How much green is Delhi as on 13 Oct 2022?

delhi_image = landsat.filter_by('OBJECTID=3866751') # 2022-10-21

Applying Natural color to verify the quality of image.

apply(delhi_image, 'Natural Color with DRA')

Creating NDVI composite for the filtered imagery

In the Landsat layer properties, pre defined "NDVI Raw" function is applied to get the NDVI composite.

ndvi_colorized = apply(delhi_image, 'NDVI Raw')
ndvi_colorized

Clipping the NDVI composite for Delhi and setting the extent of the generated raster.

delhi_clip = clip(raster=ndvi_colorized, geometry= delhi_geom)
delhi_clip.extent = area['extent']
delhi_clip

Masking NDVI composite for green area calculation

"remap" function is used to define the NDVI range for agricultural land and forest. The NDVI values between 0.4 - 0.5 represents agricultural land whereas NDVI values between 0.5 - 1 shows forest/ tree cover [2].

threshold_val = 0.5
masked = colormap(remap(delhi_clip, 
                        input_ranges=[0.4,threshold_val,     # agricultural land
                                     threshold_val, 1],      # forest area/ tree cover
                        output_values=[1, 2]),
                        colormap=[[1, 124, 252, 0], [2, 0, 102, 0]], astype='u8')

Image(masked.export_image(bbox=area['extent'], size=[1200,450], f='image'))

Visualising results on map

m = gis.map('New Delhi, India', 10)
m.add_layer(masked)
m.legend = True
m

Here, the pixels with light green color having a value of "1" represent agricultural land whereas the pixels with dark green color having a value of "2" represent forest area/ tree cover.

Area Derivation

Now we will calculate the total green cover of Delhi. It is important to note that pixels belonging to agricultural land as well as forest area/ tree cover are considered as green cover in the calculation.

pixx = (delhi_clip.extent['xmax'] - delhi_clip.extent['xmin']) / 1200
pixy = (delhi_clip.extent['ymax'] - delhi_clip.extent['ymin']) / 450

res = masked.compute_histograms(delhi_clip.extent,
                               pixel_size={'x':pixx, 'y':pixy})
numpix = 0
histogram = res['histograms'][0]['counts'][0:]
for i in histogram[1:]:
    numpix += i

sqmarea = numpix * pixx * pixy # in sq. m
acres = 0.00024711 * sqmarea   # in acres
HTML('<h3>Total Green cover in Delhi is ~ <i>{}%</i> of the total \
     geographical area of Delhi.</h3>'.format(int((acres/419004.17)*100)))

plt.title('Green Cover', y=-0.1)
plt.pie(histogram, labels=['Non green','Agricultural Land', 'Forest Cover']);
plt.axis('equal');

The pie chart clearly shows the distribution of agricultural land, forest cover and non green areas like urban areas, water, etc. Here, the non green areas contribute around 74% of Delhi's land and remaining 26% goes with forest cover and agricultural land.

Conclusion

In this study, green cover of Delhi is calculated using Landsat 8 imagery for 21 October, 2022. Normalised Difference Vegetation Index (NDVI) is used for the calculation of green areas, which is a well known and widely accepted spectral index for vegetation studies.

NDVI is used here compared to the other spectral indices because it is easy to compute and requires only two bands. Pixels with NDVI values greater than 0.4 and less than 0.5 are considered as agricultural land, parks, etc and the pixels with more than or equal to 0.5 NDVI values are considered as tree cover/ forest areas.

Finally, the area of the pixels of agricultural land and forest class is calculated to estimate overall green cover of Delhi.

The study shows how the green land cover of an area could be easily computed in few lines of code using Esri's predefined NDVI layer and Landsat 8 imagery from ArcGIS server.

References

[1] https://fsi.nic.in/forest-report-2021-details

[2] https://economictimes.indiatimes.com/news/india/environmentalists-question-govt-report-on-delhis-forest-cover/articleshow/92164943.cms?from=mdr

[3] https://earthobservatory.nasa.gov/features/MeasuringVegetation