Please Note: This library was created as a thesis project and it is no longer maintained by its author.

All in one IDW package for python

This is an example map created using the pyidw library.

Features

1. Simple IDW Interpolation.
2. IDW with external raster (eg, elevation raster) covariable.
3. Accuracy Score.
4. Built-in raster visualisation with coordinate and colour bar.

Why pyidw ?

Inverse distance weighted interpolation is one of the simplest geospatial interpolation methods available in GIS. Although it is easy to produce an idw raster using conventional desktop GIS software (eg. ArcGIS, QGIS). It was never straightforward to create such a beautiful map image using python. This is why I have created the pyidw library where you can create beautiful idw maps of your desired location using your favourite programming language 🐍

pyidw package also incorporates a clever technique to use additional raster data as a covariable using polynomial regression. For example, if you are working with temperature data, it is widely known that temperature is inversely proportional to elevation, the higher the elevation, the lower the temperature is. With pyidw, you can easily add elevation data with traditional idw calculation to obtain a different result.

---

Installation

pyidw library can be installed using a simple pip install pyidw command. However, if you are facing trouble installing pyidw on your windows machine, please try the commands below on the windows command line.

pip install wheel
pip install pipwin
pipwin refresh
pipwin install numpy
pipwin install pandas
pipwin install shapely
pipwin install gdal
pipwin install fiona
pipwin install pyproj
pipwin install six
pipwin install rtree
pipwin install geopandas
pipwin install rasterio
pip install pyidw
pipwin refresh

---

Example

If you are convinced enough to give pyidw a try, here is a simple tutorial for you. You should first download the pyidw_example.zip file. This zip file contains four files,

• pyidw_tutorial.ipynb
• Bangladesh_Temperature.shp
• Bangladesh_Border.shp
• Bangladesh_Elevation.tif

The pyidw_tutorial.ipynb file is a jupyter notebook file of this example tutorial, which you could try to run and then modify with your own data. The Bangladesh_Temperature.shp file is an ESRI point shapefile that contains maximum and minimum temperature values for 34 weather stations all over Bangladesh. Its attribute table looks something like this.

Station_Name	Station_ID	Latitude	Longitude	Max_Temp	Min_Temp
BARISAL	BGM00041950	22.75	90.37	36.75	9.60
BHOLA	41951099999	22.68	90.65	35.62	10.19
BOGRA	BGM00041883	24.85	89.37	38.62	8.29
CHANDPUR	41941099999	23.27	90.70	35.87	11.28
CHITTAGONG	BGM00041978	22.25	91.81	36.92	11.24
CHUADANGA	41926099999	23.65	88.82	37.84	8.59
COMILLA	41933099999	23.43	91.18	35.41	10.35
COXS BAZAR	BGM00041992	21.45	91.96	37.11	11.51

For those who are not familiar with shapefile, every shapefile consists of seven different files with the same name but seven different file extensions. Namely .cpg .dbf .prj .sbn .sbx .shp and .shx. If any of these files are missing then the shapefile system won't work properly. Note that Max_Temp and Min_Temp column in Bangladesh_Temperature.shp files attribute table, we will use one of these columns later when creating IDW interpolated maps.

The Bangladesh_Border.shp is an ESRI polygon shapefile that covers all the areas of the country Bangladesh. We will use this shapefile to define the calculation extent for IDW interpolation. And finally, the Bangladesh_Elevation.tif file which is a raster file containing elevation information in meter, We don't need this file for standard IDW interpolation but with regression_idw, we will use this file as an external covariable. All the files and their spatial dimension is shown below.

---

idw_interpolation()

Now the fun part begins. Write these few lines of code from below in any python interpreter while you are on pyidw_example directory.

from pyidw import idw

idw.idw_interpolation(
    input_point_shapefile="Bangladesh_Temperature.shp",
    extent_shapefile="Bangladesh_Border.shp",
    column_name="Max_Temp",
    power=2,
    search_radious=4,
    output_resolution=250,
)

It will take a few seconds to complete, then a map image like below will be shown. And a new file will be created namely Bangladesh_Temperature_idw.tif, this is the saved raster file of the interpolated map. This file is named after input_point_shapefile name with _idw.tif suffix. idw_interpolation() function take six parameters.

• The first parameter input_point_shapefile= take an ESRI point shapefile which should contain the particular data value we are interested to create an interpolation map. Also, there shouldn't be any value outside of our given extent_shapefile area.
• The second parameter extent_shapefile= take an ESRI polygon shapefile, this shapefile is used for defining the calculation and mapping boundary. The coordinate system of extent_shapefile should be the same as input_point_shapefile.
• The third parameter column_name= take the column name of a particular field as a string. This is the value upon which the IDW map will be created.
• The fourth parameter power= is an optional parameter with a default value of 2, this is the power parameter from idw equation.
• The fifth parameter search_radious= is also an optional parameter with a default value of 4, it determines how many nearest points will be used for idw calculation.
• The sixth parameter output_resolution= is also optional with default value of 250. This parameter defines the maximum height or width (which one is higher) of the resulting _idw.tif file in pixel.

Output map from idw_interpolation() function.

---

accuracy_standard_idw()

If you are interested in accuracy assessment of your interpolation then you could use accuracy_standard_idw() function from pyidw which take 6 parameters same as idw_interpolation() function. But instead of creating a idw interpolated map, the accuracy_standard_idw() function return tow python list. The first one contains actual data values from the input shapefile and the second list contains the interpolated values for those data points using LeaveOneOut cross-validation method. Then you could compare them to obtain your desired accuracy score. An example code for accuracy_standard_idw() function is given below.

from pyidw import idw
from sklearn.metrics import mean_squared_error

original_value, interpolated_value = idw.accuracy_standard_idw(
    input_point_shapefile="Bangladesh_Temperature.shp",
    extent_shapefile="Bangladesh_Border.shp",
    column_name="Max_Temp",
    power=2,
    search_radious=6,
    output_resolution=250,
)

print("RMSE:", mean_squared_error(original_value, interpolated_value, squared=False))

Output: RMSE: 1.401379

---

show_map()

We have also implemented a raster visualization function named show_map(). This function incorporates easy map visualization with a built-in colour bar and coordinate tick marks. It takes 4 parameters.

• input_raster= take raster file name as argument.
• colormap= is an optional parameter which take matplotlib colormaps parameter. By changing this, you can easily alter the looks of your map image.
• image_size= is also an optional parameter with a default value set to 1.5 which you can change to make your resulting image larger or smaller.
• The last parameter return_figure= is for those people who wish to alter the resulting image to their liking. by default it is set to false and show_map() function won't return anything other than showing the map on the screen. If set to true, then show_map() function will return figure, axes and color_bar to the user. We will see a detailed example of this in the next section.

Here is an example code of show_map() function.

from pyidw import idw

show_map(
    input_raster="Bangladesh_Temperature_idw.tif",
    colormap="nipy_spectral_r",
    image_size=1.5,
    return_figure=False,
)

Below is an example code of setting return_figure= to True and adding some extra elements to the map 🗺 image.

from pyidw import idw
from matplotlib import pyplot as plt

fig, ax, cbar = idw.show_map(
    input_raster="Bangladesh_Temperature_idw.tif",
    colormap="CMRmap",
    image_size=1.5,
    return_figure=True)

ax.set_title("Maximum temperature map")
ax.set_xlabel("Longitude")
ax.set_ylabel("Latitude")
cbar.set_label("maximum annual temperature")
plt.show()

---

regression_idw_interpolation()

This function is quite different than regular IDW interpolation as it incorporates external raster covariable, polynomial regression and r_squared value. This is an experimental method and we don't recommend using it as it doesn't always produce reliable output and accuracy score are also lower than regular idw interpolation. It take same parameters as idw_interpolation() function only with 2 extra parameters namely input_raster_file= which is the raster covariable. input_raster file should be larger than extent_shapefile. The other parameter is polynomial_degre= with a default value set to 1 which would define the polynomial regression equation. An example code for regression_idw_interpolation() is given below.

from pyidw import idw

idw.regression_idw_interpolation(
    input_point_shapefile="Bangladesh_Temperature.shp",
    input_raster_file="Bangladesh_Elevation.tif",
    extent_shapefile="Bangladesh_Border.shp",
    column_name="Min_Temp",
    power=2,
    polynomial_degree=1,
    search_radious=5,
    output_resolution=250,
)

---

accuracy_regression_idw()

This function is similar to accuracy_standard_idw() function. An example code is given below.

from pyidw import idw
from sklearn.metrics import mean_squared_error

original_value, interpolated_value = idw.accuracy_regression_idw(
    input_point_shapefile="Bangladesh_Temperature.shp",
    input_raster_file="Bangladesh_Elevation.tif",
    extent_shapefile="Bangladesh_Border.shp",
    column_name="Min_Temp",
    power=2,
    polynomial_degree=1,
    search_radious=5,
    output_resolution=250,
)

print("RMSE:", mean_squared_error(original_value, interpolated_value, squared=False))

Output: RMSE: 1.086343

---

If you have any questions, problems or suggestion, feel free to contact me at: yahyatamim0@gmail.com