![Percentage change of WUI area and wildfires with vegetation cover: (a) WUI area; (b) ignition points within WUI; (c) burned areas within WUI. The x axes represent the percentage of vegetation cover included in the WUI mapping. The colored dots represent the WUI-RS area, ignition points and wildfire perimeters included in the WUI-RS under different vegetation cover and buffer radius. To test the sensitivity of the WUI map, the vegetation cover changed by 10%\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$10\%$$\end{document} at a time, and its rate of change is on the secondary Y axis and represented by dash lines.](https://www.researchgate.net/profile/Mukesh-Kumar-57/publication/359779447/figure/fig1/AS:1142062770597894@1649300391953/Percentage-change-of-WUI-area-and-wildfires-with-vegetation-cover-a-WUI-area-b_Q320.jpg)
Access to this full-text is provided by Springer Nature.
Content available from Scientific Reports
This content is subject to copyright. Terms and conditions apply.
1
Vol.:(0123456789)
Scientic Reports | (2022) 12:5789 | https://doi.org/10.1038/s41598-022-09707-7
www.nature.com/scientificreports
Mapping the wildland‑urban
interface in California using remote
sensing data
Shu Li*, Vu Dao, Mukesh Kumar, Phu Nguyen & Tirtha Banerjee
Due to the mixed distribution of buildings and vegetation, wildland‑urban interface (WUI) areas are
characterized by complex fuel distributions and geographical environments. The behavior of wildres
occurring in the WUI often leads to severe hazards and signicant damage to man‑made structures.
Therefore, WUI areas warrant more attention during the wildre season. Due to the ever‑changing
dynamic nature of California’s population and housing, the update frequency and resolution of WUI
maps that are currently used can no longer meet the needs and challenges of wildre management
and resource allocation for suppression and mitigation eorts. Recent developments in remote
sensing technology and data analysis algorithms pose new opportunities for improving WUI mapping
methods. WUI areas in California were directly mapped using building footprints extracted from
remote sensing data by Microsoft along with the fuel vegetation cover from the LANDFIRE dataset in
this study. To accommodate the new type of datasets, we developed a threshold criteria for mapping
WUI based on statistical analysis, as opposed to using more ad‑hoc criteria as used in previous
mapping approaches. This method removes the reliance on census data in WUI mapping, and does not
require the calculation of housing density. Moreover, this approach designates the adjacent areas of
each building with large and dense parcels of vegetation as WUI, which can not only rene the scope
and resolution of the WUI areas to individual buildings, but also avoids zoning issues and uncertainties
in housing density calculation. Besides, the new method has the capability of updating the WUI map in
real‑time according to the operational needs. Therefore, this method is suitable for local governments
to map local WUI areas, as well as formulating detailed wildre emergency plans, evacuation routes,
and management measures.
e process of suburbanization in the United States, which has continued since World War II, has dramatically
increased the impact of human activities on natural ecosystems1,2. e urban sprawl caused by the migration of
population to the suburbs has intensied the ingression of human structures into wildlands, forests, and habitats3.
To monitor and evaluate the impact of these human activities on local climate and environment, the area where
human structures and wildland vegetation coexist either adjacent or interspersed with each other were dened
as Wildland-Urban Interface (WUI)4,5. In California, one of the most notable features of WUI is that they are
perceived as high-risk areas of human-caused wildres due to the accumulation of wildland vegetation, the
concentration of ammable human structures, and the strewing of sparks le by human activities6. Although
most wildres occur in uninhabited wildland, wildres ignited or spread into WUI areas pose a more signicant
threat to human lives and assets due to the proximity of human community and wildland fuels7–9. Besides, the
disturbance to local species and the introduction of invasive species caused by the construction and development
of human communities weaken local ecosystems’ resistance and resilience to wildres10. erefore, WUI area is
of great concern in wildre prevention and management.
e distribution of WUI in the United States is widespread and continues to rise. From 1990 to 2010, WUI
area in the contiguous United States increased rapidly from 7.2 to 9.5
%
, which caused a 9.6
%
increase in the
number of houses and a 8.5
%
increase in land area within in the new WUI, accompanied by the increase in
housing and population11. During the same time span, the WUI area in California increased from 26,263
km2
to 27,255
km2
, with an increase of about 3.8
%
. By 2010, the number of houses within the WUI had grown to 4.46
million, and the population had grown to 11.2 million in California, making it the state with the largest number
of houses and population in the WUI12. e growth of WUI not only increased the risk of re ignitions but also
increased the diculty of re extinction, because in these areas, reghters’ top priority is to protect people and
OPEN
Department of Civil and Environmental Engineering, University of California, Irvine, Irvine, CA 92697, USA. *email:
shul15@uci.edu
Content courtesy of Springer Nature, terms of use apply. Rights reserved
2
Vol:.(1234567890)
Scientic Reports | (2022) 12:5789 | https://doi.org/10.1038/s41598-022-09707-7
www.nature.com/scientificreports/
assets13. Given the above, accurately mapping and timely updating the WUI region to provide reghters and
emergency responders with more complete and practical WUI maps is of prime importance14.
Based on the denition of WUI from the Federal Register15, and the current common operational mapping
method16,17, areas with housing density greater than 6.17 houses/
km2
and vegetation cover greater than
50%
are
classied as WUI. e housing density is calculated using housing counts in Census blocks, and the threshold is
1 house/40 acre (6.17 houses/
km2
). e vegetation proportion is calculated by extracting vegetation area from
the National Land Cover Database (NLCD) within each Census block, the threshold of vegetation is
>50%
for
intermix and
>75%
for interface15. It means that the WUI area should meet requirements of high-density houses
surrounded by high-density wildland vegetation (50
%
of vegetation area in the Census block) or high-density houses
adjacent to, that is within 1.5 miles (2.4 km) of a large tract of contiguous wildland vegetation (75
%
of vegetation
area in the Census block).
e current common WUI mapping methods heavily rely on the threshold of building density and vegeta-
tion proportion. However, because of the limitations in the update-frequency (10 years for Census data) and
the precision of housing and vegetation data, there are uncertainties in the current data sources and mapping
methods, especially in calculating housing density17. Radelo etal.5 mapped WUI areas of all states in the
contiguous United States based on housing statistics provided by the Census data in 2010 and vegetation cover
provided by the NLCD. e minimum unit of housing density in this zonal mapping method was Census block,
and the zone modiable areal unit problem cannot be avoided17. Meanwhile, this method removes all the area
of public land to get a more precise housing density. However, there are also human structures on public lands,
such as power-lines and airports, and their eect is completely ignored. Subsequently, as an improvement to the
zonal mapping method, Platt18 used points mapped from the parcel centroids representing building locations
to calculate housing density in WUI mapping. e limitation of this method is that these points were not the
actual locations of the buildings and would still be erroneous. Bar-Massada etal.17 mapped WUI directly from
the buildings’ location and calculated housing density by “circular moving window analysis”. It divides the studied
region by grids rst and then overlaying a circular moving window to calculate the mean housing density for all
the grids in the window. e mean housing density would be assigned to the central grid. Although WUI maps
can be customized based on needs, dening appropriate window sizes in dierent spatial scale is a challenge.
Besides, traditional housing density data are updated at long intervals, as Census data are collected and updated
every ten years. e current update frequency of WUI maps is far behind its growth rate.
Eorts to digitize spatial housing locations and use them for WUI mapping or wildre risk assessment
include extracting building locations from GPS records19, and mapping structures from high-resolution digital
orthophotographs20. Moreover, with the improvement of the quality of remote sensing data in terms of acquisi-
tion eciency and resolution, it has become possible to extract detailed housing and vegetation boundaries
from such datasets. For example, Caggiano etal.21 explored the feasibility of detecting individual buildings from
National Aerial Image Program imagery using Object Based Image Analysis; Johnston and Flannigan22 used
structure locations from CanVec+, which were derived from a variety of remote sensing products, and wildland
fuels from Land Cover 2000 dataset, which was based on Landsat 5 and Landsat 7 iamges, to map the interface
areas in Canada; Alcasena etal.23 used remote sensing data provided by Spanish national topographic platform
(BTN25) map to identify the housing locations and the vegetaion map from land parcel identication system
to identify forest land polygons.
Over the past few years, Microso has made signicant eorts in applying deep learning, computer vision and
AI to mapping and leveraging the power of machine learning in analyzing satellite imagery to trace the shape of
buildings across the country. e building footprint dataset released by Microso in 2018 contains 129,591,852
buildings, covering the entire United States; and it is available to download free of charge. It used Deep Neural
Network and residual neural network (ResNet34) with segmentation techniques (ReneNet up-sampling) to
detect individual building footprints from their imagery data24. In terms of the vegetation data, the LANDFIRE
program from United States Geological Survey (USGS) utilizes Landsat 7 Enhanced ematic Mapper Plus and
Landsat 8 Operational Land Imager products to provide national scale vegetation, fuel, and re regime data25.
e fuel vegetation cover data from the LANDFIRE database not only update more frequently than NLCD maps,
but also directly provide pixel-level vegetation cover percentage, simplifying and reducing the computational
cost in WUI mapping.
Using the above-mentioned data, the WUI map in California as well as the WUI mapping method can be
updated. Our goal in this paper is to (1) use housing footprints from Microso and fuel vegetation data from
the LANDFIRE program, combining the denition of WUI and, based on the characteristics of the data, design
a practical WUI rendering method; (2) compare the new WUI mapping method against previous methods, and
analyze if there are improvements to the previous maps; (3) combining with the historical wildre record, analyze
the usefulness of the new WUI map in wildre risk assessment.
Results and discussion
Threshold determination. e threshold of the two main components in the WUI denition—vegetation
cover and the distance between housing and high-density vegetation, were tested using the one factor at a time
(OFAT) approach16. e validity of the map was evaluated by three indicators: the percentage change of WUI
area with changes in the parameter value, wildre ignition points within WUI and wildre perimeters within
WUI. When the percentage changes of the indicators are smaller than the percentage change of the WUI com-
ponents, the threshold can be seen as stable5. e results are shown in Fig.1. For the vegetation cover, when the
threshold is lower than
40%
, the change in the new WUI area is relatively stable; on the other hand, when the
threshold is lower than
50%
, the change in the ignition points and the perimeters are relatively stable. us
40%
of vegetation cover was selected as the threshold. More specically, the housing within or next to the wildland
Content courtesy of Springer Nature, terms of use apply. Rights reserved
3
Vol.:(0123456789)
Scientic Reports | (2022) 12:5789 | https://doi.org/10.1038/s41598-022-09707-7
www.nature.com/scientificreports/
area with vegetation coverage higher than
40%
were seen as WUI area. In terms of how close the housing should
be to the wildland vegetation, results show that the change of
50%
increase or decrease based on 2.4 km does
not have a drastic eect on the three validity indicators, and the range of percentage change in each indicator is
within
50%
. erefore, 2.4 km is reasonable both conceptually and operationally in this method. Since there is
no restriction on the density of houses in this method, the area of buer directly determines whether the houses
are included in WUI. Overall, WUI area in this new operational method (WUI-Remote Sensing, abbreviated as
WUI-RS) was dened as the area in which man made structures are surrounded by or within 2.4 km of wildlands
where the vegetation cover is higher than
40%
.
Latest WUI maps. Based on the above optimal thresholds, we mapped the WUI area in California using
Microso housing footprint data in 2018 and LANDFIRE fuel vegetation cover in 2018, which we designate as
remote sensing based WUI (WUI-RS). Since the WUI map generated with remote sensing data can be updated
at any time as long as we delineate the housing footprints from the satellite images, we did not label the year the
map was drawn.
ere are currently two common WUI maps used in California: one was released by Martinuzzi etal.12 in
2015 using Census data in 2010 NLCD data in 2006, hereinaer WUI-USFS (2010); the other was released by
the State of California and the Department of Forestry and Fire Protection (CAL FIRE) using several internal
data sources for the 2015 Assessment of Forest and Rangelands, hereinaer WUI-FRAP (2010). To compare our
mapping approach with the existing mapping approaches we have to overcome one obstacle which is that the
WUI-USFS is from 2010 as they used the 2010 Census data. In order to facilitate a fair comparison, we updated
the WUI-USFS map to 2020 as well. We followed the common zonal-based mapping method12,16 and used the
latest Census data in 2020 to map the WUI area in California as well and designate it WUI-USFS (2020).
e map of the WUI-RS is shown in Fig.2a. e WUI-RS map covers an area of 28,575
km2
in California, with
55.21
%
intermix area (15,776
km2
) and 44.79
%
interface area (12,799
km2
), accounting for 6.74
%
of the land area
in CA (423,971
km2
). About ve million housing units, which accounts for 45.13
%
of total housing in California
are included in the WUI-RS. e distribution of the new WUI is concentrated along the western coastline and
to the west of the Sierra Nevada Mountain range. It is sparse in the central and southeastern California, because
most of the San Joaquin Valley in the Central California have been developed and planted, and Southeastern
California has vast tracts of barren land and very few human structures.
At the county level, Fig.2b,c shows the area and percentage of WUI in each county. e San Diego (SD), Los
Angeles (LA) and Sonoma (SON) counties contains largest WUI area which are 1909
km2
, 1400
km2
, 1225
km2
separately. e Contra Costa (CC), Sacramento (SAC) and Santa Cruz (SCZ) counties in the northern California
have the highest percent of WUI which accounts for 36.18
%
, 33.64
%
and 31.44
%
of their land area separately.
e map of the WUI-USFS(2020) is shown in Fig.3a. e total area of the updated WUI-USFS map is about
29,343
km2
, including 70.22
%
intermix area (20,605
km2
) and 29.78
%
interface area (8738
km2
). It shows the
similar spatial patterns with the WUI-RS map, which are along the western coast and to the west of the Sierra
Nevada mountain, while the density of the WUI area is much lower than WUI-RS. e reason for this dierence
is that the new method put more emphasis on the presence of man made-structures rather than structure density.
us the WUI-RS map include the area with high density vegetation and low density human-structures as well.
Compared to the WUI map of California in 2010 released by USFS, the total area of WUI in 2020 increased
by approximately 2000
km2
, with the similar spatial distribution. e growth in WUI area over the decade was
mainly driven by an increase in the number of houses in California.
Figure3b,c show the statistics of updated zonal-based WUI area and their percentage of land area in each
county. ree counties in sourthern California: the San Bernardino (SBD), Riverside (RIV) and San Diego (SD)
county have the largest WUI area, especially compared to the WUI-RS map. However aer dividing the total
land area in each county, the spatial distribution of the percentage of the WUI areas, which is shown in Fig.3c,
Figure1. Percentage change of WUI area and wildres with vegetation cover: (a) WUI area; (b) ignition points
within WUI; (c) burned areas within WUI. e x axes represent the percentage of vegetation cover included
in the WUI mapping. e colored dots represent the WUI-RS area, ignition points and wildre perimeters
included in the WUI-RS under dierent vegetation cover and buer radius. To test the sensitivity of the WUI
map, the vegetation cover changed by
10%
at a time, and its rate of change is on the secondary Y axis and
represented by dash lines.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
4
Vol:.(1234567890)
Scientic Reports | (2022) 12:5789 | https://doi.org/10.1038/s41598-022-09707-7
www.nature.com/scientificreports/
has similar patterns with WUI-RS. e counties in the central California and on both side of the Sierra Nevada
Mountain have high percentage of WUI area.
e overlaps between the most recent WUI-RS map and WUI-USFS map is shown in Fig.4. e total area of
the overlap is 22,245
km2
, which accounts for 77.85
%
of the total area of WUI-RS and 75.81
%
of the total area
of WUI-USFS (2020). e blue areas in Fig.4 are the unique part in the WUI-RS map, which are concentrated in
central California, west of the Sierra Nevada Mountains. ey are not included in the WUI-USFS map because
of the low housing density. e orange areas in Fig.4 show the unique part in the WUI-USFS map, which are
concentrated in southern California. ese areas did not meet the vegetation density threshold of the WUI-RS
map. In general, the high percentage of overlaps between these two maps indicate the validity of the WUI-RS
map drawn by using the selected remote sensing data and the tested threshold in this study.
Comparison between dierent WUI maps. Dierent types of WUIs vary widely in their denitions
and applications, each of which has its own scope of application. We compared the rendering methods, applica-
Figure2. WUI-RS map in 2018 and the spatial distribution of WUI-RS area in California: (a) WUI-RS map in
2018; (b) WUI-RS area in each county; (c) percentage of WUI-RS cover in each county.
Figure3. WUI-USFS map in 2020 and the spatial distribution of WUI-USFS area in California: (a) WUI-USFS
map in 2020; (b) WUI-USFS area in each county; (c) percentage of WUI-USFS cover in each county.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
5
Vol.:(0123456789)
Scientic Reports | (2022) 12:5789 | https://doi.org/10.1038/s41598-022-09707-7
www.nature.com/scientificreports/
tions and spatial patterns of WUI-RS, WUI-USFS (2020), WUI-USFS (2010) and WUI-FRAP (2010) in Table1
and Fig.5. e comparison here is to demonstrate the dierences and analyze the causes. Additionally, the extent
and coverage of dierent WUI maps are dependent to some extent on the criteria used to map them in the rst
place. However, in our mapping approach the selection of the threshold criteria are not completely arbitrary and
based on statistical approaches such as the OFAT technique as discussed earlier.
In general, all types of WUI areas are concentrated along the Sierra Nevada Mountain and the West Coast.
ese common features demonstrate that although the new mapping method simplies the operation steps, it
still successfully captures the distribution characteristics of WUI. In Fig.5e,f, we selected two representative con-
centrated WUI regions in northern and southern California, presenting two types of general spatial distribution
dierences among the WUI maps. Figure5e shows an area in Northern California which has small fragments
of WUI-RS but not the other types of WUI. is region belongs to the Eldorado National Forest, dominated
by high density of trees. Despite the low housing density, the wildre ignition points were concentrated here
between from 2000 to 2019, and there have been large wildres in the two decades. e WUI-RS map includes
such low-density houses surrounded by high-density vegetation, supplementing these piecemeal WUI areas with
potentially high wildre risk that would have been missed by previous methods. Besides, Fig.5f shows evident
dierences in dierent types of WUI close to the developed areas in Southern California which are dominated
by shrubs. In particular, the northeastern region close to the mountains in this gure was treated as non-WUI
in the WUI-RS map. is is due to the fact that most of the shrubs cover in this area are under 40
%
and are not
continuous. At the same time, the vast developed area impede this whole region being classied as high-density
wildland vegetation. From the perspective of wildland re risk, no human-caused wildres ignited in this area
from 2010 to 2019, thus this dierence is deemed acceptable.
Compared with WUI mapping methods of WUI-USFS and WUI-FRAP predening a threshold of housing
density, WUI-RS treated each house independently, as the location and the perimeter of houses are known from
the building footprint data. e current common WUI mapping methods follow the denition of WUI from
the Federal Register, adopting 6.17 houses per
km2
as the minimum housing density threshold in WUI, which
eliminate the areas with low density buildings adjacent to or within wildland. However, the study by Syphard26
demonstrate that the smaller and more isolated housing clusters have higher risk of property loss due to wildres.
erefore we treated all the buildings within a certain distance of high-density fuel vegetation cover as WUI in
this study. is treatment also eliminates the requirement of housing density calculation which is complicated by
zoning issues and the uncertainty due the the variation of housing density threshold16. Using the fuel vegetation
cover (FVC) data from LANDFIRE directly, our new method also gets rid of the vegetation density calculation.
To contextualize the WUI mapping with historical wildre activity, the number of historical wildre ignition
points and the burned area within WUIs were calculated within ten years of each of the three maps being most
eective. As shown in Table2, WUI-RS captures the majority of human-caused ignition points, and it has the
highest percentage to the total ignition points among four WUI maps. In terms of the burned area within WUI,
none of the four maps included much of human-caused wildre area (less than 10
%
in general). is is because
WUI maps are primarily used to identify houses and human structures with a high risk of igniting wildres or
Figure4. Overlaps and dierences between WUI-RS and WUI-USFS (2020).
Content courtesy of Springer Nature, terms of use apply. Rights reserved
6
Vol:.(1234567890)
Scientic Reports | (2022) 12:5789 | https://doi.org/10.1038/s41598-022-09707-7
www.nature.com/scientificreports/
be aected by wildres. When the burned area of a wildland re overlaps with the WUI perimeter, it means
that the wildre has caused damage to a human community, which has been increasing in recent times but still
fairly uncommon9. e majority of wildres were still conned to the wildland. Speaking of the risk of human
community being eected by wildres, the number of houses contained within the WUI was also calculated, and
the results show that WUI-RS contains the highest proportion of houses, and the housing percentage in WUI-
USFS is close to that in WUI-RS. Given that houses in WUI areas are at a high risk of rebrand ignition, these
results demonstrate that the WUI-RS can be used to compute the risk from wildland res near communities.
Conclusion
e development of remote sensing technologies and the wide applications of remote sensing data provides
opportunities to improve the accuracy and update frequency of WUI maps. So far, remote sensing data with 30
m resolution has been readily available, with the highest resolution reaching up to 3–5 m, which is enough to
clearly delineate the boundary of man-made structures and rendering ne-scale WUI maps. With the ability to
Table 1. Comparison of dierent WUI mapping methods.
WUI-RS (Our approach) WUI-USFS WUI-FRAP
Update frequency In real time 10 years (limited by Census data) Only released one map for
the 2015 Assessment of Forest
and Rangelands
Inputs Housing footprints (Microso);
fuel vegetation cover (LANDFIRE)
Housing density (Census housing
data);
wildland vegetation (NLCD)
Housing density;
Fire Hazard Severity Zones;
Unimproved Parcels;
and Vegetation Cover
(all the data are from other
FRAP programs)
Operation denition
e area in which structures are
surrounded by (interface) or
within (intermix) 2.4 km of wild-
lands
where the vegetation cover
≥
40
%
(obtained using statistical analysis)
Interface:
at least one structure per 40 acre,
and located <2.4 km of an area
≥
5
km2
in size that is
≥
75
%
vegetated
Intermix:
at least one structure per 40 acre,
and wildland vegetation cover
≥
50
%
Interface:
at least one structure per 20 acre;
in moderate, high, or very high Fire
Hazard Severity Zone;
Not dominated by wildland vegeta-
tion;
Spatially contiguous groups of 30m
cells that are 10 acres and larger
Intermix:
not interface;
one structure per 20 acres to one
structure per 5 acres;
at least one structure per 5 acres and
dominated by wildland vegetation;
In moderate, high or very high Fire
Hazard Severity Zone;
Improved parcels only;
Spatially contiguous groups of 30m
cells that are 25 acres and larger
Scope of application
Provide the spatial patterns of WUI
accurate to individual houses in
California, pinpoint specic WUI
areas to show the re risk and
develop
targeted management strategies
Provide the extent and locations of
WUI
areas at the state level, and sum-
marize
the statistics at the state to federal
level
Provide the overall pattern of WUI
development at the county level in
California, and compare counties in
terms of development patterns
Advantages
(1). e operation denition is
straight-
forward and simplied, eliminat-
ing the
calculation of housing density and
vegetation cover;
(2). WUI-RS can be updated in real-
time as needed;
(3). WUI-RS is accurate to an
individual
building and and removes the
reliance
on ad-hoc thresholding criteria.
(1). e input data were available
and consistent nationwide
(2). It is feasible to provide the large-
scale statistics summaries
(1). Take re hazards in to account
in mapping process
(2). Instead of using a single housing
density threshold, it renes the
housing
density into four levels.
Limitations
(1). e vegetation cover data were
derived
from the LANDFIRE database
directly,
we did not calibrate it in this
research;
(2). e new set of thresholds in
mapping
method only applicable in California,
mapping other areas will require
area specic calibration.
(1). e maps are not directly
comparable
with those from earlier decades
due to
changes in census block
boundaries12;
(2). e shape and area of census
blocks
are inconsistent, and it will introduce
bias17;
(3). e low density development
areas
were not factored into the mapping
process18.
(1). Until the dataset is rened
through
a eld review process, it is not suited
for WUI designations for individual
houses or neighborhoods;
(2). Both the input data and the
mapping
method are applicable only in
California.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
7
Vol.:(0123456789)
Scientic Reports | (2022) 12:5789 | https://doi.org/10.1038/s41598-022-09707-7
www.nature.com/scientificreports/
obtain high-resolution satellite images, WUI maps are expected to be generated in real time and automatically,
which is also the direction of this research in the future.
e new WUI mapping method proposed in this research goes back to the essential denition of WUI, identi-
es appropriate thresholds for WUI operational denition, and simplies the requirements of WUI mapping by
using modern advanced databases. e vegetation cover of 40
%
and the distance of 2.4 km between the wildland
vegetation and the housing were validated as the threshold for WUI mapping.
In the new WUI map (WUI-RS), the WUI area in the San Diego county is prominent. When it comes to WUI
percentage, the counties with the highest percentage of WUI are concentrated in northern California, west of the
Sierra Nevada Mountains. Compared to the two previous WUI maps from USFS, the new WUI map overlaps with
them, but with the addition of low density housing clusters surrounded by high-density vegetation, which have
Figure5. Comparison between (a) WUI-RS, (b) WUI-USFS (2020), (c) WUI-USFS (2010) and (d) WUI-
FRAP (2010). e detailed comparison among these four maps of two selected enlarged regions are shown in
(e) and (f). Maps were generated by ArcGIS Pro 2.4.0 (https:// www. esri. com/ en- us/ arcgis/ produ cts/ arcgis- pro/
overv iew).
Table 2. Wildre ignition points, burned area and housing within dierent WUIs in CA. Wildre ignition
points from USFS and wildre perimeters from CAL FIRE were extracted to evaluate the percentage of
wildres ignited or burned within dierent WUI maps. WUI-RS and WUI-USFS (2020) used wildre data in
2010–2019, WUI-USFS (2010) and WUI-FRAP(2010) used wildre data in 2000–2019.
Total area (km
2
)Number of res
igniting in WUI Percentage of res
igniting in WUI (
%
)Burned area in WUI
(km
2
)
Percentage of
burned area in WUI
(
%
)Number of houses
in WUI Percentage of
houses in WUI (
%
)
WUI-RS 28,575 4277 86.25 524.45 4.09 4,959,028 45.13
WUI-USFS (2020) 29,343 4523 66.50 963.29 7.16 4,841,685 43.28
WUI-USFS (2010) 27,255 4414 65.71 951.06 6.70 4,457,884 40.57
WUI-FRAP (2010) 9581 2877 42.83 3.9 0.03 709,198 6.45
Content courtesy of Springer Nature, terms of use apply. Rights reserved
8
Vol:.(1234567890)
Scientic Reports | (2022) 12:5789 | https://doi.org/10.1038/s41598-022-09707-7
www.nature.com/scientificreports/
been shown to be at a higher wildre risk than areas with high-density houses. Besides, the WUI-RS captures
the highest percentage of ignition points of human-caused wildres and highest percentage of housing among
the three maps, demonstrating its capability to be used for wildre risk calculation. rough the detailed re
risk analysis within WUI areas, we can conclude that the WUI map generated using our proposed method have
the ability to capture the area with high re risks.
e innovation of this approach is mainly reected in three aspects: (1) Using remote sensing data to map
WUI removes the reliance on census data which is only available every ten years and cannot be updated fre-
quently to keep up with the pace of rapid WUI proliferation. (2) e ability to map individual household foot-
prints provides us the ability to map WUI much more precisely and changes the paradigm from previous WUI
mapping approaches which could only locate zones based on housing density as opposed to individual buildings.
Using our approach, local authorities will be able to identify wildre risk down to the level of individual streets
and buildings. Although the use of Microso building footprint data in WUI mapping has been proposed by
other studies27, their approach only extracted building locations to calculate the housing density. Moreover,
they still followed the existing WUI mapping method in the US. In contrast, we used building footprint data
in WUI mapping directly allowing for more precise mapping. (3) We developed the threshold criteria for map-
ping WUI based on statistical analysis (the OFAT approach) to accommodate the new type of data as opposed
to using ad-hoc criteria as used in previous mapping approaches. In this way, WUI maps can incorporate the
presence of man made-structures within or at the vicinity of high density vegetation areas that are characterized
by high risks of igniting res regardless of the density of structures. Mapping the WUI areas directly using the
housing footprints makes both the new method and the new map intuitive and easier to interpret. Removing the
housing calculation not only simplies the WUI mapping process signicantly, but also gets rid of the zoning
issue that both the zonal-based and the point-based mapping methods would be troubled by. e rapid update
frequency of remote sensing data also makes it possible to update WUI maps frequently without having to rely
on the availability of census and zoning data. In terms of the wildland vegetation identication, we collected the
fuel vegetation cover from the LANDFIRE database, which provides the vegetation cover in percentage directly
based on the satellite image using deep learning algorithm. Adopting this dataset we remove the requirement for
vegetation density calculation and do not need to check the land use classications or the ownership of lands.
While the new method greatly simplies the calculation of WUI, especially eliminating the need to calculate
housing density and vegetation cover, we did not calibrate the vegetation cover data from LANDFIRE. In addi-
tion, the new threshold criteria for housing density and vegetation cover developed in this study is only applicable
to California, and it needs to be specically calibrated if mapping in other regions. erefore, the new method
is suitable for mapping high-precision WUI maps to assist in developing targeted management strategies, but is
not ideal for large spatial scale (national) WUI mapping and comparison. Meanwhile, the new WUI map cannot
be directly compared with the previous WUI maps to show the increase or decrease of WUI area or the change
of the spatial pattern.
Considering the application of the WUI map in re risk visualization, in the future, re occurrence statistics
can also be incorporated into WUI mapping. In this way, re hazard can be classied more intuitively. Leverag-
ing the rapid development of statistical methods, especially the development of semantic segmentation, using
machine learning techniques to streamline and automate the WUI identication process would be a promising
future direction. It will reduce computational cost signicantly and improve the updating eciency of WUI
maps. Besides, although human activities have become one of the main causes of wildres, wildre risk is not a
simple combination of human activities and ammable fuels. In our future research, we will combine multiple
data layers such as wildre risk, human activities and ammable fuels, coupled with the WUI mapping approach
developed in this work, which will be easier to apply and generalize on a national scale. In addition, WUI maps
could have other applications beyond determining wildre risk, such as analyzing de-urbanization trends or
observing the proliferation of invasive plants. For WUI maps with dierent application purposes, corresponding
data or layers can be added in the mapping process to enhance its pertinence.
Methods
Housing footprint data. One of the most important components of WUI is the presence of human popu-
lation, which was represented using housing footprints in this study. e housing footprint data in GeoJSON
format were obtained from Microso United States building footprints database released in 2018. It contains
10,988,525 computer generated building footprints in California, which were extracted from satellite imagery
by semantic segmentation and converted to polygons by polygonization24. Based on this algorithm, aerial data
can also be used to identify building footprints in the future. us, remote sensing data in this research refers to
the aerial and satellite images of the Earth taken from the air and from space. Due to the large size of GeoJSON
format data (2537 MB), it was rst split into multiple les in Python before converting to shapeles in the so-
ware ArcGIS Pro (hereinaer ArcGIS), then the shapele fragments were merged into a complete le. To reduce
the computation time and cost, the polygons of housing footprints in the shapele were converted to 5 meter
resolution rasters in the mapping process. e distribution of housing footprints in the entire state of California
is shown in Fig.6a, the inset zoom into a random human community, showing the detailed shape and the layout
of extracted housing footprint.
Vegetation data. To simplify the calculation of vegetation density, the fuel vegetation cover in 2018 from
the Landscape Fire and Resource Management Planning Tools (LANDFIRE)28 were used as the vegetation infor-
mation. e 30-m resolution fuel vegetation cover layer (FVC) from the LANDFIRE program adopted “plot-
level ground-based visual assessments and lidar observations”, providing the information of the canopy cover of
herbaceous, shrub and tree in percentages28. Simplied high-density wildland vegetation proles can be deline-
Content courtesy of Springer Nature, terms of use apply. Rights reserved
9
Vol.:(0123456789)
Scientic Reports | (2022) 12:5789 | https://doi.org/10.1038/s41598-022-09707-7
www.nature.com/scientificreports/
ated by selecting an appropriate threshold of vegetation cover percentage. e percentage from 20 to 80
%
with
stride of 10
%
were tested in this study by sensitivity analysis. Moreover, the land cover information except for
the fuel vegetation, such as the developed area, the barren land, the open water and the snow/ice etc., were also
included, which were used to identify the non-wildland area. e distributions of the original FVC as well as the
wildland and non-wildland areas are shown in Fig.6b, the inset zoom into the same resolution as the inset in
Fig.6b, showing the detailed distribution of fuel vegetation within the same region.
WUI mapping method. Based on the denition of WUI from the Federal Register15 and the operational
denitions from the research of Stewart etal.16 and the research from Kumar etal.9, WUI refers to the area where
the houses are adjacent to (interface) or surrounded by (intermix) high-density fuel vegetation cover. us,
WUI maps in this study delineated the perimeters of the WUI area and show the classication of intermix and
interface area.
WUI was determined based on the presence of houses, the percentage of vegetation cover and the distance
between houses and high-density vegetation. To determine the threshold of fuel vegetation cover density and the
distance from housing footprints to high-density vegetation, the one factor at a time (OFAT) method was adopted
to evaluate the eect of their variations on the total WUI area and the percentage of the wildre ignitions and
burned area contained in the corresponding WUI to the total ignitions and burned area5,16. e vegetation cover
percentage were tested from greater than 10
%
to greater than 90
%
with stride of 10
%
. In terms of the distance
from vegetation to housing footprint, conceptually, the distance of 2.4 km (1.5 mile) is appropriate because it
represents statistically how far the rebrands can y from the re front and was evaluated and used in several
previous studies5,29. To prove that the 2.4 km is still applicable in this method, the distance of 1.2 km and 4.8 km
were also tested to show the variation in the resultant WUI.
When mapping the WUI, the high-resolution distribution of fuel vegetation cover was extracted from the
LANDFIRE dataset based on the selected threshold at the beginning. To exclude the parks, recreational green
spaces and green belts in the urban area, the vegetation area with a continuous area of less than 5
km2
were
removed. Subsequently, buers with 2.4 km (or 1.2 km, 4.8 km) radius were created around the vegetation prole.
Selecting all the houses located in the high-density vegetation cover (intermix) and buer zone (interface), the
new WUI map was obtained aer aggregating the dispersed houses to a continuous area, and it is referred to as
WUI-Remote Sensing (abbreviated as WUI-RS). e owchart of mapping process is shown in Fig.7.
Figure6. Housing footprint and fuel vegetation cover data in California: (a) Housing footprint distribution
from Microso United States (US) building footprints database, the inset shows the detailed housing layout
and shape; (b) fuel vegetation cover distribution from LANDFIRE database. e dark blue boundary represents
the simplied wildland vegetation boundary which was delineated by selecting pixels with vegetation cover
percentage of 50
%
or higher, the inset shows the detailed distribution of fuel vegetation. Maps were generated by
ArcGIS Pro 2.4.0 (https:// www. esri. com/ en- us/ arcgis/ produ cts/ arcgis- pro/ overv iew).
Content courtesy of Springer Nature, terms of use apply. Rights reserved
10
Vol:.(1234567890)
Scientic Reports | (2022) 12:5789 | https://doi.org/10.1038/s41598-022-09707-7
www.nature.com/scientificreports/
e new WUI map (WUI-RS) generated in this study was compared to the WUI data products from 2010
(when the last Census data were available) developed by the United States Forest Service (WUI-USFS (2010))
and the California Department of Forestry and Fire Protection’s Fire and Resource Assessment Program (WUI-
FRAP (2010)). Using Census data in 2020, we also updated the traditional WUI map following the mapping
method used by USFS (WUI-USFS (2020)). e last three maps used threshold of housing density as one of the
basic criteria. e criterion in WUI-USFS is 6.17 houses per
km2
, which is equivalent to one house per 40 acres,
whereas the minimum value of housing density within WUI-FRAP is one house per 20 acres. Regarding wildland
vegetation, WUI-USFS calculated the percentage of vegetation in each Census block using NLCD data, while
WUI-FRAP used the layer of vegetation cover and re hazard zone from CAL FIRE to determine whether the
area is dominated by vegetation. However, the WUI-FRAP added a new classication of “Inuence zone”, which
involved the vegetation within 1.5 miles (2.4 km) from the interface and the intermix. In general, the criteria
for WUI-FRAP are more stringent than that for WUI-USFS. Consequently the total WUI area in California in
the WUI-USFS (2010) is 27,255
km2
while in the WUI-FRAP (2010) is 9581
km2
. And the area including the
inuence zone in the WUI-FRAP (2010) is 71,609.22
km2
.
Geospatial analysis methods. To show the spatial distribution of WUI area in each county in California
and provide basic statistics information, we split the WUI area along California county boundary. WUI areas
in each county were counted and the percentage of WUI areas in each county were calculated by dividing WUI
areas by the county area.
To show the consistency and inconsistency between the WUI-RS and WUI-USFS (2020), the overlaps between
these two maps were extracted. Also the proportion of overlap areas in the two maps were calculated by dividing
the overlap areas by total WUI areas.
To evaluate the re risk and its trend in WUI areas, the wildre ignition points from USFS30 and the wild-
re perimeters from CAL FIRE31 in California were plotted in the maps. e re ignition point database from
USFS dates back to 1970. It includes the ignition points of individual wildres under the jurisdiction of USFS.
e burned area of individual res varies from smaller than 0.01–410,203 acres (1660
km2
). e re perimeter
Figure7. Flowchart of WUI-RS mapping method. FVC represents the fuel vegetation cover and WUI
represents the wildland urban interface.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
11
Vol.:(0123456789)
Scientic Reports | (2022) 12:5789 | https://doi.org/10.1038/s41598-022-09707-7
www.nature.com/scientificreports/
database from CAL FIRE collected large wildres under the jurisdiction of CAL FIRE and USFS from 1950.
e minimum burned area of the collected wildres is 10 acres. Since the WUI area has grown rapidly in the
last decade since 199011, the 2000–2009 wildres were used to calculate the percentage of wildres within the
WUI area for the 2010 WUI maps (WUI-USFS and WUI-FRAP), while the new WUI map (WUI-RS) used the
2010–2019 wildres. e spatial distributions of wildre ignition points and perimeters between 2000–2009
and 2010–2019 are shown in Fig.8.
Received: 28 July 2021; Accepted: 25 March 2022
References
1. Hammer, R. B., Stewart, S. I. & Radelo, V. C. Demographic trends, the wildland-urban interface, and wildre management. Soc.
Nat . Re sour. 22, 777–782 (2009).
2. Simon, G. L. Vulnerability-in-production: a spatial history of nature, auence, and re in oakland, california. Ann. Assoc. Am.
Geogr. 104, 1199–1221 (2014).
3. Zhang, Y., He, H. S. & Yang, J. e wildland-urban interface dynamics in the southeastern us from 1990 to 2000. Landsc. Urban
Plan. 85, 155–162 (2008).
4. Stein, S.M. etal. Wildre, wildlands, and people: understanding and preparing for wildre in the wildland-urban interface-a
forests on the edge report. Gen. Tech. Rep. RMRS-GTR-299. Fort Collins, CO. US Department of Agriculture, Forest Service, Rocky
Mountain Research Station. 36 p.299 (2013).
5. Radelo, V. C. et al. e wildland-urban interface in the united states. Ecol. Appl. 15, 799–805 (2005).
6. Syphard, A. D. et al. Human inuence on california re regimes. Ecol. Appl. 17, 1388–1402 (2007).
7. Mercer, D. E. & Prestemon, J. P. Comparing production function models for wildre risk analysis in the wildland-urban interface.
Forest Policy Econ. 7, 782–795 (2005).
8. Kramer, H. A., Mockrin, M. H., Alexandre, P. M., Stewart, S. I. & Radelo, V. C. Where wildres destroy buildings in the us relative
to the wildland-urban interface and national re outreach programs. Int. J. Wildland Fire 27, 329–341 (2018).
9. Kumar, M., Li, S., Nguyen, P. & Banerjee, T. Revisiting the existing denitions of wildland-urban interface for california. Earth
and Space Science Open Archive (2020).
10. Syphard, A. D., Radelo, V. C., Hawbaker, T. J. & Stewart, S. I. Conservation threats due to human-caused increases in re frequency
in mediterranean-climate ecosystems. Conserv. Biol. 23, 758–769 (2009).
11. Radelo, V. C. et al. Rapid growth of the us wildland-urban interface raises wildre risk. Proc. Natl. Acad. Sci. 115, 3314–3319
(2018).
12. Martinuzzi, S. etal. e 2010 wildland-urban interface of the conterminous united states. Research Map NRS-8. Newtown Square,
PA: US Department of Agriculture, Forest Service, Northern Research Station. 124 p.[includes pull-out map].8, 1–124 (2015).
13. Ager, A. A. et al. Wildre exposure to the wildland urban interface in the western us. Appl. Geogr. 111, 102059 (2019).
14. Calkin, D. E., Cohen, J. D., Finney, M. A. & ompson, M. P. How risk management can prevent future wildre disasters in the
wildland-urban interface. Proc. Natl. Acad. Sci. 111, 746–751 (2014).
15. Glickman, D. & Babbitt, B. Urban wildland interface communities within the vicinity of federal lands that are at high risk from
wildre. Fed. Reg. 66, 751–777 (2001).
16. Stewart, S. I., Radelo, V. C., Hammer, R. B. & Hawbaker, T. J. Dening the wildland-urban interface. J. Forest. 105, 201–207 (2007).
17. Bar-Massada, A., Stewart, S. I., Hammer, R. B., Mockrin, M. H. & Radelo, V. C. Using structure locations as a basis for mapping
the wildland urban interface. J. Environ. Manage. 128, 540–547 (2013).
18. Platt, R. V. e wildland-urban interface: Evaluating the denition eect. J. Forest. 108, 9–15 (2010).
Figure8. Human-caused wildre ignition points and perimeters between (a) 2000–2009 and (b) 2010–2019 in
California.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
12
Vol:.(1234567890)
Scientic Reports | (2022) 12:5789 | https://doi.org/10.1038/s41598-022-09707-7
www.nature.com/scientificreports/
19. Calkin, D. E., Rieck, J. D., Hyde, K. D. & Kaiden, J. D. Built structure identication in wildland re decision support. Int. J. Wildland
Fire 20, 78–90 (2011).
20. Lowell, K. et al. Assessing the capabilities of geospatial data to map built structures and evaluate their bushre threat. Int. J. Wildland
Fire 18, 1010–1020 (2009).
21. Caggiano, M. D., Tinkham, W. T., Homan, C., Cheng, A. S. & Hawbaker, T. J. High resolution mapping of development in the
wildland-urban interface using object based image extraction. Heliyon 2, e00174 (2016).
22. Johnston, L. M. & Flannigan, M. D. Mapping canadian wildland re interface areas. Int. J. Wildland Fire 27, 1–14 (2017).
23. Alcasena, F. J., Evers, C. R. & Vega-Garcia, C. e wildland-urban interface raster dataset of catalonia. Data Brief 17, 124–128
(2018).
24. Microso. Us building footprints. https:// github. com/ micro so/ USBui lding Footp rints (2018).
25. Picotte, J.J., Long, J., Peterson, B. & Nelson, K.J. Landre 2015 remap–utilization of remotely sensed data to classify existing
vegetation type and structure to support strategic planning and tactical response. Earthzine (2017).
26. Syphard, A. D., Keeley, J. E., Massada, A. B., Brennan, T. J. & Radelo, V. C. Housing arrangement and location determine the
likelihood of housing loss due to wildre. PLoS ONE 7, e33954 (2012).
27. Bar-Massada, A. A comparative analysis of two major approaches for mapping the wildland-urban interface: A case study in
california. Land 10, 679 (2021).
28. LANDFIRE. Landre fuel vegetation cover layer. http:// land re. cr. usgs. gov/ viewer/ (2013, June - last update).
29. Lampin-Maillet, C., Jappiot, M., Long, M., Morge, D. & Ferrier, J.-P. Characterization and mapping of dwelling types for forest re
prevention. Comput. Environ. Urban Syst. 33, 224–232 (2009).
30. USFS. Fire occurrence restat yearly. http:// data. fs. usda. gov/ geoda ta/ edw/ datas ets. php (2020).
31. FRAP, C.F. California re perimeters. https:// frap. re. ca. gov/ frap- proje cts/ re- perim eters/ (2018).
Acknowledgements
TB acknowledges the funding support from the University of California Oce of the President (UCOP) grant
LFR-20-653572 (UC Lab-Fees); the National Science Foundation (NSF) grants NSF-AGS-PDM-2146520
(CAREER), NSF-OISE-2114740 (AccelNet) and NSF-EAR-2052581 (RAPID); the United States Department
of Agriculture (USDA) grant 2021-67022-35908 (NIFA); and a cost reimbursable agreement with the USDA
Forest Service 20-CR-11242306-072.
Author contributions
T.B. designed research; T.B. and P.N. supervised research activities; S.L., V.D. and M.K. performed research; S.L.
wrote the paper; T.B. and P.N. led revisions.
Competing interests
e authors declare no competing interests.
Additional information
Correspondence and requests for materials should be addressed to S.L.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access is article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© e Author(s) 2022
Content courtesy of Springer Nature, terms of use apply. Rights reserved
1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com
Content uploaded by Mukesh Kumar
Author content
All content in this area was uploaded by Mukesh Kumar on May 18, 2022
Content may be subject to copyright.
