BBox problem with rasterio - Converting KML file to GeoTiff file on Python

Question

At my work, I need to convert different kinds of files into Geotiff files. Now I have to convert a banch o KML files. Those came with the coordinates from the corners of the polygon. Using rasterio to convert to GeoTiff, I need to use West, East, South and North coordinates, which is giving me a couple of meters of error from the rigth location. Any advice? My code and the image on Google Earth:

How the coordinates come on KML file:

[['POLYGON ((-68.02573239434068 10.458541609679036'], [' -67.97139569083748 10.447019877002532'], [' -67.9631296429911 10.495261426980813'], [' -68.01557210143345 10.506382845074937'], [' -68.02573239434068 10.458541609679036))']

Code:

  south = np.float64(lat_min)
  north = np.float64(lat_max)
  east = np.float64(lon_min)
  west = np.float64(lon_max)

  with rio.open(png, 'r') as img:
  img = img.read([1])
  img = img.astype('uint16')
  bounds = rio.transform.from_bounds(west, south, east, north, 
  img.shape[1], 
  img.shape[2])
  crs = '+proj=latlong'
  show(img) #shows true color
  with rio.open('{}.tiff'.format(file),'w', 
                 driver='GTiff',
                 count=1,
                 height=img.shape[1],
                 width=img.shape[2],
                 dtype=img.dtype,
                 crs=crs,
                 transform= bounds,
                 ) as dst:
        dst.write(img)

Answer 1

rio.transform.from_bounds takes in order width, height. You are specifiying in order shape[1] then shape[2]

But in the other hand, in open you are specifying: height=img.shape[1], width=img.shape[2]

I think there lies the issue of your code, you should have written:

 tf = rio.transform.from_bounds(
   west, 
   south, 
   east, 
   north, 
   width=img.shape[2], 
   height=img.shape[1]
)

---

I also think you should use EPSG:4326 as KML crs, as specified in GDAL: https://gdal.org/drivers/vector/kml.html