Homer and others (2007) indicate that approximately 76 percent of the continental United States is classified as having less than 1 percent impervious coverage, 11 percent as having 1 to 10 percent impervious coverage, 4 percent as having estimated impervious coverage of 11 to 20 percent, 4.4 percent as having estimated impervious coverage of 21 to 40 percent, and approximately 4.4 percent as having estimated impervious coverage greater than 40%.

Total impervious area (TIA), commonly referred to as impervious coverage (IC) in calculations, can be expressed as a fraction (zero to one) or a percentage. There are many methods for estimating TIA, including (for the United States) the use of the National Land Cover Data Set (NLCD)[1]​ with a geographic information system (GIS), land use categories with categorical estimates of TIA, a generalized percentage of urbanized area, and relationships between population density and TIA.

The US NLCD Impervious Surface Data Set can provide a high-quality, nationally consistent land cover data set in a geographic information system (GIS) compatible format that can be used to estimate the TIA value. The NLCD consistently quantifies the percentage of anthropogenic TIA for the NLCD at 30 meter (900 m²) pixel resolution across the country.

Within the data set, each pixel is quantized with a TIA value that varies between 0 and 100 percent. The TIA estimates made with the NLCD array represent an aggregated TIA value for each pixel rather than a TIA value for an individual impervious feature.

For example, a 2-lane road in a grass field has a TIA value of 100 percent, but the pixel containing the road would have a TIA value of 26 percent. If the road extends (equally) along the two-pixel boundary, each pixel would have a TIA value of 13 percent. Data quality analysis of the NLCD 2001 set with manually delimited TIA sample areas indicates that the average error of predicted versus actual TIA can vary between 8.8 and 11.4 percent.

TIA estimates from land uses are made by identifying categories of these uses for large areas of land, adding the total area of ​​each category and multiplying each area by a characteristic TIA coefficient. These categories are commonly used to estimate TIA because areas with common land use can be identified from field surveys, maps, planning and zoning information, and remote imagery.

Land use coefficient methods are commonly used because planning and zoning maps that identify similar areas are increasingly available in GIS formats. Additionally, these uses are taken into account to estimate the effects on the TIA of the planned development with planning maps that quantify the projected changes in these uses.

There are substantial differences between real and estimated TIAs in different studies in the literature. Terms like "low density" and "high density" can have different meanings in different areas. A residential density of half an acre (an acre is approximately 40% of a hectare) per house can be classified as high density in a rural area, medium density in a suburban area, and low density in an urban area. Granato (2010)[2]​ provides a table with TIA values ​​for different land uses based on 30 studies.

The percentage of urbanized area (PDA) is commonly used to estimate the TIA manually through maps. The Multi-Resolution Land Features Consortium (MRLCC) defines an urbanized area as being covered by at least 30% built materials. Southard (1986) described undeveloped areas as natural, agricultural, or dispersed residential. Using the log power function with data from 23 river basins in Missouri, he was able to arrive at a regression equation to predict the TIA using the percentage of urbanized area (Table 6-1). He noted that this method was advantageous because large basins could be quickly delineated and TIA estimated by hand based on available maps.

Granato (2010)[2]​ developed another regression equation using data from 262 river basins in 10 metropolitan areas of the continental United States with watersheds ranging from 0.35 to 216 square miles and PDA values ​​ranging from 0.16 to 99.06 percent.

The TIA can also be calculated from population density data, estimating the population in the area of ​​interest and using regression equations to calculate the associated TIA. In the United States, population density can be used because consistent, national-scale census block data are available in GIS formats. If this information is available in other countries, this system can be used. Population density can also be used to predict the potential effects of future urbanization. Although there may be substantial variation in the relationships between population density and TIA, the precision of such estimates tends to improve with increasing drainage area as local variations are averaged out.

Granato (2010) [2]​ provides a table with 8 population density relationships from the literature and a new equation developed using data from 6,255 river basins in the United States Geological Survey (USGS) GAGESII (Geospatial Attributes of Gages for Evaluating Streamflow) data set. Granato (2010)[2]​ also provides 4 equations to estimate the TIA from housing density, which is related to population density.

The TIA can also be estimated from imperviousness maps extracted by remote sensing. Remote sensing has been widely used to detect impervious surfaces.[3]​[4]​ Detection of impervious areas using deep learning (an artificial intelligence technique) coupled with satellite imagery has emerged as a transformative method in remote sensing and environmental monitoring. Deep learning algorithms, particularly convolutional neural networks (CNN), have revolutionized the ability to identify and quantify impervious surfaces from high-resolution satellite images. These algorithms can automatically extract complex spatial and spectral features, allowing them to distinguish impermeable from permeable surfaces with high accuracy.[5][6][7]​.

Naturally impervious areas are defined here as land covers that can contribute a substantial amount of runoff during small and large storms, but are typically classified as permeable areas. Naturally impervious areas are not typically considered a major source of stormwater in most highway and urban runoff quality studies, yet they can produce a substantial amount of water. These natural impervious areas can include open water, wetlands, rock outcrops, barren soils (natural soils with low permeability), and areas of compacted soils. Natural impervious areas, depending on their type and background conditions, can produce significant stormwater flows from excess infiltration, saturation surface flow, or direct precipitation. The effects of natural impervious areas on runoff generation are expected to be more important in low TIA areas than in highly urbanized areas.

The National Land Cover Data Set (NLCD) provides percentages of impervious land that can be used as a qualitative measure of the prevalence of different land covers that can act as natural impervious areas.

Open water (for example a lake) can act as a natural impermeable area if rain in an area reaches it quickly through channels.

Wetlands can act as a natural impermeable area during storms if the flow escaping from them through underground aquifers and surface streams is significantly less than the amount of stormwater.

Unvegetated riparian areas can act as a natural impervious area during storms because, when under normal vegetation, these areas are a sink for excess surface flows.

Apparently permeable areas that have been affected by urbanization activities can act as impermeable areas and generate excess surface flows. These stormwaters can occur even during storms that do not meet precipitation volume or intensity criteria to produce runoff based on nominal infiltration rates.

Partially urbanized permeable areas can behave as impermeable areas, because urbanization and subsequent use tend to compact soils and reduce infiltration rates. For example, Felton and Lull (1963) measured infiltration rates of forest soils and turfgrasses and indicated a possible 80 percent reduction in infiltration as a result of development activities. Similarly, Taylor (1982) conducted infiltrometer tests in areas before and after urbanization and observed that topsoil disturbance and compaction from construction activities reduced infiltration rates by more than 77 percent.

This article incorporates public domain material from websites or documents of the United States Geological Survey and the Federal Highway Administration.

Homer and others (2007) indicate that approximately 76 percent of the continental United States is classified as having less than 1 percent impervious coverage, 11 percent as having 1 to 10 percent impervious coverage, 4 percent as having estimated impervious coverage of 11 to 20 percent, 4.4 percent as having estimated impervious coverage of 21 to 40 percent, and approximately 4.4 percent as having estimated impervious coverage greater than 40%.

Total impervious area (TIA), commonly referred to as impervious coverage (IC) in calculations, can be expressed as a fraction (zero to one) or a percentage. There are many methods for estimating TIA, including (for the United States) the use of the National Land Cover Data Set (NLCD)[1]​ with a geographic information system (GIS), land use categories with categorical estimates of TIA, a generalized percentage of urbanized area, and relationships between population density and TIA.

The US NLCD Impervious Surface Data Set can provide a high-quality, nationally consistent land cover data set in a geographic information system (GIS) compatible format that can be used to estimate the TIA value. The NLCD consistently quantifies the percentage of anthropogenic TIA for the NLCD at 30 meter (900 m²) pixel resolution across the country.

Within the data set, each pixel is quantized with a TIA value that varies between 0 and 100 percent. The TIA estimates made with the NLCD array represent an aggregated TIA value for each pixel rather than a TIA value for an individual impervious feature.

For example, a 2-lane road in a grass field has a TIA value of 100 percent, but the pixel containing the road would have a TIA value of 26 percent. If the road extends (equally) along the two-pixel boundary, each pixel would have a TIA value of 13 percent. Data quality analysis of the NLCD 2001 set with manually delimited TIA sample areas indicates that the average error of predicted versus actual TIA can vary between 8.8 and 11.4 percent.

TIA estimates from land uses are made by identifying categories of these uses for large areas of land, adding the total area of ​​each category and multiplying each area by a characteristic TIA coefficient. These categories are commonly used to estimate TIA because areas with common land use can be identified from field surveys, maps, planning and zoning information, and remote imagery.

Land use coefficient methods are commonly used because planning and zoning maps that identify similar areas are increasingly available in GIS formats. Additionally, these uses are taken into account to estimate the effects on the TIA of the planned development with planning maps that quantify the projected changes in these uses.

There are substantial differences between real and estimated TIAs in different studies in the literature. Terms like "low density" and "high density" can have different meanings in different areas. A residential density of half an acre (an acre is approximately 40% of a hectare) per house can be classified as high density in a rural area, medium density in a suburban area, and low density in an urban area. Granato (2010)[2]​ provides a table with TIA values ​​for different land uses based on 30 studies.

The percentage of urbanized area (PDA) is commonly used to estimate the TIA manually through maps. The Multi-Resolution Land Features Consortium (MRLCC) defines an urbanized area as being covered by at least 30% built materials. Southard (1986) described undeveloped areas as natural, agricultural, or dispersed residential. Using the log power function with data from 23 river basins in Missouri, he was able to arrive at a regression equation to predict the TIA using the percentage of urbanized area (Table 6-1). He noted that this method was advantageous because large basins could be quickly delineated and TIA estimated by hand based on available maps.

Granato (2010)[2]​ developed another regression equation using data from 262 river basins in 10 metropolitan areas of the continental United States with watersheds ranging from 0.35 to 216 square miles and PDA values ​​ranging from 0.16 to 99.06 percent.

The TIA can also be calculated from population density data, estimating the population in the area of ​​interest and using regression equations to calculate the associated TIA. In the United States, population density can be used because consistent, national-scale census block data are available in GIS formats. If this information is available in other countries, this system can be used. Population density can also be used to predict the potential effects of future urbanization. Although there may be substantial variation in the relationships between population density and TIA, the precision of such estimates tends to improve with increasing drainage area as local variations are averaged out.

Granato (2010) [2]​ provides a table with 8 population density relationships from the literature and a new equation developed using data from 6,255 river basins in the United States Geological Survey (USGS) GAGESII (Geospatial Attributes of Gages for Evaluating Streamflow) data set. Granato (2010)[2]​ also provides 4 equations to estimate the TIA from housing density, which is related to population density.

The TIA can also be estimated from imperviousness maps extracted by remote sensing. Remote sensing has been widely used to detect impervious surfaces.[3]​[4]​ Detection of impervious areas using deep learning (an artificial intelligence technique) coupled with satellite imagery has emerged as a transformative method in remote sensing and environmental monitoring. Deep learning algorithms, particularly convolutional neural networks (CNN), have revolutionized the ability to identify and quantify impervious surfaces from high-resolution satellite images. These algorithms can automatically extract complex spatial and spectral features, allowing them to distinguish impermeable from permeable surfaces with high accuracy.[5][6][7]​.

Naturally impervious areas are defined here as land covers that can contribute a substantial amount of runoff during small and large storms, but are typically classified as permeable areas. Naturally impervious areas are not typically considered a major source of stormwater in most highway and urban runoff quality studies, yet they can produce a substantial amount of water. These natural impervious areas can include open water, wetlands, rock outcrops, barren soils (natural soils with low permeability), and areas of compacted soils. Natural impervious areas, depending on their type and background conditions, can produce significant stormwater flows from excess infiltration, saturation surface flow, or direct precipitation. The effects of natural impervious areas on runoff generation are expected to be more important in low TIA areas than in highly urbanized areas.

The National Land Cover Data Set (NLCD) provides percentages of impervious land that can be used as a qualitative measure of the prevalence of different land covers that can act as natural impervious areas.

Open water (for example a lake) can act as a natural impermeable area if rain in an area reaches it quickly through channels.

Wetlands can act as a natural impermeable area during storms if the flow escaping from them through underground aquifers and surface streams is significantly less than the amount of stormwater.

Unvegetated riparian areas can act as a natural impervious area during storms because, when under normal vegetation, these areas are a sink for excess surface flows.

Apparently permeable areas that have been affected by urbanization activities can act as impermeable areas and generate excess surface flows. These stormwaters can occur even during storms that do not meet precipitation volume or intensity criteria to produce runoff based on nominal infiltration rates.

Partially urbanized permeable areas can behave as impermeable areas, because urbanization and subsequent use tend to compact soils and reduce infiltration rates. For example, Felton and Lull (1963) measured infiltration rates of forest soils and turfgrasses and indicated a possible 80 percent reduction in infiltration as a result of development activities. Similarly, Taylor (1982) conducted infiltrometer tests in areas before and after urbanization and observed that topsoil disturbance and compaction from construction activities reduced infiltration rates by more than 77 percent.

This article incorporates public domain material from websites or documents of the United States Geological Survey and the Federal Highway Administration.