I have created a GeoTIFF file using the code below.
Part1: code for GeoTIFF creation.
import numpy as np
import pandas as pd
import rasterio as rio
# Define the range of latitudes and longitudes
lat_range = (39, 41)
lon_range = (-76, -73)
# Define the resolution of the grid
resolution = 0.1
# Create arrays of the latitude and longitude values
lats = np.arange(lat_range[0], lat_range[1], resolution)
lons = np.arange(lon_range[0], lon_range[1], resolution)
# Create a meshgrid of the latitude and longitude values
lon_mesh, lat_mesh = np.meshgrid(lons, lats)
# Flatten the meshgrids into 1D arrays of latitude and longitude values
lat_values = lat_mesh.flatten()
lon_values = lon_mesh.flatten()
# Generate random z values for each latitude and longitude pair
z_values = np.random.rand(len(lat_values))
# Create a pandas DataFrame to store the data
data = pd.DataFrame({'lat': lat_values, 'lon': lon_values, 'z': z_values})
# Define the GeoTIFF file parameters
width = len(lons)
height = len(lats)
transform = rio.transform.from_bounds(lon_range[0], lat_range[0], lon_range[1], lat_range[1], width, height)
crs = rio.crs.CRS.from_epsg(4326) # WGS 1984
# Write the GeoTIFF file
with rio.open("output.tif", "w", driver="GTiff", width=width, height=height, count=1, dtype=np.float32, nodata=0, transform=transform, crs=crs) as dst:
# Reshape the z values into a 2D array and write them to the GeoTIFF file
z_array = z_values.reshape((height, width))
dst.write(z_array, 1)
Then used the following code to extract the values from the above geotiff file stored as output.tif.
Part2: code for extraction.
import rasterio as rio
import pandas as pd
import geopandas as gpd
import numpy as np
# Load the GeoTIFF file
with rio.open(r"E:\\Machine_Learning_for_Himalaya_IEEE_GRSL\\plots\\output.tif") as src:
# Extract the metadata for the file
meta = src.meta
print("meta:",meta)
# Define the latitude and longitude ranges
lat_range = (39, 41)
lon_range = (-76, -73)
# Create a grid of points with a 0.1 degree interval within the specified ranges
lats = np.arange(lat_range[0], lat_range[1], 0.1)
lons = np.arange(lon_range[0], lon_range[1], 0.1)
lon_mesh, lat_mesh = np.meshgrid(lons, lats)
geometry = gpd.points_from_xy(lon_mesh.ravel(), lat_mesh.ravel())
points = gpd.GeoDataFrame(geometry=geometry)
# Extract the z values at the point locations using rasterio.sample
zs = list(src.sample(zip(points.geometry.x, points.geometry.y)))
# Create a Pandas DataFrame to store the point coordinates and values
df = pd.DataFrame({'Latitude': lat_mesh.ravel(),
'Longitude': lon_mesh.ravel(),
'Value': zs})
print(df)
The outputs are as follows: for the second part following is output:
meta: {'driver': 'GTiff', 'dtype': 'float32', 'nodata': 0.0, 'width': 30, 'height': 20, 'count': 1, 'crs': CRS.from_epsg(4326), 'transform': Affine(0.1, 0.0, -76.0,
0.0, -0.1, 41.0)}
Latitude Longitude Value
0 39.0 -76.0 [0.0]
1 39.0 -75.9 [0.0]
2 39.0 -75.8 [0.0]
3 39.0 -75.7 [0.0]
4 39.0 -75.6 [0.0]
.. ... ... ...
595 40.9 -73.5 [0.5188609]
596 40.9 -73.4 [0.589194]
597 40.9 -73.3 [0.5691334]
598 40.9 -73.2 [0.8172998]
599 40.9 -73.1 [0.5687353]
[600 rows x 3 columns]
If we print the dataframe i.e., data from the Part1 code, the output is as below.
lat lon z
0 39.0 -76.0 0.613836
1 39.0 -75.9 0.002668
2 39.0 -75.8 0.432013
3 39.0 -75.7 0.912882
4 39.0 -75.6 0.892871
.. ... ... ...
595 40.9 -73.5 0.170039
596 40.9 -73.4 0.340669
597 40.9 -73.3 0.809385
598 40.9 -73.2 0.501880
599 40.9 -73.1 0.085377
[600 rows x 3 columns]
We can clearly see the values are different for same coordinates in the above outputs. How to sort this and why this is happening?
