ax[0].axis('off')
ax[1].imshow((bgr* 255).astype(np.uint8), cmap='Blues')
ax[1].axis('off')
plt.show()

We can look for parks, water surfaces or buildings in city area

from cscale_notebooks_functions import clipper, coor_converter
coor_converter("geojson/prague.geojson","geojson/prague_GPScut.geojson",32633)
Prague_reg = "geojson/prague_GPScut.geojson"

roicut_path = path + "/images/cut_out/"
if os.path.isdir(roicut_path) != True:
    os.mkdir(roicut_path)

clipper(folder_img, Prague_reg, roicut_path, whole = "yes")

img = load_sentinel_image(roicut_path, ["B03", "B04", "B09_res"])
img1 = load_sentinel_image(roicut_path, ["B03", "B04", "B09_res"])
rgb = image_rgb(img, 'B03', 'B04', 'B09_res', alpha=5.5)
bgr = image_rgb(img1, 'B04', 'B09_res', 'B03', alpha=5.5)

fig, ax = plt.subplots(1, 2, figsize= (15, 10))
ax[0].imshow((rgb* 255).astype(np.uint8))
ax[0].axis('off')
ax[1].imshow((bgr* 255).astype(np.uint8))
ax[1].axis('off')
plt.show()

diff = normalized_difference(img,'B09_res','B03',)

plt.figure(figsize=(10,5))
plt.imshow(diff, cmap = 'Greens_r')
plt.axis("off")
plt.show()

C-scale appendix

NOTE: In case of L1C data the channel B09 is resampled with factor of 6, whereas L2A data needs resampling of ONLY channel B09 with the factor of 3!

NOTE: In case of L1C data we need to enter B05_res, B06_res channels. For L2A just B05 and B06.

We can play more with the bands and their order while displaying.

We can look for parks, water surfaces or buildings in city area

On this page