Skip To Content ArcGIS for Developers Sign In Dashboard

ArcGIS API for Python

Download the samples Try it live

Calculate Impervious Surfaces from Spectral Imagery

Introduction and objective

Ground surfaces that are impenetrable to water can cause serious environmental problems, including flooding and contaminated runoff. Many local government agencies use impervious surface calculations to compute the storm water bill for properties. That depends on how much impervious area is there in each property.

Map highlighting impervious surface areas: driveways, private walks, roof tops, and parking lots [1].

In this notebook, we’ll use a high resolution land cover map obtained from Chesapeake Conservancy to determine which parts of the ground are pervious and impervious.

Impervious surfaces are generally human-made: buildings, roads, parking lots, brick, or asphalt. Pervious surfaces include vegetation, water bodies, and bare soil.

We will reclassify those land-use types into either impervious or pervious surfaces to derive the stormwater bill for each property. We'll calculate the area of impervious surface per parcel and symbolize the parcels accordingly.

Necessary Imports

In [1]:
import pandas as pd
from datetime import datetime as dt

import arcgis
import arcpy
from arcgis.gis import GIS
from arcgis.raster.functions import RFT           

gis = GIS(url='', username='arcgis_python', password='amazing_arcgis_123')

Get data for analysis

For this analysis we need the following datasets:

  • Land Cover of Kent County, Delaware
  • Kent County parcel feature layer
  • Raster Function Template to extract necessary class labels from the land cover layer

In the cells below, we access these datasets from the GIS.

In [2]:
rft_to_reclassify ="title: RFT_Reclassify_RasterFunction owner:api_data_owner",
                                       "Raster Function Template")[0]
RFT_Reclassify_RasterFunctionRaster function template by api_data_owner
Last Modified: February 28, 2020
0 comments, 0 views
In [3]:
classified_raster ='Land Cover of Kent County Delaware owner:api_data_owner')[0]
Land Cover of Kent County Delaware
This dataset was developed to support land-cover mapping and modeling initiatives in the Chesapeake Bay Watershed and Delaware River Basin. At the time of its...Imagery Layer by api_data_owner
Last Modified: February 28, 2020
0 comments, 0 views
In [4]:

Twelve land cover classes were mapped:

  1. Water
  2. Emergent Wetlands
  3. Tree Canopy
  4. Scrub/Shrub
  5. Low Vegetation
  6. Barren
  7. Structures
  8. Other Impervious Surfaces
  9. Roads
  10. Tree Canopy over Structures
  11. Tree Canopy over Other Impervious Surfaces
  12. Tree Canopy over Roads

The complete class definitions and standards can be viewed at the link below.

In [5]:
parcels ='kent_county_parcels owner:api_data_owner', 'feature layer')[0]
In [6]:
Feature Layer Collection by api_data_owner
Last Modified: February 28, 2020
0 comments, 2 views
In [7]:
parcel_lyr = parcels.layers[0]

Classify pixel values into pervious and impervious class

Create the RFT object from raster function template portal item. This object serves as a python function corresponding to the Raster Function Template.

In [8]:
rft_to_reclassify_ob = RFT(rft_to_reclassify)

Once an RFT class object has been created, a single question mark before, or after the RFT object will show help relative to it. The help document would display the parameters that were marked as public by the author of the RFT.

In [9]:
RFT to classify pixel values to pervious and impervious class


Type : 3
Rows : 5

Imagery Layer, on which the function chain is applied 

We can visualize the function chain and the help document specific to this RFT inorder to understand the input parameters and the functions involved.

To view the raster function chain visually, we install graphviz Python library.

In [ ]:
# ! conda install graphviz -y
In [11]:

Provide the values for the input parameters in the RFT object to create the Imagery Layer.

In [12]:
output = rft_to_reclassify_ob(Raster=classified_raster.layers[0])
In [13]:
persisted_output ='generate_reclassified_raster' + str(
In [14]:
impervious_lyr = persisted_output.layers[0]
In [15]:

Tabulate area of impervious surfaces

We'll determine the area of impervious surfaces within each parcel of land in the neighborhood. We'll first calculate the area and store the results in a stand-alone table. Then, we'll join the table to the Parcels layer.

In [16]:
arcpy.CreateFileGDB_management(r"\\notebook\output\impervious", "output_table.gdb")

Impervious field shows the area (in feet) of impervious surfaces per parcel, while Pervious shows the area of pervious surfaces. We will calculate the impervious area with the name field which denotes Parcel_ID.

In [17]:
tabulate_op =, 

Now we have the area of impervious surfaces per parcel, but only in a stand-alone table. Next, we'll join the stand-alone table to the Parcels attribute table. A table join updates the input table with the attributes from another table based on a common attribute field. Because we created the impervious area table with the name field from the Parcels layer, we'll perform the join based on that field.

In [18]:, 

Compute stormwater bill

We will now use add_to_definition method which adds a new field to an existing feature layer. We will then calculate percentage of impervious area to determine the tax value.

In [19]:
{'success': True}
In [20]:
parcel_lyr.calculate(where="fid > 0", 
                                       "sqlExpression":" ( impervious / ( pervious + impervious ) ) * 100"}])
parcel_lyr.calculate(where="fid > 0", 
                                       "sqlExpression":" impervious * 0.04"}])
{'success': True,
 'updatedFeatureCount': 83142,
 'recordCount': 83142,
 'layerName': 'kent_county_parcels'}

percent_imp is the percentage of impervious area.

rate is the tax value calculated for $0.04 per square ft of impervious area.

In [21]:
{'success': True}

Visualize results

In [22]:
sdf = pd.DataFrame.spatial.from_layer(parcel_lyr)
sdf[['name', 'mailingadd', 'ownercity', 'propertyus', 'impervious', 'pervious', 'percent_imp', 'rate']]
name mailingadd ownercity propertyus impervious_12 pervious_12 percent_imp rate
0 6-00-19700-01-1800-00001 OF HARRY RONALD WEBB GREENWOOD Agricultural-Land 330.0 645297.0 0.051113 13.20
1 6-00-19700-01-1802-00001 4318 MOUNT HOLLY RD GREENWOOD Vacant Land 0.0 47.0 0.000000 0.00
2 6-00-19700-01-1200-00001 25 ADAMSVILLE RD GREENWOOD Manuf Home/Ret.Title 2001.0 1773.0 53.020668 80.04
3 6-00-19600-01-5600-00001 143 LIDEN SCHOOL RD GREENWOOD Single-Fam Residential 935.0 6145.0 13.206215 37.40
4 6-00-19700-01-5900-00001 4299 MT HOLLY ROAD GREENWOOD Vacant Land 89.0 2226.0 3.844492 3.56
... ... ... ... ... ... ... ... ...
83137 1 291.0 94103.0 0.308282 11.64
83138 1 291.0 94103.0 0.308282 11.64
83139 1 291.0 94103.0 0.308282 11.64
83140 2-05-06818-01-2402-00001 2472.0 4624.0 34.836528 98.88
83141 2-05-06818-01-2400-00001 3500 S CLARK ST ARLINGTON Miscellaneous Imps. 19834.0 2270.0 89.730366 793.36

83142 rows × 8 columns

Symbolize the parcels

In [23]:
m1 ='Deleware, kent county')