remote sensing

Article
Urban Land Surface Temperature Downscaling in Chicago:
Addressing Ethnic Inequality and Gentrification
Jangho Lee 1, * , Max Berkelhammer 1 , Matthew D. Wilson 2 , Natalie Love 3 and Ralph Cintron 4

1 Department of Earth and Environmental Sciences, University of Illinois Chicago, Chicago, IL 60607, USA;
[email protected]
2 Great Cities Institute, University of Illinois Chicago, Chicago, IL 60607, USA; [email protected]
3 Department of Biological Sciences, University of Illinois Chicago, Chicago, IL 60607, USA; [email protected]
4 Department of Latin American and Latino Studies, University of Illinois Chicago, Chicago, IL 60607, USA;
[email protected]
* Correspondence: [email protected]

Abstract: In this study, we developed a XGBoost-based algorithm to downscale 2 km-resolution land

surface temperature (LST) data from the GOES satellite to a finer 70 m resolution, using ancillary
variables including NDVI, NDBI, and DEM. This method demonstrated a superior performance over
the conventional TsHARP technique, achieving a reduced RMSE of 1.90 ◦ C, compared to 2.51 ◦ C
with TsHARP. Our approach utilizes the geostationary GOES satellite data alongside high-resolution
ECOSTRESS data, enabling hourly LST downscaling to 70 m—a significant advancement over
previous methodologies that typically measure LST only once daily. Applying these high-resolution
LST data, we examined the hottest days in Chicago and their correlation with ethnic inequality. Our
analysis indicated that Hispanic/Latino communities endure the highest LSTs, with a maximum LST
that is 1.5 ◦ C higher in blocks predominantly inhabited by Hispanic/Latino residents compared to
those predominantly occupied by White residents. This study highlights the intersection of urban
development, ethnic inequality, and environmental inequities, emphasizing the need for targeted
urban planning to mitigate these disparities. The enhanced spatial and temporal resolution of our
LST data provides deeper insights into diurnal temperature variations, crucial for understanding and
addressing the urban heat distribution and its impact on vulnerable communities.
Citation: Lee, J.; Berkelhammer, M.;
Wilson, M.D.; Love, N.; Cintron, R.
Keywords: land surface temperature; downscaling; GOES-R; ECOSTRESS; climate justice; environ-
Urban Land Surface Temperature mental justice; inequality
Downscaling in Chicago: Addressing
Ethnic Inequality and Gentrification.
Remote Sens. 2024, 16, 1639. https://
doi.org/10.3390/rs16091639 1. Introduction
Academic Editor: Janet Nichol Remotely sensed estimates of land surface temperature (LST) in urban areas are
a critical metric for evaluating urban climate conditions, particularly in examining the
Received: 22 February 2024 surface urban heat island effect and understanding the influence of urban climate on
Revised: 20 April 2024
societal and environmental aspects [1–4]. The high-resolution mapping of LST in urban
Accepted: 3 May 2024
areas has emerged as a vital tool for assessing environmental justice [5,6] and varying heat
Published: 4 May 2024
exposures at the neighborhood scale [7,8], as LST is highly sensitive to urban elements such
as land covers [9,10] and vegetation [11–13]. However, a detailed estimation of LST with a
neighborhood-scale spatial resolution and temporal resolution to capture diurnal cycles
Copyright: © 2024 by the authors.
and extreme events are lacking, underscoring a gap in how high LST in cities varies with
Licensee MDPI, Basel, Switzerland. socioeconomic factors.
This article is an open access article Polar-orbiting satellites such as MODIS [14–17] or the Landsat series [2,12,18] are
distributed under the terms and extensively used for detailed urban LST estimates due to their high spatial resolution (up to
conditions of the Creative Commons 60 m). However, their primary limitation is that they are designed with a fixed return time—
Attribution (CC BY) license (https:// that they revisit the same region on Earth at fixed hour of the day—restricting continuous
creativecommons.org/licenses/by/ LST monitoring or diurnal cycle analysis. ECOSTRESS, mounted on the International Space
4.0/). Station (ISS), has the potential to address this by revisiting locations irregularly, offering

Remote Sens. 2024, 16, 1639. https://doi.org/10.3390/rs16091639 https://www.mdpi.com/journal/remotesensing

Remote Sens. 2024, 16, 1639 2 of 16

insights into the diurnal characteristics of LST while maintaining the high spatial resolution
of 70 m. Although it cannot provide continuous estimates and has long return periods,
it has been widely utilized for LST analysis [19–23]. For example, it has been utilized
for analyzing the diurnal dynamics of surface urban heat island [24], impact of urban
landscapes on LST [25], and various factors affecting LST [26].
On the other hand, geostationary satellites, especially the GOES-R series, stand out
for their ability to continuously monitor a fixed point on Earth, making them ideal for the
sub-hourly LST data capture over North America, despite their lower spatial resolution
(2 km) compared to polar-orbiting satellites. This capability has led to their increasing use
in urban LST monitoring [27–30].
To maximize the benefits of these varied satellite systems, previous studies have
presented methods to improve the spatial resolution of LST estimates through downscaling.
This process often involves using statistical method with ancillary parameters, such as the
Normalized Difference Vegetation Index (NDVI) [19,31,32], available at finer resolutions.
One prominent example using NDVI for downscaling is the TsHARP approach [32] which
utilize NDVI to establish a linear relationship with LST for downscaling. However, in
diverse urban environments, these NDVI-based methods can be subject to large errors due
to the influence of other factors like elevation and urban structures [33–35]. Thus, other
ancillary parameters such as the Normalized Difference Built-up Index (NDBI), and Digital
Elevation Models (DEM) [36,37] are often used to address this limitation.
Additionally, more sophisticated statistical methods incorporating multiple predic-
tor variables and non-linear relationships have been explored. These include machine-
learning and deep-learning techniques such as Random Forest [12,38], Artificial Neural
Networks [39,40], and Extreme Gradient Boosting (XGBoost) [41]. These methods are par-
ticularly effective in complex spatial data analysis due to their ability to capture non-linear
interactions and handle multicollinearity among predictors. However, they also present
challenges, including the potential for overfitting and increased computation demands.
Thus, the detailed pre-processing and robust tuning of the data and models are essential
during the development phase.
One of the primary motivations to generate downscaled LST products is to develop
a better understanding of how exposure to heat—the primary cause of weather-related
mortality [42]—varies with factors such as ethnicity and economic status. Demographic
gradients in cities are steep—often varying significantly block to block—and, thus, data at
this resolution are needed. The unequal exposure to LST among different socioeconomic
groups has recently been the focus of multiple studies [5,6,43–45]. Especially, environmental
inequality from an ethnic perspective has been a concern, especially in large cities in the
US [46,47]. It is critical to document the exposure of minoritized and/or marginalized
groups to higher temperatures because these groups often lack the resources and domestic
infrastructure to mitigate the impact of heat. This disparity makes existing socioeconomic
inequalities more significant and poses risks to public health, particularly during heat
waves and extreme weather events. Additionally, the vulnerability of these communities to
the urban heat island effect is further intensified by their limited access to cooling facilities
and green spaces, a situation exemplified in documented cases from Chicago [48,49].
In this study, our first objective is to develop an XGBoost-based downscaling method
for LST and compare its performance with the conventional TsHARP model, specifically
focusing on the Chicago region in the summertime of 2019–2023. Secondly, our study
extends the outcome of the downscaling to illustrate practical applications of an hourly
70 m LST for a major urban center. We demonstrate how these downscaled LST datasets can
be effectively utilized to classify ethnic inequalities associated with exposure to high LST.

2. Data
2.1. LST Data
In this study, LST data come from two primary sources. The first is the LST estimates
from GOES-16, 17, and 18 satellites, hereafter collectively referred to as GOES [50]. These
 unique orbit of the ISS, not all parts of Chicago are observed in each pass. Therefore, we
have utilized ECOSTRESS data whenever it captures part or all our study region within
its scene. Over the five-year span of our study period, this approach yielded a total of 221
scenes. Previous studies have reported a mean absolute error of the ECOSTRESS LST
Remote Sens. 2024, 16, 1639 product to be below 0.5 K [54]. 3 of 16
One inherent limitation of satellite-based LST estimates is the inability to measure
LST under a cloud cover. To address this issue, we have utilized a cloud mask and quality
control products
estimates, which from
serveboth
as theGOES
basisand ECOSTRESS,
for our using
downscaling, onlya aspatial
have ‘best-quality’ LST
resolution ofprod-
2 km
uct—with a minimal cloud edge and shadow effect. By applying these cloud
and provide hourly estimates. For our analysis, we extract data specific to the urban, masks, we
effectively filter out any LST data obscured by clouds. Consequently, the
suburban, and preurban regions around Chicago during the summer months (June, July, LST data we use
for
andboth GOES
August) and ECOSTRESS
spanning in our
5 years from 2019study canThe
to 2023. be considered as cloud-free
specific study region andLST esti-
its local
mates.
climate The
zonesaverage
[51] areLST for ECOSTRESS
depicted in Figure 1. and GOES
Previous for the
studies entire
have studyerror
reported period can be
margin of
found
GOES in Figure
LST to be 2a and 2b.K [52,53].
below

Figure 1. Location
Figure 1. Location and
and land
land use
use characteristics of the
characteristics of the study
study area.
area. (a)
(a) A
A red
red box
box on
on the
the map
map of
of the
the US
US
highlights the study region. (b) Local climate zones within the study area, providing an overview
highlights the study region. (b) Local climate zones within the study area, providing an overview of
of different land use patterns.
different land use patterns.

2.2. Ancillary Variables

The second source of LST data in our study is from ECOSTRESS LST estimates onboard
In this study,Space
the International we utilize three
Station ancillary
(ISS). variables:
ECOSTRESS has athe Normalized
finer Difference
spatial resolution Vegeta-
of 70 m but
tion Index
has an (NDVI),
irregular the Normalized
revisit period for the Difference Built-up
Chicago area. Index (NDBI),
Additionally, and
due to the
the digitalorbit
unique ele-
of the ISS,
vation mapnot all parts
(DEM). of Chicago
These variables are observed
are selectedin eachthey
since pass.
areTherefore, we have
closely related utilized
to the LST
ECOSTRESS
variation, data whenever
especially in urbanit captures
settingspart or all ourThe
[10,55–57]. study region
NDVI and within
NDBIitsdatasets
scene. Over
are
the five-year
sourced fromspan of our study
Sentinel-2 imageryperiod,
[58],this approach
which yielded
provides a total
a high of 221
spatial scenes. Previous
resolution of 10 m
studies
and haveperiod
a revisit reported a mean
of 10 days. absolute error ofofthe
The calculation ECOSTRESS
NDVI and NDBILST areproduct
performedto be
as below
Equa-
0.5 K [54].
tions (1) and (2):
One inherent limitation of satellite-based LST estimates is the inability to measure
LST under a cloud cover. To address this issue, we have utilized a cloud mask and
quality control products from both GOES and ECOSTRESS, using only a ‘best-quality’ LST
product—with a minimal cloud edge and shadow effect. By applying these cloud masks,
we effectively filter out any LST data obscured by clouds. Consequently, the LST data
we use for both GOES and ECOSTRESS in our study can be considered as cloud-free LST
estimates. The average LST for ECOSTRESS and GOES for the entire study period can be
found in Figure 2a,b.

2.2. Ancillary Variables

In this study, we utilize three ancillary variables: the Normalized Difference Vegetation
Index (NDVI), the Normalized Difference Built-up Index (NDBI), and the digital elevation
map (DEM). These variables are selected since they are closely related to the LST variation,
especially in urban settings [10,55–57]. The NDVI and NDBI datasets are sourced from
Sentinel-2 imagery [58], which provides a high spatial resolution of 10 m and a revisit period
of 10 days. The calculation of NDVI and NDBI are performed as Equations (1) and (2):

NIR − RED
NDVI = (1)
NIR + RED
SWIR − NIR
NDBI = (2)
SWIR + NIR
 NIR − RED
NDVI = (1)
NIR + RED
Remote Sens. 2024, 16, 1639 4 of 16
SWIR − NIR
NDBI = (2)
SWIR + NIR
In Equations
In Equations(1) (1)and
and(2),
(2),NIR,
NIR,RED,
RED,and andSWIR
SWIRcorresponds
correspondstotothe the near-infrared
near-infrared (cen-
(central
wavelength: 0.842 µm), red (0.665 µm), and short-wave infrared (1.610 µm) bands of
tral wavelength: 0.842 μm), red (0.665 μm), and short-wave infrared (1.610 μm) bands of
Sentinel-2,respectively.
Sentinel-2, respectively.NDVI
NDVIexhibits
exhibitsa aclear
clearseasonal
seasonal cycle
cycle asas it related
it is is related to the
to the vegeta-
vegetation,
tion, whereas
whereas NDBINDBIshowsshows
minimalminimal variation
variation duringduring the summer
the summer months. months.
GivenGiven
this, forthis,
NDVIfor
NDVI values,
values, we usewe usevalues.
daily daily values. Since values
Since NDVI NDVI values are available
are available in a 10-dayin a 10-day frequency
frequency due to
duerevisit
the to the frequency
revisit frequency of Sentinel-2,
of Sentinel-2, we temporally
we temporally interpolate
interpolate the NDVI the NDVI
values values to a
to a daily
daily resolution. For the NDBI, we use annual values which are
resolution. For the NDBI, we use annual values which are the median NDBI values for the the median NDBI values
for thesummer
entire entire summer
months of months of that
that year. Overyear. Over the
the 5-year 5-year
study study
period, thisperiod, thisresults
approach approach in a
results in a total of 460 NDVI data maps (92 days × 5 years)
total of 460 NDVI data maps (92 days × 5 years) and 5 NDBI maps (5 years). and 5 NDBI maps (5 years).
The DEM
The DEM dataset
dataset comes
comes from
from the
the 1/3
1/3 arc-second
arc-second (approximately
(approximately 10 10 m)
m) map
mapfromfromthe the
US Geological
US Geological Survey
Survey [59,60].
[59,60]. WeWe have
have chosen
chosen the the 2017
2017 DEM
DEM dataset
datasetfor forour
ourstudy
studyregion,
region,
as itit provides
as provides thethe most
most up-to-date
up-to-date data
data available
available forfor our
our area
area ofof interest.
interest. These
These 2017
2017 DEMDEM
data are
data areemployed
employedas asaastatic
staticvariable
variablemap,
map,consistently
consistently used
used across
across thethe entire
entire study
study pe-
period
riod from 2019 to 2023. From the DEM map, we extract the elevation
from 2019 to 2023. From the DEM map, we extract the elevation (ELEV) for our analysis. (ELEV) for our anal-
ysis. Visualizations
Visualizations of theseofancillary
these ancillary variables,
variables, including including NDVI,and
NDVI, NDBI, NDBI,
ELEV,and
areELEV,
presented are
presented
in Figure 2c–e. in Figure 2c–e.

Figure 2.
Figure 2. Data
Datavisualizations
visualizationsforforthe study,
the with
study, withpermanent
permanent water bodies
water masked
bodies out out
masked in navy. (a)
in navy.
Cloud-free average LST from ECOSTRESS with 70 m resolution for the entire study
(a) Cloud-free average LST from ECOSTRESS with 70 m resolution for the entire study period. period. (b) Same
as (a), but for GOES estimates with 2 km resolution. (c) Average NDVI data with 10 m resolution
(b) Same as (a), but for GOES estimates with 2 km resolution. (c) Average NDVI data with 10 m
for the study period. (d) Similar to (c), but for NDBI. Note that NDBI values range from 0 to 1.
resolution for the study period. (d) Similar to (c), but for NDBI. Note that NDBI values range from 0
However, for better visualization, the color bar is scaled from 0 to 0.1, since NDBI map contains a
to 1. However,
high frequencyfor
of better
valuesvisualization,
near zero. (e)the
10 color bar is scaled
m-resolution ELEV from
map0 for
to 0.1,
thesince
year NDBI map contains
2017, derived from
aDEM
highdata.
frequency of values near zero. (e) 10 m-resolution ELEV map for the year 2017, derived from
DEM data.

2.3. Collocation and Aggregation of Data

Given the irregular observation frequency of ECOSTRESS and the continuous hourly
measurements provided by GOES, we first align the GOES data temporally with ECOSTRESS
scenes. To achieve this, we identify two GOES scenes that temporally bracket each
Remote Sens. 2024, 16, 1639 5 of 16

ECOSTRESS scene. We then calculate a weighted average of these two GOES scenes,
using weights inversely proportional to their temporal distance from the corresponding
ECOSTRESS scene. This process results in GOES data that are temporally aligned to the
ECOSTRESS timestamps. The collocated LST estimates from GOES and ECOSTRESS will
be referred to as LST2km and LST70m , respectively.
As a next step, we aggregate ancillary variables to align with the GOES and ECOSTRESS
data. For each GOES grid point, we define a circle with a radius of 1 km, which is half of
the GOES’s native resolution. Within this circle, we average the 10 m-resolution values of
NDVI, NDBI, and ELEV to derive aggregated values at the 2 km resolution for each of these
variables. This method ensures that the higher-resolution data are appropriately scaled
to match the coarser spatial resolution of the GOES dataset. These averaged variables are
then denoted as NDVI2km , NDBI2km , and ELEV2km , respectively. This aggregation process
ensures that the ancillary data are consistent with the spatial resolution of the LST data
from GOES. The same procedure is applied to ECOSTRESS grid points with a 35 m radius
to generate NDVI70m , NDBI70m , and ELEV70m .

2.4. Socioeconomic Data

In addition to meteorological and land cover data, we utilize socioeconomic data in
this study. The ethnic demographic data for Chicago are sourced from the 2020 Decennial
Census of the U.S. Census Bureau. Organized at the block level, this dataset covers
38,783 city blocks with an average block size of 0.015 km2 and standard deviation of area
of 0.103 km2 . Each block details population counts for total, White, Black, Asian, and
Hispanic/Latino groups. The ethnic composition of Chicago’s population is as follows:
32.5% White, 29.61% Black, 7.14% Asian, and 30.80% Hispanic/Latino. Additionally, we
utilize the economic hardship and education level (population with high school diploma)
which come in at a block group level, which is a group of multiple-block-level data.

3. Downscaling Methods
3.1. TsHARP-Based Method
In our study, we employ a modified version of the TsHARP method [32], a widely used
method to downscale LST [31,61–63], as a baseline model. The TsHARP method establishes
a linear relationship between the NDVI and LST, using it to generate LST data at the finer
resolution of the NDVI map. Mathematically, this can be expressed as Equation (3):

LSTfine (s) = α(s) × (NDVIcoarse (s) − NDVIfine (s)) + LSTcoarse (s) + ε(s) (3)

In Equation (3), LSTfine and LSTcoarse each represents the LST at fine and coarse
resolutions, respectively, corresponding to ECOSTRESS and GOES LST products in this
study. NDVIfine and NDVIcoarse represents the aggregated NDVI value in a fine and coarse
resolution. α is a linear regression coefficient, and ε is an error term. Lastly, s is an indicator
for scene, implying that this relationship is tailored for each individual scene (221 scenes in
this study).
However, the traditional TsHARP method is scene-specific, meaning that the linear
relationship of NDVI and LST is established individually for each collocated time step. The
reason for the scene-specific nature of the TsHARP method lies in the potential variation in
the NDVI–LST relationship across different scenes, likely influenced by factors like the time
of day. To meet our objective of creating LST data even outside these collocated time steps,
we have generalized the approach. To address this, we generate the linear relationship for
each hour of the day rather than for each individual scene, allowing for a more generalized
application. This results in 24 individual TsHARP models representing each hour of the
day. Mathematically, this can be expressed as Equation (4):

LSTfine (h) = α(h) × (NDVIcoarse (h) − NDVIfine (h)) + LSTcoarse (h) + ε(h) (4)
Remote Sens. 2024, 16, 1639 6 of 16

In Equation (4), all variables remain consistent with Equation (3), except for h, an
hour-of-the-day indicator. With the TsHARP method, we can calculate the downscaled LST
at the 70 m-resolution as LSTTsHARP .

3.2. Extreme Gradient Boosting (XGB)

Extreme Gradient Boosting (XGBoost, hereafter XGB) is the primary machine-learning
algorithm utilized in our study [64]. As part of the gradient boosting family, it operates by
constructing multiple decision trees sequentially, where each subsequent tree corrects the
errors of the previous ones. This iterative approach allows XGB to effectively learn from
complex data patterns and improve prediction accuracy.
Here, we utilize nine predictor variables—LST2km , NDVI70m , NDVI2km , NDBI70m ,
NDBI2km , ELEV70m , ELEV2km , hour of the day (HOD), and day of year (DOY)—to predict
LST70m . A notable advantage of our XGB is the handling of multicollinearity, particularly
as NDVI and NDBI are often anti-correlated. Furthermore, by using information from both
the 2 km- and 70 m-resolutions, rather than just their differences as in TsHARP, XGB can
capture potential non-linearities in the influence of NDVI, NDBI, and ELEV on LST.
For the hyperparameter tuning on the XGB model, we employ a Bayesian optimization
algorithm [65] to optimize the selection of hyperparameters, with an objective to minimize
the root mean squared error (RMSE) of the model. This process involved 20 initial points,
followed by 80 iterations, making a total of 100 estimation on the hyperparameter space.
A list of hyperparameters with their descriptions and search ranges on the Bayesian
optimization algorithm is shown in Table 1. Additionally, to avoid the risk of overfitting,
we implement a five-fold cross-validation process. With the XGB method, we can calculate
the downscaled LST at the 70 m-resolution as LSTXGB .

Table 1. Hyperparameter description and optimization details. The “Hyperparameter” and “De-
scription” column show the name and description of hyperparameter utilized. The “Search Range”
column indicates the range within which the Bayesian optimization algorithm sought optimal values.
The “Selected Values” column shows the final hyperparameter values chosen by the optimizer.

Hyperparameter Description Search Range Selected Value

max_depth Maximum depth of a tree [3, 10] 4
learning_rate Boosting learning rate [0.01, 0.3] 0.05
Number of gradient-boosted
n_estimators trees. Equivalent to number of [50, 1200] 700
boosting rounds
Minimum loss reduction
required to make a further
gamma [0, 5] 1
partition on a leaf node of
the tree
Subsample ratio of the training 0.7
subsample instance [0.5, 1]
Subsample ratio of columns
colsample_bytree [0.5, 1] 0.75
when constructing each tree

3.3. Accuracy Verification with Evaluation Set

To assess the performance of the TsHARP and XGB, we divide the data into training
and evaluation sets. We select four specific scenes from the available 221 collocated scenes,
representing the hours of 0, 6, 12, and 18, as our evaluation sets, while all other scenes
are selected as training sets. This selection is aimed at evaluating the models’ ability to
capture LST variations at different times and their diurnal cycles. Furthermore, these scenes
are chosen based on having the second highest number of valid LST estimates in their
respective hours, ensuring minimal cloud cover and maximum coverage by ECOSTRESS.
Due to this selection threshold, the number of pixels in the training set is 92% of the total
data, and the pixels in the evaluation set is 8% of the total data. The schematic of the model
development, validation, and application are depicted in Figure 3.
 LST variations at different times and their diurnal cycles. Furthermore, these scenes are
chosen based on having the second highest number of valid LST estimates in their respec-
tive hours, ensuring minimal cloud cover and maximum coverage by ECOSTRESS. Due
to this selection threshold, the number of pixels in the training set is 92% of the total data,
Remote Sens. 2024, 16, 1639 and the pixels in the evaluation set is 8% of the total data. The schematic of the model 7 of 16
development, validation, and application are depicted in Figure 3.

Figure 3.3.Schematic
Figure Schematicofofdata collection,
data model
collection, development,
model and application
development, of XGBoost
and application LST down-
of XGBoost LST
scaling model.model.
downscaling

4.
4. Downscaling Results
After
After training
trainingthetheTsHARP
TsHARPand andXGB
XGB with thethe
with designated
designatedtraining sets sets
training (217(217
scenes, 92%
scenes,
of
92%data) with awith
of data) five-fold cross-validation
a five-fold process,
cross-validation we compare
process, their performance
we compare against
their performance
LST 70m from
against LST70m ECOSTRESS, in the evaluation
from ECOSTRESS, sets (foursets
in the evaluation scenes,
(four8% of data).
scenes, 8% The initialThe
of data). RMSEini-
between ECOSTRESS and the nearest-neighbor-interpolated GOES
tial RMSE between ECOSTRESS and the nearest-neighbor-interpolated GOES data in the data in the evaluation
set was 3.14 ◦ to 2.46 ◦ Ctofor ◦C
evaluation set C.
was However, the RMSEthe
3.14 °C. However, improved
RMSE improved 2.46TsHARP and to 2.00
°C for TsHARP and to
for
2.00XGB.
°C for XGB.
As
As aa next
next step,
step, we
we train
train TsHARP
TsHARP and and XGB
XGB using
using all
all available
available collocated
collocated data
data for for
broader application. We maintained the same hyperparameters for
broader application. We maintained the same hyperparameters for the XGB as in the train- the XGB as in the
training
ing phasephase to ensure
to ensure there
there is is no overfitting.
no overfitting. Ultimately,
Ultimately, the XGBthe XGBshowed
model model ashowed
superior a
superior performance compared to TsHARP, with a reduction in the RMSE by 0.61 ◦ C and
performance compared to TsHARP, with a ◦reduction in the RMSE by 0.61 °C and in the
in the absolute
mean mean absolute error (MAE)
error (MAE) by 0.40by °C,0.40 C, as depicted
as depicted in Figurein 4.
Figure 4. The consistency
The consistency of these
of these results across both the evaluation set and the entire
results across both the evaluation set and the entire dataset suggests minimal dataset suggests minimal
overfitting
overfitting in the XGB model and its enhanced effectiveness over TsHARP in predicting
in the XGB model and its enhanced effectiveness over TsHARP in predicting LST. Detailed
LST. Detailed statistics including r-squared value (RSQ) and MAE, as well as examples
statistics including r-squared value (RSQ) and MAE, as well as examples from the valida-
from the validation scenes, are presented in Figure 4.
tion scenes, are presented in Figure 4.
Although XGBoost operates as a ‘black box’ model, preventing direct access to the
Although XGBoost operates as a ‘black box’ model, preventing direct access to the
equations of individual trees, we can still discern the significance of predictor variables
equations of individual trees, we can still discern the significance of predictor variables
through the model’s feature importance function. Specifically, we employ the “Gain”
through the model’s feature importance function. Specifically, we employ the “Gain”
function. This function quantifies the importance of each predictor by measuring the
function. This function quantifies the importance of each predictor by measuring the av-
average gain of a feature when it is used in trees. By analyzing the gain values, we can
erage gain of a feature when it is used in trees. By analyzing the gain values, we can de-
determine which features contribute most to the model’s decision-making process, offering
termine which features contribute most to the model’s decision-making process, offering
insights into the underlying factors that drive the predictions, despite the inherent opacity
insights
of into thealgorithm.
the XGBoost underlying factors
Table that drive
2 presents thethe predictions,
feature despite
importance of allthe inherentvariables
predictor opacity
of the XGBoost algorithm. Table 2 presents the feature importance
used in the development of the XGBoost algorithm. As observed in Table 2, HOD emerges of all predictor varia-
bles used in the development of the XGBoost algorithm. As observed
as the most significant variable in our model, followed by NDBI, DOY, NDVI, and ELEV. in Table 2, HOD
emerges as the most significant variable in our model, followed by NDBI, DOY, NDVI,
and ELEV.
Table 2. Relative importance of predictor variables in the XGBoost model, calculated with the average
gain of each feature when used in constructing the model’s trees.

Variable Abbreviation Feature Importance

Hour of the day HOD 0.1016
Day of year DOY 0.0345
Normalized Difference Built-up Index NDBI 0.0532
Normalized Difference Vegetation Index NDVI 0.0142
Elevation ELEV 0.0029
Remote Sens.
Remote Sens. 2024,
2024, 16,
16, 1639
x FOR PEER REVIEW 88of 17
of 16

Figure 4.
Figure 4. Comparative
Comparative analysis
analysis of of LST
LST downscaling
downscaling methods.
methods. (a–d)
(a–d) Downscaling
Downscaling result
result for
for 15 July
15 July
2020, at 00:22, with (a) LST data from ECOSTRESS (LST70m), (b) GOES LST data (LST2km), (c)
2020, at 00:22, with (a) LST data from ECOSTRESS (LST70m ), (b) GOES LST data (LST2km ), (c) TsHARP-
TsHARP-downscaled LST (LSTTsHARP), and (d) XGB-downscaled LST (LSTXGB). Panels (e–h) show
downscaled LST (LSTTsHARP ), and (d) XGB-downscaled LST (LSTXGB ). Panels (e–h) show the same
the same set for 23 June 2022, at 12:13. (i) Depiction of a heatmap comparing LST70m with LSTTsHARP
set for the
across 23 June 2022,
entire at 12:13.dataset,
collocated (i) Depiction of aaheatmap
including linear fitcomparing
equation inLST 70m with
orange, and LST
statistical across
TsHARPmetrics
the entire collocated dataset, including a linear fit equation in orange, and statistical metrics
for the evaluation set in blue and for the final model in red. (j) Same as (i), but for LST70m and LSTXGB for the.
evaluation set in blue
R squared, RMSE, andandMAE forvalues
the final
are model in red. in
also depicted (j)the
Same as (i),
figure forbut
the for LST70m set
Evaluation and(Eval,
LSTXGB .R
using
92% of data
squared, for training
RMSE, and MAE andvalues
testedareon 8%
alsoofdepicted
data) and inthe
thefinal model
figure (Final,
for the using allset
Evaluation available data
(Eval, using
for training
92% of data and testing).and tested on 8% of data) and the final model (Final, using all available data
for training
for training and testing).
Table 2. Relative importance of predictor variables in the XGBoost model, calculated with the aver-
ageEthnic
5. gain ofInequality
each featureof when
LSTused in constructing
Exposure in Chicagothe model’s trees.
5.1. City-Scale EthnicVariable
Inequality Abbreviation Feature Importance
Employing Hour
the XGB downscaling
of the day technique enablesHOD us to refine GOES0.1016 LST data to a
higher resolution (70 m), irrespective
Day of year of the availability of
DOY ECOSTRESS LST (application
0.0345
stage Normalized
of the schematic in Figure 3).
Difference Built-up Index With GOES providing
NDBI hourly LST estimates
0.0532 over
Chicago, we can now obtain LST estimates at a 70 m resolution on an hourly basis. This
Normalized Difference Vegetation Index NDVI 0.0142
enhanced spatiotemporal resolution of the LST data facilitates in-depth assessments at the
Elevation ELEV 0.0029
neighborhood level. Utilizing this capability, we first apply our method to evaluate ethnic
inequality in terms of the high LST in Chicago.
5. Ethnic Inequality of LST Exposure in Chicago
For our study, we selected target days from the GOES dataset over a five-year period
5.1. City-Scale
based Ethnic
on specific Inequality
criteria. Firstly, we focused on days during which more than 90% of
the urban area ofthe
Employing Chicago was under clear
XGB downscaling sky conditions
technique enables us fortothe entire
refine GOES24 hLST
period.
dataThis
to a
threshold was determined to be optimal after evaluating various
higher resolution (70 m), irrespective of the availability of ECOSTRESS LST (application levels, balancing data
availability
stage of the with the quantity
schematic of suitable
in Figure 3). With observation
GOES providing days. hourly
Additionally, we focusover
LST estimates on days
Chi-
with a minimum temperature exceeding 26 ◦ C, representing the 75th percentile of the daily
cago, we can now obtain LST estimates at a 70 m resolution on an hourly basis. This en-
average LST in Chicago’s
hanced spatiotemporal urban regions.
resolution of the LSTApplying these thresholds
data facilitates in-depthresults in 10 days
assessments of
at the
nearly clear skies
neighborhood and
level. high LST,
Utilizing thiswe effectively
capability, weidentify what
first apply ourcan be considered
method to evaluateasethnic
some
of the hottest
inequality days of
in terms in the
Chicago.
high LST Following
in Chicago. this, we apply the XGB downscaling, which
provides us with hourly estimates of LST
For our study, we selected target days from at a 70 mtheresolution
GOES dataset for these
overselected 10 days.
a five-year period
We calculated the average LST
based on specific criteria. Firstly, we XGB focused on days during which more than 90%in
for each block across all 240 h included of our
the
selected 10-day period, capturing the 24
urban area of Chicago was under clear sky conditionsh diurnal LST XGB cycle during Chicago’s
for the entire 24 h period. This hottest
days at thewas
threshold block level. We then
determined to becomputed
optimal afterthe ethnicity-specific
evaluating various diurnal LSTbalancing
levels, cycle for each
data
block by applying a weighted average approach to LST XGB , where the
availability with the quantity of suitable observation days. Additionally, we focus on days weights correspond
to theapercentage
with minimumoftemperature
each ethnic group
exceedingwithin 26the°C,block. Additionally,
representing we conducted
the 75th percentile aofsame
the
analysis
daily average LST in Chicago’s urban regions. Applying these thresholds of
using LST 2km , obtained through the nearest-neighbor interpolation GOESinLST
results 10
data. This method allows us to compare the effects of downscaling on the diurnal LST cycles
days of nearly clear skies and high LST, we effectively identify what can be considered as
some of the hottest days in Chicago. Following this, we apply the XGB downscaling,
Remote Sens. 2024, 16, 1639 9 of 16

Remote Sens. 2024, 16, x FOR PEER REVIEW 10 of 17

specific to different ethnic groups. Chicago’s ethnic distribution and its corresponding
diurnal cycle of LST2km and LSTXGB are depicted in Figure 5a–d.

Figure 5.
Figure 5. (a–d)
(a–d) Distribution
Distribution of
of ethnicity
ethnicity (map)
(map) and
and corresponding
correspondingdiurnal
diurnalcycle
cycleofofLST
LSTobtained
obtained
from LST 2km (blue) and LSTXGB (red) for (a) White, (b) Black, (c) Asian, and (d) Hispanic/Latino pop-
from LST2km (blue) and LSTXGB (red) for (a) White, (b) Black, (c) Asian, and (d) Hispanic/Latino
ulation. (e-h) Heat metric of each ethnicity, calculated as relative to White population. Metric calcu-
population. (e–h) Heat metric of each ethnicity, calculated as relative to White population. Metric
lated with LSTXGB are in gray bar, while values in LST2km are in red line. (e) Mean, (f) maximum,
calculated
(g) minimumwithLST,
LSTXGB areDTR
and (h) in gray bar, while values in LST2km are in red line. (e) Mean, (f) maximum,
of LST.
(g) minimum LST, and (h) DTR of LST.
5.2. Regional Ethnicity Analysis on Humboldt Park
In Figure 5a–d, the diurnal LST cycle derived from the LSTXGB model demonstrates a
more To delvevariation
distinct deeper into ethnic inequalities
compared to that fromwith an .emphasis
LST2km on variability
This higher Hispanic/Latino com-
is consistent
munities, our analysis now focuses on the Humboldt Park region of Chicago.
across all ethnic demographics, showcasing the increased sensitivity of capturing the Humboldt
Park hasevolution
diurnal become aoffocal
LSTpoint
whenfor discussions
using in Chicago,
the downscaling regarding gentrification, which
method.
presents
For a focused examination of ethnicity-specific heat exposure, weand
a complex tapestry of cultural shifts, economic development, displacement
compute four LST
pressures [66–68]. Furthermore, by concentrating our research on a
metrics utilizing LSTXGB data: mean, maximum, minimum, and the diurnal temperaturemore confined area of
range (DTR). Figure 5e–h present bar graphs that illustrate the variance in these heat to
Humboldt Park, rather than the entire expanse of Chicago, we enhance our ability dis-
metrics
cern the nuances and localized impacts of downscaling.
among ethnic groups, with comparisons made relative to the White population. Both the
meanFigure 6a displays
and maximum LSTa(Figure
demographic map of
5e,f) indicate Humboldt
higher Park, highlighting
temperatures ethnic dis-
for the Hispanic/Latino
parities at the block level. This is achieved by calculating the difference
populations compared to Whites. Conversely, the minimum LST differences are negligible between His-
panic/Latino
among and groups,
the ethnic White population percentages
with a variation of less in
thaneach
0.1block. The
◦ C. The resulting
DTR values,
depicted where
in Figure 5h
larger numbers signify a higher prevalence of Hispanic/Latino over White
reveals pronounced disparities, particularly for the Hispanic/Latino community, mostly populations,
reveal a significant west–east gradient in the distribution of these ethnic groups. Moreo-
ver, areas with a Hispanic/Latino majority are characterized by distinct socio-economic
patterns, as depicted in Figure 6b (economic hardship) and Figure 6c (educational
Remote Sens. 2024, 16, 1639 10 of 16

due to the high maximum LST during the day. Overall, we found that Hispanic/Latino
communities are more exposed to higher LST and DTR.
However, while the LSTXGB captures greater diurnal variability in ethnic-specific LST
compared to LST2km , it does not highlight more significant ethnic differences in the mean,
maximum, and DTR of LST. This is attributed to the fact that, although the XGB method
detects a higher variation, the magnitude of these differences between ethnicities are less
pronounced compared to those observed with LST2km .

5.2. Regional Ethnicity Analysis on Humboldt Park

To delve deeper into ethnic inequalities with an emphasis on Hispanic/Latino com-
munities, our analysis now focuses on the Humboldt Park region of Chicago. Humboldt
Park has become a focal point for discussions in Chicago, regarding gentrification, which
presents a complex tapestry of cultural shifts, economic development, and displacement
pressures [66–68]. Furthermore, by concentrating our research on a more confined area
of Humboldt Park, rather than the entire expanse of Chicago, we enhance our ability to
discern the nuances and localized impacts of downscaling.
Figure 6a displays a demographic map of Humboldt Park, highlighting ethnic dis-
parities at the block level. This is achieved by calculating the difference between His-
panic/Latino and White population percentages in each block. The resulting values, where
larger numbers signify a higher prevalence of Hispanic/Latino over White populations,
reveal a significant west–east gradient in the distribution of these ethnic groups. Moreover,
areas with a Hispanic/Latino majority are characterized by distinct socio-economic pat-
terns, as depicted in Figure 6b (economic hardship) and Figure 6c (educational attainment).
Note that the analysis of these socio-economic metrics is conducted at the block group level,
offering a coarser resolution compared to the ethnic data.
Figure 6b presents the hardship index, a measure of economic adversity where higher
scores denote greater hardship. Figure 6c illustrates the educational metric, which quan-
tifies the proportion of the population with a high school diploma. The data illustrated
in these figures indicate that, within Humboldt Park, communities predominantly com-
prising Hispanic/Latino individuals tend to face greater economic disadvantages and
exhibit lower educational attainment levels. Both exhibit strong correlations with the
Hispanic/Latino population percentage, with R² values of 0.56 and 0.49, respectively, un-
derscoring a profound association between ethnic composition and socio-economic factors
in the region.
We match these data with the maximum LSTXGB during the 10 hottest days in Chicago,
as calculated from the previous section (Figure 6d). In the Humboldt Park region, there is
a strong, positive relationship between the block-level Hispanic/Latino population ratio
and maximum LST (Figure 6e). Block-groups with predominantly Hispanic/Latino popu-
lations tend to experience greater maximum temperatures than blocks with predominantly
White populations. Calculating the linear fit (R2 = 0.52) between the ethnic difference and
maximum LST, we find that blocks completely constituted with Hispanic/Latino residents
show a maximum LST that is 1.5 ◦ C higher compared to blocks with solely White residents.
We calculate the same statistics using LST2km . The results, evident in Figure 6f,g,
demonstrate that employing LST2km notably reduces the spatial variability of LST. When
calculating the linear relationship between ethnic differences and LST2km , both the slope
and the R2 value decrease. This suggests that, while using a downscaled LST may not
significantly impact the assessment of ethnic differences at the city level (as shown in
Figure 5), it plays a crucial role in analyzing ethnic inequality at a regional level, such as in
Humboldt Park.
 Remote
Remote Sens.
Sens. 2024,
2024, 16,16, x FOR PEER REVIEW
1639 12 of1117of 16

Figure6.6. (a)
Figure (a) Ethnic
Ethnic map
map ofof Humboldt
HumboldtPark Parkregion,
region,where
where ethnic
ethnicdifference is calculated
difference as His-
is calculated as His-
panic/Latino percentage minus White percentage. Red colors show blocks predominantly inhabited
panic/Latino percentage minus White percentage. Red colors show blocks predominantly inhabited
by Hispanic/Latino residents, while blue shows the blocks that are home to White residents. (b) Map
byofHispanic/Latino
economic hardship residents,
index forwhile blue shows
Humboldt the blocks
Park region. that of
(c) Map areeducational
home to White residents.
attainment (b) Map
in Hum-
ofboldt
economic hardship index for Humboldt Park region. (c) Map of educational
Park region. (d) Maximum LSTXGB map for Humboldt Park region, calculated from 10 clear-attainment in Hum-
boldt Park region.
sky hottest day in(d) Maximum
Chicago. LSTXGB map
(e) Scatterplot andfor Humboldt
linear Park region,between
fit of relationship calculated fromdifference
ethnic 10 clear-sky
and maximum
hottest LSTXGB. (f,g)
day in Chicago. Same as (d,e),
(e) Scatterplot andbut with fit
linear LSTof2km .
relationship between ethnic difference and
maximum LSTXGB . (f,g) Same as (d,e), but with LST2km .
Remote Sens. 2024, 16, 1639 12 of 16

6. Discussions and Conclusions

In this study, we developed an XGBoost-based algorithm to downscale 2 km-resolution
LST data from the GOES satellite to a finer 70 m resolution, utilizing ancillary variables
including NDVI, NDBI, and DEM. This method outperformed the conventional TsHARP
technique, achieving a lower RMSE of 1.90 ◦ C compared to 2.51 ◦ C for TsHARP. The distinc-
tive aspect of our approach is its reliance on data from the geostationary satellite GOES and
the high-resolution estimate from ECOSTRESS with an irregular revisit period, facilitating
hourly LST downscaling, which is vital for acquiring high-resolution LST data consistently
throughout the day. Prior research predominantly focused on LST measurements taken
once daily [8,45,69], an approach that may not adequately capture the complete diurnal
cycle of LST. Additionally, our methodology has the potential for broader application across
North America, leveraging the global observational reach of ECOSTRESS and the North
America-focused capabilities of GOES.
To demonstrate the utility of these data, we assessed a 70 m-resolution LST on
Chicago’s hottest days and investigated its association with ethnic inequality. Our findings
reveal that the Hispanic/Latino communities endure the highest LST and variability, evi-
denced by an elevated maximum LST and DTR. Although LST is not a direct indicator of
human exposure to high temperatures, its relevance as a proxy is supported by the previous
research on thermal comfort [70–72]. Given the previous research indicating the limited
healthcare access for Hispanic/Latino communities in Chicago [73], the observed temper-
ature extremes could disproportionately amplify the risk of temperature-related health
issues within these populations [74,75]. Furthermore, in Chicago, gentrification frequently
impacts Hispanic/Latino communities [76–78], a process associated with reduced access to
green spaces [79] that can increase LST [80–82] and exacerbate health outcomes [83].
To deepen our understanding of the impact of gentrification, we focus on Chicago’s
Humboldt Park, an area notably affected by such changes. Within this region, we observed
that blocks predominantly inhabited by Hispanic/Latino residents, who are economically
and educationally disadvantaged, display a maximum LST that is 1.5 ◦ C higher than blocks
predominantly occupied by White residents. This disparity highlights the intersection
of urban development and ethnic inequality, revealing how gentrification can lead to
environmental inequities. The use of downscaled LST data, derived from our developed
XGBoost downscaling method, enhances the spatial resolution of our analysis, allowing for
a more precise and localized examination of these temperature disparities. Additionally,
unlike data from ECOSTRESS, which is limited by its return period, our downscaled
method enables the estimation of high-resolution LST on an hourly basis, providing a more
continuous temporal resolution. This capability allows for a more detailed understanding
of how LST varies throughout the day, offering insights into the diurnal patterns that affect
urban heat distribution and the resulting impacts on the community’s residents. This
refined approach offers a clearer, more granular understanding of how urban changes
affect different segments of the population, highlighting the critical need for targeted urban
planning and interventions.
Overall, our findings indicate that, in Chicago, the populations with the fewest re-
sources to combat high temperatures are those most exposed to elevated LST. This discovery
is pivotal for policymakers, as it underscores the urgency of targeted urban planning inter-
ventions, such as augmenting green spaces within underrepresented communities. The
significance of such measures is compounded by the existing inequality in the distribution
of green spaces across Chicago [48], further emphasizing the need for equitable access to
these critical urban infrastructures.
While this study provides significant insights, there is uncertainty involved and room
for improvement. Firstly, the accuracy of LST estimates from both GOES and ECOSTRESS
can cause uncertainty [52,55,84–87] as factors such as different observation angles can
influence the LST measurements [88]. Since the purpose of this study is to downscale the
2 km satellite-based LST to 70 m satellite-based LST estimates, inherent uncertainties in
satellite LST measurements are unavoidable. To address this, in situ measurements of LST
Remote Sens. 2024, 16, 1639 13 of 16

from meteorological stations of flux towers could be utilized. However, incorporating these
measurements is outside the scope of the current study, although it represents a promising
direction for future research. Secondly, incorporating metrics beyond LST, such as a 2 m
air temperature and humidity, could enhance the direct analysis of health outcomes and
community impacts. These metrics, which cannot be directly estimated from satellite data,
would need additional research and methodological advancements, such as incorporating
ground-based lidar or a temperature measuring campaign [89]. Lastly, LST measurements
are perforumed under clear sky conditions. Other downscaling methods rather than using
satellite estimates can overcome this limitation [90–92].

Author Contributions: Conceptualization, J.L., M.B., M.D.W. and R.C.; methodology, J.L.; formal
analysis, J.L.; investigation, J.L.; resources, M.B.; data curation, J.L., M.D.W. and N.L.; visualization,
J.L.; writing—original draft preparation, J.L.; writing—review and editing, M.B., M.D.W., N.L.
and R.C.; funding acquisition, M.B. All authors have read and agreed to the published version of
the manuscript.
Funding: This material is based upon work supported by the U.S. Department of Energy, Office
of Science, Office of Biological and Environmental Research’s Urban Integrated Field Laboratories
CROCUS project research activity, under Award Number DE-SC0023226.
Data Availability Statement: All data used in this study are publicly available. GOES-16, 17,
and 18 data are available on Amazon Web Service (AWS) S3 Explorer (GOES-16: https://noaa-
goes16.s3.amazonaws.com/index.html (accessed on 15 December 2023); GOES-17: https://noaa-
goes17.s3.amazonaws.com/index.html (accessed on 15 December 2023); GOES-18: https://noaa-
goes18.s3.amazonaws.com/index.html (accessed on 15 December 2023)). ECOSTRESS data are
available in NASA’s EarthData portal (https://search.earthdata.nasa.gov/search (accessed on 15
December 2023)). Sentinel-2 imagery used to calculate NDVI and NDBI are from Google Earth Engine
(https://developers.google.com/earth-engine/datasets/catalog/sentinel-2 (accessed on 12 January
2024)). DEM maps are available on USGS website (https://data.usgs.gov/datacatalog/data/USGS:
3a81321b-c153-416f-98b7-cc8e5f0e17c3 (accessed on 18 January 2024)). Socioeconomic and ethnic
data for Chicago region can be accessed on the U.S. Census Bureau website (https://data.census.gov/
(accessed on 1 December 2023)).
Conflicts of Interest: The authors declare no conflicts of interest.

References
1. Yang, J.; Yang, Y.; Sun, D.; Jin, C.; Xiao, X. Influence of urban morphological characteristics on thermal environment. Sustain.
Cities Soc. 2021, 72, 103045. [CrossRef]
2. Chen, F.; Yang, S.; Yin, K.; Chan, P. Challenges to quantitative applications of Landsat observations for the urban thermal
environment. J. Environ. Sci. 2017, 59, 80–88. [CrossRef] [PubMed]
3. Sharifi, A.; Pathak, M.; Joshi, C.; He, B.-J. A systematic review of the health co-benefits of urban climate change adaptation.
Sustain. Cities Soc. 2021, 74, 103190. [CrossRef]
4. Moonen, P.; Defraeye, T.; Dorer, V.; Blocken, B.; Carmeliet, J. Urban Physics: Effect of the micro-climate on comfort, health and
energy demand. Front. Archit. Res. 2012, 1, 197–228. [CrossRef]
5. Mashhoodi, B. Environmental justice and surface temperature: Income, ethnic, gender, and age inequalities. Sustain. Cities Soc.
2021, 68, 102810. [CrossRef]
6. Renteria, R.; Grineski, S.; Collins, T.; Flores, A.; Trego, S. Social disparities in neighborhood heat in the Northeast United States.
Environ. Res. 2022, 203, 111805. [CrossRef]
7. Mitchell, B.C.; Chakraborty, J. Exploring the relationship between residential segregation and thermal inequity in 20 US cities.
Local Environ. 2018, 23, 796–813. [CrossRef]
8. Dialesandro, J.; Brazil, N.; Wheeler, S.; Abunnasr, Y. Dimensions of thermal inequity: Neighborhood social demographics and
urban heat in the Southwestern US. Int. J. Environ. Res. Public Health 2021, 18, 941. [CrossRef] [PubMed]
9. Feng, X.; Myint, S.W. Exploring the effect of neighboring land cover pattern on land surface temperature of central building
objects. Build. Environ. 2016, 95, 346–354. [CrossRef]
10. Alexander, C. Normalised difference spectral indices and urban land cover as indicators of land surface temperature (LST). Int. J.
Appl. Earth Obs. Geoinf. 2020, 86, 102013. [CrossRef]
11. Tetali, S.; Baird, N.; Klima, K. A multicity analysis of daytime Surface Urban Heat Islands in India and the US. Sustain. Cities Soc.
2022, 77, 103568. [CrossRef]
Remote Sens. 2024, 16, 1639 14 of 16

12. Guha, S.; Govil, H.; Dey, A.; Gill, N. Analytical study of land surface temperature with NDVI and NDBI using Landsat 8 OLI and
TIRS data in Florence and Naples city, Italy. Eur. J. Remote Sens. 2018, 51, 667–678. [CrossRef]
13. Gage, E.A.; Cooper, D.J. Urban forest structure and land cover composition effects on land surface temperature in a semi-arid
suburban area. Urban For. Urban Green. 2017, 28, 28–35. [CrossRef]
14. Fu, P.; Weng, Q. Variability in annual temperature cycle in the urban areas of the United States as revealed by MODIS imagery.
ISPRS J. Photogramm. Remote Sens. 2018, 146, 65–73. [CrossRef]
15. Al-Hamdan, M.Z.; Quattrochi, D.A.; Bounoua, L.; Lachir, A.; Zhang, P. Using Landsat, MODIS, and a biophysical model to
evaluate LST in urban centers. Remote Sens. 2016, 8, 952. [CrossRef]
16. Mukherjee, S.; Joshi, P.; Garg, R.D. Analysis of urban built-up areas and surface urban heat island using downscaled MODIS
derived land surface temperature data. Geocarto Int. 2017, 32, 900–918. [CrossRef]
17. Keramitsoglou, I.; Kiranoudis, C.T.; Ceriola, G.; Weng, Q.; Rajasekar, U. Identification and analysis of urban surface temperature
patterns in Greater Athens, Greece, using MODIS imagery. Remote Sens. Environ. 2011, 115, 3080–3090. [CrossRef]
18. Elmes, A.; Healy, M.; Geron, N.; Andrews, M.; Rogan, J.; Martin, D.; Sangermano, F.; Williams, C.; Weil, B. Mapping spatiotemporal
variability of the urban heat island across an urban gradient in Worcester, Massachusetts using in-situ Thermochrons and Landsat-
8 Thermal Infrared Sensor (TIRS) data. GISci. Remote Sens. 2020, 57, 845–864. [CrossRef]
19. Hulley, G.; Shivers, S.; Wetherley, E.; Cudd, R. New ECOSTRESS and MODIS land surface temperature data reveal fine-scale heat
vulnerability in cities: A case study for Los Angeles County, California. Remote Sens. 2019, 11, 2136. [CrossRef]
20. Chang, Y.; Xiao, J.; Li, X.; Middel, A.; Zhang, Y.; Gu, Z.; Wu, Y.; He, S. Exploring diurnal thermal variations in urban local climate
zones with ECOSTRESS land surface temperature data. Remote Sens. Environ. 2021, 263, 112544. [CrossRef]
21. Liu, Y.; Xu, X.; Wang, F.; Qiao, Z.; An, H.; Han, D.; Luo, J. Exploring the cooling effect of urban parks based on the ECOSTRESS
land surface temperature. Front. Ecol. Evol. 2022, 10, 1031517. [CrossRef]
22. Yao, X.; Zeng, X.; Zhu, Z.; Lan, Y.; Shen, Y.; Liu, Q.; Yang, F. Exploring the diurnal variations of the driving factors affecting
block-based LST in a “Furnace city” using ECOSTRESS thermal imaging. Sustain. Cities Soc. 2023, 98, 104841. [CrossRef]
23. Chang, Y.; Xiao, J.; Li, X.; Zhou, D.; Wu, Y. Combining GOES-R and ECOSTRESS land surface temperature data to investigate
diurnal variations of surface urban heat island. Sci. Total Environ. 2022, 823, 153652. [CrossRef] [PubMed]
24. Chang, Y.; Xiao, J.; Li, X.; Weng, Q. Monitoring diurnal dynamics of surface urban heat island for urban agglomerations using
ECOSTRESS land surface temperature observations. Sustain. Cities Soc. 2023, 98, 104833. [CrossRef]
25. Han, D.; An, H.; Cai, H.; Wang, F.; Xu, X.; Qiao, Z.; Jia, K.; Sun, Z.; An, Y. How do 2D/3D urban landscapes impact diurnal land
surface temperature: Insights from block scale and machine learning algorithms. Sustain. Cities Soc. 2023, 99, 104933. [CrossRef]
26. Lin, Z.; Xu, H.; Yao, X.; Yang, C.; Ye, D. How does urban thermal environmental factors impact diurnal cycle of land surface
temperature? A multi-dimensional and multi-granularity perspective. Sustain. Cities Soc. 2024, 101, 105190. [CrossRef]
27. Chang, Y.; Xiao, J.; Li, X.; Frolking, S.; Zhou, D.; Schneider, A.; Weng, Q.; Yu, P.; Wang, X.; Li, X. Exploring diurnal cycles of surface
urban heat island intensity in Boston with land surface temperature data derived from GOES-R geostationary satellites. Sci. Total
Environ. 2021, 763, 144224. [CrossRef] [PubMed]
28. Hrisko, J.; Ramamurthy, P.; Yu, Y.; Yu, P.; Melecio-Vázquez, D. Urban air temperature model using GOES-16 LST and a diurnal
regressive neural network algorithm. Remote Sens. Environ. 2020, 237, 111495. [CrossRef]
29. Hrisko, J.; Ramamurthy, P.; Melecio-Vázquez, D.; Gonzalez, J.E. Spatiotemporal Variability of Heat Storage in Major US Cities—A
Satellite-Based Analysis. Remote Sens. 2020, 13, 59. [CrossRef]
30. Beale, C.; Norouzi, H.; Sharifnezhadazizi, Z.; Bah, A.R.; Yu, P.; Yu, Y.; Blake, R.; Vaculik, A.; Gonzalez-Cruz, J. Comparison of
diurnal variation of land surface temperature from GOES-16 ABI and MODIS instruments. IEEE Geosci. Remote Sens. Lett. 2019,
17, 572–576. [CrossRef]
31. Ha, W.; Gowda, P.H.; Howell, T.A. Downscaling of land surface temperature maps in the Texas high plains with the TsHARP
method. GISci. Remote Sens. 2011, 48, 583–599. [CrossRef]
32. Agam, N.; Kustas, W.P.; Anderson, M.C.; Li, F.; Colaizzi, P.D. Utility of thermal sharpening over Texas high plains irrigated
agricultural fields. J. Geophys. Res. Atmos. 2007, 112, D19110. [CrossRef]
33. Chen, X.; Li, W.; Chen, J.; Rao, Y.; Yamaguchi, Y. A combination of TsHARP and thin plate spline interpolation for spatial
sharpening of thermal imagery. Remote Sens. 2014, 6, 2845–2863. [CrossRef]
34. Inamdar, A.K.; French, A. Disaggregation of GOES land surface temperatures using surface emissivity. Geophys. Res. Lett. 2009,
36, L02408. [CrossRef]
35. Hais, M.; Kučera, T. The influence of topography on the forest surface temperature retrieved from Landsat TM, ETM+ and ASTER
thermal channels. ISPRS J. Photogramm. Remote Sens. 2009, 64, 585–591. [CrossRef]
36. Maeda, E.E. Downscaling MODIS LST in the East African mountains using elevation gradient and land-cover information. Int. J.
Remote Sens. 2014, 35, 3094–3108. [CrossRef]
37. Zhang, X.; Zhao, H.; Yang, J. Spatial downscaling of land surface temperature in combination with TVDI and elevation. Int. J.
Remote Sens. 2019, 40, 1875–1886. [CrossRef]
38. Hutengs, C.; Vohland, M. Downscaling land surface temperatures at regional scales with random forest regression. Remote Sens.
Environ. 2016, 178, 127–141. [CrossRef]
39. Pu, R. Assessing scaling effect in downscaling land surface temperature in a heterogenous urban environment. Int. J. Appl. Earth
Obs. Geoinf. 2021, 96, 102256. [CrossRef]
Remote Sens. 2024, 16, 1639 15 of 16

40. Li, W.; Ni, L.; Li, Z.-l.; Duan, S.-B.; Wu, H. Evaluation of machine learning algorithms in spatial downscaling of MODIS land
surface temperature. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2019, 12, 2299–2307. [CrossRef]
41. Tu, H.; Cai, H.; Yin, J.; Zhang, X.; Zhang, X. Land surface temperature downscaling in the karst mountain urban area considering
the topographic characteristics. J. Appl. Remote Sens. 2022, 16, 034515. [CrossRef]
42. Anderson, B.G.; Bell, M.L. Weather-related mortality: How heat, cold, and heat waves affect mortality in the United States.
Epidemiology 2009, 20, 205. [CrossRef] [PubMed]
43. Mashhoodi, B. Feminization of surface temperature: Environmental justice and gender inequality among socioeconomic groups.
Urban Clim. 2021, 40, 101004. [CrossRef]
44. Zhu, Y.; Myint, S.W.; Schaffer-Smith, D.; Muenich, R.L.; Tong, D.; Li, Y. Formulating operational mitigation options and examining
intra-urban social inequality using evidence-based urban warming effects. Front. Environ. Sci. 2022, 9, 795474. [CrossRef]
45. Mitchell, B.C.; Chakraborty, J. Landscapes of thermal inequity: Disproportionate exposure to urban heat in the three largest US
cities. Environ. Res. Lett. 2015, 10, 115005. [CrossRef]
46. Benz, S.A.; Burney, J.A. Widespread race and class disparities in surface urban heat extremes across the United States. Earth's
Future 2021, 9, e2021EF002016. [CrossRef]
47. Hsu, A.; Sheriff, G.; Chakraborty, T.; Manya, D. Disproportionate exposure to urban heat island intensity across major US cities.
Nat. Commun. 2021, 12, 2721. [CrossRef]
48. Liu, D.; Kwan, M.-P.; Kan, Z. Analysis of urban green space accessibility and distribution inequity in the City of Chicago. Urban
For. Urban Green. 2021, 59, 127029. [CrossRef]
49. Williams, T.G.; Logan, T.M.; Zuo, C.T.; Liberman, K.D.; Guikema, S.D. Parks and safety: A comparative study of green space
access and inequity in five US cities. Landsc. Urban Plan. 2020, 201, 103841. [CrossRef]
50. Schmit, T.J.; Gunshor, M.M.; Menzel, W.P.; Gurka, J.J.; Li, J.; Bachmeier, A.S. Introducing the next-generation Advanced Baseline
Imager on GOES-R. Bull. Am. Meteorol. Soc. 2005, 86, 1079–1096. [CrossRef]
51. Ching, J.; Mills, G.; Bechtel, B.; See, L.; Feddema, J.; Wang, X.; Ren, C.; Brousse, O.; Martilli, A.; Neophytou, M. WUDAPT: An
urban weather, climate, and environmental modeling infrastructure for the anthropocene. Bull. Am. Meteorol. Soc. 2018, 99,
1907–1924. [CrossRef]
52. Yu, Y.; Tarpley, D.; Privette, J.L.; Flynn, L.E.; Xu, H.; Chen, M.; Vinnikov, K.Y.; Sun, D.; Tian, Y. Validation of GOES-R satellite land
surface temperature algorithm using SURFRAD ground measurements and statistical estimates of error properties. IEEE Trans.
Geosci. Remote Sens. 2011, 50, 704–713. [CrossRef]
53. Yu, Y.; Yu, P. Land surface temperature product from the GOES-R series. In The GOES-R Series; Elsevier: Amsterdam, The
Netherlands, 2020; pp. 133–144.
54. Hulley, G.C.; Göttsche, F.M.; Rivera, G.; Hook, S.J.; Freepartner, R.J.; Martin, M.A.; Cawse-Nicholson, K.; Johnson, W.R. Validation
and quality assessment of the ECOSTRESS level-2 land surface temperature and emissivity product. IEEE Trans. Geosci. Remote
Sens. 2021, 60, 5000523. [CrossRef]
55. Guha, S.; Govil, H.; Gill, N.; Dey, A. A long-term seasonal analysis on the relationship between LST and NDBI using Landsat
data. Quat. Int. 2021, 575, 249–258. [CrossRef]
56. Guo, A.; Yang, J.; Sun, W.; Xiao, X.; Cecilia, J.X.; Jin, C.; Li, X. Impact of urban morphology and landscape characteristics on
spatiotemporal heterogeneity of land surface temperature. Sustain. Cities Soc. 2020, 63, 102443. [CrossRef]
57. Guha, S.; Govil, H. Seasonal variability of LST-NDVI correlation on different land use/land cover using Landsat satellite sensor:
A case study of Raipur City, India. Environ. Dev. Sustain. 2022, 24, 8823–8839. [CrossRef]
58. Gascon, F.; Bouzinac, C.; Thépaut, O.; Jung, M.; Francesconi, B.; Louis, J.; Lonjou, V.; Lafrance, B.; Massera, S.; Gaudel-Vacaresse,
A. Copernicus Sentinel-2A calibration and products validation status. Remote Sens. 2017, 9, 584. [CrossRef]
59. Maune, D.F.; Kopp, S.; Zerdas, C. Digital elevation model technologies and applications. In The DEM Users Manual; American
Society for Photogrammetry and Remote Sensing: Bethesda, MD, USA, 2007.
60. Gesch, D.B.; Oimoen, M.J.; Evans, G.A. Accuracy Assessment of the US Geological Survey National Elevation Dataset, and Comparison
with other Large-Area Elevation Datasets: SRTM and ASTER; US Department of the Interior, US Geological Survey: Sioux Falls, SD,
USA, 2014; Volume 1008.
61. Sattari, F.; Hashim, M.; Sookhak, M.; Banihashemi, S.; Pour, A.B. Assessment of the TsHARP method for spatial downscaling of
land surface temperature over urban regions. Urban Clim. 2022, 45, 101265. [CrossRef]
62. Mukherjee, S.; Joshi, P.; Garg, R.D. A comparison of different regression models for downscaling Landsat and MODIS land
surface temperature images over heterogeneous landscape. Adv. Space Res. 2014, 54, 655–669. [CrossRef]
63. Sattari, F.; Hashim, M.; Pour, A.B. Thermal sharpening of land surface temperature maps based on the impervious surface index
with the TsHARP method to ASTER satellite data: A case study from the metropolitan Kuala Lumpur, Malaysia. Measurement
2018, 125, 262–278. [CrossRef]
64. Chen, T.; Guestrin, C. Xgboost: A scalable tree boosting system. In Proceedings of the 22nd ACM Sigkdd International Conference
on Knowledge Discovery and Data Mining, San Francisco, CA, USA, 13–17 August 2016; pp. 785–794.
65. Snoek, J.; Larochelle, H.; Adams, R.P. Practical bayesian optimization of machine learning algorithms. Adv. Neural Inf. Process.
Syst. 2012, 25, 2951.
66. Mumm, J. Gentrification in color and time: White and Puerto Rican racial histories at work in Humboldt Park. Cent. J. 2016, 28,
88–125.
Remote Sens. 2024, 16, 1639 16 of 16

67. Wilson, D.; Grammenos, D. Gentrification, discourse, and the body: Chicago’s Humboldt Park. Environ. Plan. D Soc. Space 2005,
23, 295–312. [CrossRef]
68. Rinaldo, R. Space of resistance: The Puerto Rican cultural center and Humboldt Park. Cult. Crit. 2002, 50, 135–174. [CrossRef]
69. Chakraborty, T.; Hsu, A.; Manya, D.; Sheriff, G. Disproportionately higher exposure to urban heat in lower-income neighborhoods:
A multi-city perspective. Environ. Res. Lett. 2019, 14, 105003. [CrossRef]
70. Goldblatt, R.; Addas, A.; Crull, D.; Maghrabi, A.; Levin, G.G.; Rubinyi, S. Remotely sensed derived land surface temperature
(LST) as a proxy for air temperature and thermal comfort at a small geographical scale. Land 2021, 10, 410. [CrossRef]
71. Patel, S.; Indraganti, M.; Jawarneh, R.N. A comprehensive systematic review: Impact of Land Use/Land Cover (LULC) on Land
Surface Temperatures (LST) and outdoor thermal comfort. Build. Environ. 2023, 249, 111130. [CrossRef]
72. Imran, H.; Hossain, A.; Islam, A.S.; Rahman, A.; Bhuiyan, M.A.E.; Paul, S.; Alam, A. Impact of land cover changes on land surface
temperature and human thermal comfort in Dhaka city of Bangladesh. Earth Syst. Environ. 2021, 5, 667–693. [CrossRef]
73. Lee, H.; Caldwell, J.T.; Maene, C.; Cagney, K.A.; Saunders, M.R. Racial/ethnic inequities in access to high-quality dialysis
treatment in Chicago: Does neighborhood racial/ethnic composition matter? J. Racial Ethn. Health Disparities 2020, 7, 854–864.
[CrossRef]
74. Lee, J.; Dessler, A.E. Future Temperature-Related Deaths in the US: The Impact of Climate Change, Demographics, and Adaptation.
GeoHealth 2023, 7, e2023GH000799. [CrossRef]
75. Davis, R.E.; Hondula, D.M.; Sharif, H. Examining the diurnal temperature range enigma: Why is human health related to the
daily change in temperature? Int. J. Biometeorol. 2020, 64, 397–407. [CrossRef] [PubMed]
76. Betancur, J.J. The politics of gentrification: The case of West Town in Chicago. Urban Aff. Rev. 2002, 37, 780–814. [CrossRef]
77. Mumm, J. The racial fix: White currency in the gentrification of black and Latino Chicago. Focaal 2017, 2017, 102–118. [CrossRef]
78. Orta, D. “Mexicans Built This Neighborhood!” Gentrification, Organizations, and the Role of Place-Based Identity in Latinx
Chicago. Soc. Sci. 2021, 10, 304. [CrossRef]
79. Francis, J.; Disney, M.; Law, S. Monitoring canopy quality and improving equitable outcomes of urban tree planting using LiDAR
and machine learning. Urban For. Urban Green. 2023, 89, 128115. [CrossRef]
80. Schwaab, J.; Meier, R.; Mussetti, G.; Seneviratne, S.; Bürgi, C.; Davin, E.L. The role of urban trees in reducing land surface
temperatures in European cities. Nat. Commun. 2021, 12, 6763. [CrossRef] [PubMed]
81. Wang, J.; Zhou, W.; Jiao, M. Location matters: Planting urban trees in the right places improves cooling. Front. Ecol. Environ. 2022,
20, 147–151. [CrossRef]
82. Zhou, W.; Wang, J.; Cadenasso, M.L. Effects of the spatial configuration of trees on urban heat mitigation: A comparative study.
Remote Sens. Environ. 2017, 195, 1–12. [CrossRef]
83. Cole, H.V.; Triguero-Mas, M.; Connolly, J.J.; Anguelovski, I. Determining the health benefits of green space: Does gentrification
matter? Health Place 2019, 57, 1–11. [CrossRef]
84. Li, K.; Guan, K.; Jiang, C.; Wang, S.; Peng, B.; Cai, Y. Evaluation of four new land surface temperature (LST) products in the
US corn belt: ECOSTRESS, GOES-R, landsat, and sentinel-3. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2021, 14, 9931–9945.
[CrossRef]
85. Sun, D.; Yu, Y.; Fang, L.; Liu, Y. Toward an operational land surface temperature algorithm for GOES. J. Appl. Meteorol. Climatol.
2013, 52, 1974–1986. [CrossRef]
86. Xu, H.; Yu, Y.; Tarpley, D.; Göttsche, F.-M.; Olesen, F.-S. Evaluation of GOES-R land surface temperature algorithm using SEVIRI
satellite retrievals with in situ measurements. IEEE Trans. Geosci. Remote Sens. 2013, 52, 3812–3822. [CrossRef]
87. Meng, X.; Cheng, J.; Yao, B.; Guo, Y. Validation of the ECOSTRESS land surface temperature product using ground measurements.
IEEE Geosci. Remote Sens. Lett. 2021, 19, 3005305. [CrossRef]
88. Bian, Z.; Roujean, J.-L.; Lagouarde, J.-P.; Cao, B.; Li, H.; Du, Y.; Liu, Q.; Xiao, Q.; Liu, Q. A semi-empirical approach for modeling
the vegetation thermal infrared directional anisotropy of canopies based on using vegetation indices. ISPRS J. Photogramm. Remote
Sens. 2020, 160, 136–148. [CrossRef]
89. Kraemer, R.; Remmler, P.; Bumberger, J.; Kabisch, N. Running a dense air temperature measurement field campaign at the urban
neighbourhood level: Protocol and lessons learned. MethodsX 2022, 9, 101719. [CrossRef] [PubMed]
90. Zhang, X.; Zhou, J.; Liang, S.; Wang, D. A practical reanalysis data and thermal infrared remote sensing data merging (RTM)
method for reconstruction of a 1-km all-weather land surface temperature. Remote Sens. Environ. 2021, 260, 112437. [CrossRef]
91. Long, D.; Yan, L.; Bai, L.; Zhang, C.; Li, X.; Lei, H.; Yang, H.; Tian, F.; Zeng, C.; Meng, X. Generation of MODIS-like land surface
temperatures under all-weather conditions based on a data fusion approach. Remote Sens. Environ. 2020, 246, 111863. [CrossRef]
92. Duan, S.-B.; Lian, Y.; Zhao, E.; Chen, H.; Han, W.; Wu, Z. A Novel Approach to All-Weather LST Estimation using XGBoost Model
and Multi-Source Data. IEEE Trans. Geosci. Remote Sens. 2023, 61, 5004614. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.