Reclassifying UrbanWatch rasters using QGIS and Rasterio

README.md

Reclassifying UrbanWatch rasters using QGIS and Rasterio

See the video demo at https://youtu.be/t-pb5-1Q_Qg

The files here were used to convert UrbanWatch from RGB images to numeric rasters with simple landcover values 0-9, as follows:

Value	Category	Color
0	No Data	Black
1	Building	Red
2	Road	Gray
3	Parking Lot	Pink
4	Tree Canopy	Dark Green
5	Grass/Shrub	Light Green
6	Agriculture	Yellow
7	Water	Blue
8	Barren	Brown
9	Other	White

To save the files below, right-click the corresponding "Raw" button then "Save link as..."

BONUS: Each file can also be converted to a COG (Cloud-Optimized GeoTIFF) using rio cogeo using a command like:

rio cogeo create in.tif out.tif

compress_and_add_colormap.py

import rasterio

files = [
    "urbanwatch-atlanta-2017.tif",
    "urbanwatch-boston-2014.tif",
    "urbanwatch-charlotte-2016.tif",
    "urbanwatch-chicago-2017.tif",
    "urbanwatch-dallas-2016.tif",
    "urbanwatch-denton-county-2020.tif",
    "urbanwatch-denver-2017.tif",
    "urbanwatch-des-moines-2017.tif",
    "urbanwatch-detroit-2014.tif",
    "urbanwatch-houston-2016.tif",
    "urbanwatch-los-angeles-2014.tif",
    "urbanwatch-miami-2017.tif",
    "urbanwatch-minneapolis-2017.tif",
    "urbanwatch-new-york-city-2017.tif",
    "urbanwatch-philadelphia-2017.tif",
    "urbanwatch-phoenix-2015.tif",
    "urbanwatch-raleigh-2016.tif",
    "urbanwatch-riverside-2014.tif",
    "urbanwatch-san-diego-2014.tif",
    "urbanwatch-san-francisco-2014.tif",
    "urbanwatch-seattle-2017.tif",
    "urbanwatch-st-louis-2014.tif",
    "urbanwatch-washington-dc-2017.tif"
]

with rasterio.Env():

    for f in files:

        with rasterio.open('input/{}'.format(f)) as src:
            orig = src.read(1)
            meta = src.meta
            meta.update(compress='deflate', nodata=0)
            # deflate is better than the lzw used in the video

        with rasterio.open('output/{}'.format(f), 'w', **meta) as dst:
            dst.write(orig, indexes=1)
            dst.write_colormap(
                1, {
                    0: (0, 0, 0, 255),
                    1: (255, 0, 0, 255),
                    2: (133, 133, 133, 255),
                    3: (255, 0, 192, 255),
                    4: (34, 139, 34, 255),
                    5: (128, 236, 104, 255),
                    6: (255, 193, 37, 255),
                    7: (0, 0, 255, 255),
                    8: (128, 0, 0, 255),
                    9: (255, 255, 255, 255) })

reclass_table.csv