JUNE 2019 SHARIFNEZHADAZIZI ET AL.

1279

A Global Analysis of Land Surface Temperature Diurnal Cycle Using

MODIS Observations

ZAHRA SHARIFNEZHADAZIZI
City College of New York, City University of New York, New York, New York

HAMID NOROUZI
New York City College of Technology, City University of New York, Brooklyn, and Earth and
Environmental Sciences, Graduate Center, City University of New York, New York, New York

SATYA PRAKASH
Divecha Centre for Climate Change, Indian Institute of Science, Bengaluru, India

CHRISTOPHER BEALE
New York City College of Technology, City University of New York, Brooklyn, New York

REZA KHANBILVARDI
City College of New York, City University of New York, New York, New York

(Manuscript received 27 September 2018, in final form 28 March 2019)

ABSTRACT

Diurnal variations of land surface temperature (LST) play a vital role in a wide range of applications such as
climate change assessment, land–atmosphere interactions, and heat-related health issues in urban regions.
This study uses 15 years (2003–17) of daily observations of LST Collection 6 from the Moderate Resolution
Imaging Spectroradiometer (MODIS) instruments on board the Aqua and the Terra satellites. A spline in-
terpolation method is used to estimate half-hourly global LST from the MODIS measurements. A pre-
liminary assessment of interpolated LST with hourly ground-based observations over selected stations of
North America shows bias and an error of less than 1 K. Results suggest that the present interpolation method
is capable of capturing the diurnal variations of LST reasonably well for different land-cover types. The
diurnal cycle of LST and time of occurrence of maximum temperature are computed from the spatially and
temporally consistent interpolated diurnal LST data at a global scale. Regions with higher variability in the
timing of maximum LST hours and diurnal amplitude are identified in this study. The global desert regions
show generally small variability of the monthly mean diurnal LST range, whereas larger areas of the global
land exhibit rather higher variability in the diurnal LST range during the study period. Moreover, the changes
in diurnal temperature range for the study period are examined for distinct land-cover types. Analysis of the
15-yr time series of the diurnal LST record shows an overall decrease of 0.5 K in amplitude over the Northern
Hemisphere. However, the diurnal LST range shows variant changes in the Southern Hemisphere.

1. Introduction island monitoring, evaluation of land surface models, and

assessment of climate change (Braganza et al. 2004;
Reliable estimates of land surface temperature (LST)
Tomlinson et al. 2011; Li et al. 2013; Zhan et al. 2013;
and its diurnal variations are vital for the adequate un-
Thorne et al. 2016). LST has been considered as one of the
derstanding of surface energy budget and land–atmosphere
high-priority parameters of the International Geosphere–
interactions, for the detection of forest fires, urban heat
Biosphere Programme. The solar insolation, atmospheric
state, and land surface characteristics (e.g., vegetation
Corresponding author: Hamid Norouzi, hnorouzi@citytech.cuny.edu cover, soil moisture, land-cover type) regulates the diurnal

DOI: 10.1175/JAMC-D-18-0256.1
Ó 2019 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright
Policy (www.ametsoc.org/PUBSReuseLicenses).
1280 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58

cycle of LST (Ignatov and Gutman 1999; Duan et al. model to construct the diurnal cycle of MODIS-derived
2014a). The diurnal cycle of skin temperature over the LST over the Mediterranean region. Zhou et al. (2013)
ocean is essentially different from that over the land. utilized a diurnal temperature cycle genetic algorithm
Moreover, the diurnal cycle patterns of LST vary locally with multisource satellite data to generate hourly LST at
with season and land-cover type, and are different in 1-km spatial resolution for the study of diurnal variations
magnitudes from those of near-surface air temperature of urban heat islands in China. Weng and Fu (2014)
and upper-layer soil temperature (Norouzi et al. 2012; developed a support vector machine method for down-
Prakash et al. 2017, 2019; Shati et al. 2018). scaling coarse-resolution infrared LST into finer spatio-
Ground-based observations of LST are generally sparse temporal resolution for monitoring of surface urban heat
at global scale and hence appear to be inadequate to study islands in the Los Angeles region.
its variability at larger spatial scale. However, satellite Ruzmaikin et al. (2017) examined the changes in spa-
remote sensing provides an effective way to measure LST tial patterns of diurnal temperature range (DTR) from
uniformly at global scale (Tomlinson et al. 2011; Li et al. the skin temperature measurements of the Atmospheric
2013). LST is measured from broadly two types of Earth- Infrared Sounder (AIRS) sensor on board the Aqua
observation satellite sensors: (i) thermal infrared sensors satellite using empirical mode decomposition. They no-
and (ii) microwave sensors. The LST derived from the ticed an increase of about 0.2 K in the global DTR be-
thermal infrared measurements has higher spatial reso- tween 2002 and 2015 that is possibly due to decrease in
lution and retrieval accuracy. However, infrared-based nighttime low cloud fraction. However, DTR was found
measurements are restricted to clear-sky conditions only to be nonmonotonic over the oceanic regions and in the
and obscured by presence of clouds and atmospheric extremely hot areas of Earth. The spatial resolution of the
disturbances. Although passive microwave measurements AIRS is rather coarser (;100 km) than the MODIS
essentially provide LST estimates for all weather condi- sensor. Additionally, MODIS sensors are mounted on
tions, they suffer from rather coarser spatial and temporal both Aqua and Terra satellites and provide generally four
resolutions. Moreover, uncertainty in the passive micro- observations daily at a given location, which may help to
wave retrieved LST during precipitation has shown to be accurately estimate LST diurnal variations. Recently,
rather larger than infrared-based estimates. The syner- Prakash et al. (2019) comprehensively assessed the dif-
gistic use of infrared and microwave measurements has ferences between MODIS- and AIRS-derived clear-sky
been recommended for the optimal estimation of LST for LST estimates over the global land areas. They also
all weather conditions (Li et al. 2013; Duan et al. 2017). showed that LST exhibits stronger diurnal variability
The thermal infrared-based LST estimates from the than the surface air temperature.
polar-orbiting satellites have received more attention in The objective of this study is twofold: 1) to generate
last two decades for several global and local applications. spatially consistent global maps of subdaily LST from the
For instance, LST estimates from the Moderate Resolu- MODIS measurements under clear-sky condition and
tion Imaging Spectroradiometer (MODIS) sensors have 2) to assess the variability of DTR for different vegetation
been used for urban heat island detection, assessment of types at global scale for the 15-yr period of 2003–17.
land surface changes, evaluation of land surface model Section 2 describes the datasets used in this study, and the
outputs, land and atmospheric parameters retrieval, and method for the computation of diurnal cycle of LST is
several other applications (Wan 2014; Norouzi et al. 2015;
presented in section 3. Results are presented and discussed
Prakash et al. 2016, 2018; Didari et al. 2017; Mildrexler
in section 4, and conclusions are outlined in section 5.
et al. 2018). However, MODIS sensors are unable to
provide complete global coverage of diurnal LST esti-
mates, and a spatiotemporal interpolation is required to 2. Data used
study the global LST variability. Several interpolation
a. Land surface temperature data
methods have been reported in literature for satellite
remote sensing data including infrared-based LST (Aires The MODIS sensors on board the Aqua and Terra
et al. 2004; Duan et al. 2014a; Norouzi et al. 2012; Zhou satellites measure high-spatial-resolution atmospheric,
et al. 2013; Weng and Fu 2014; Ruzmaikin et al. 2017). oceanic, and land surface parameters at 36 visible and
Aires et al. (2004) analyzed and constructed the diurnal infrared channels with a wide spectral range of 0.4–
cycle of infrared-based LST using a principal component 14.4 mm (Justice et al. 1998). These two sun-synchronous
analysis at a quasi-global scale using the International satellites are polar orbiting and provide global coverage
Satellite Cloud Climatology Project (ISCCP) skin tem- in nearly two days. The Terra satellite crosses the
perature product (Rossow and Schiffer 1999). Duan et al. equator from north to south at about 1030 local time,
(2014a) used a semiempirical diurnal temperature cycle whereas the Aqua satellite crosses the equator from
JUNE 2019 SHARIFNEZHADAZIZI ET AL. 1281

south to north at about 1330 local time. These satellites TABLE 1. Land-cover types used in this study and their percentage
take measurements twice a day (e.g., ascending and de- coverage over the global land.
scending orbits) at a specific location. The MODIS Percentage
products from the Terra and the Aqua satellites are Code Vegetation type coverage
available since March 2000 and July 2002, respec- LC01 Tropical/subtropical evergreen 8.00
tively. The cloud-free daily global LST products (e.g., broadleaf forest
MOD11C1 from the Terra and MYD11C1 from the LC02 Deciduous forest 11.12
Aqua satellites) Collection 6 (Wan 2014) available at a LC03 Evergreen broadleaf and 9.48
needleleaf forest
0.058 (about 5 km at the equator) climate modeling grid
LC04 Deciduous woodland 5.10
were obtained from the Land Processes Distributed LC05 Sclerophyllous woodland and 3.18
Active Archive Center (LP DAAC). Collection 6 LST forest
products incorporate improvements in retrieval algo- LC06 Wooded and nonwooded 23.31
rithms and proven to be more accurate than its pre- grassland
LC07 Tundra and mossy bog 3.50
decessor Collection 5 LST products (Wan 2014; Duan
LC08 Boreal and xeromorphic 10.30
et al. 2018). The differences between Collection 6 and shrubland
Collection 5 products were remarkable over the arid re- LC09 Nonvegetated desert 11.06
gions (Prakash et al. 2018). Moreover, the accuracy of LC10 Ice 9.83
Collection 6 LST products was found to be approximately
2 times better than those of Collection 5 products over bare
soil areas, when validated using the radiance-based method reprojected to an equal-area grid (equivalent to 0.258 at
(Duan et al. 2018). The combination of ascending and equator) and land-only pixels were extracted for compu-
descending overpasses of both MODIS sensors provides tational efficiency. LST products from the MODIS are
generally four measurements of LST at a specific grid each often influenced by clouds and other atmospheric distur-
day at about 0130, 1030, 1330, and 2230 local times. bances, which resulted in remarkable data loss. Several
The global LST products for a 15-yr period (e.g., January techniques have been developed so far to compute
2003–December 2017) were considered in this study. equivalent LST values at the missing grids of the MODIS
For the preliminary evaluation of diurnal cycle of LST swath. But, most of these techniques are suitable for rather
derived from the MODIS products, ground-based ob- smaller regions and fail at the global scale (Kilibarda et al.
servations of LST from the U.S. Climate Reference 2014; Yu et al. 2015). In addition, geostatistical interpola-
Network (USCRN) were used. The USCRN is an au- tion methods basically depend on the values of neighbor-
tomated network of more than 140 stations designed to ing grids; hence, they are not effective for a larger area
provide high-quality continuous observations of climate having missing values. As a longer time series (e.g.,
variables (Bell et al. 2013; Diamond et al. 2013). Hourly 15 years) of LST is being considered in this study, a statis-
LST observations from 18 selected stations for the year tical method based on temporal variations was utilized for
2017 were used in this study. filling the data gap at different time of the day. The goal of
present study is to analyze the global diurnal cycle of LST
b. Land-cover type data
when observations from clear-sky conditions are utilized.
A global static classification of 10 vegetation types Based on previous studies (Norouzi et al. 2015), it is
(Prigent et al. 1998; Norouzi et al. 2015) derived from 32 reasonable to assume that general characteristics of LST
different land-cover types (Matthews 1983) has been used diurnal variations such as amplitude and timing of
in this study. The name of each vegetation class and their maximum temperature do not change significantly in a
percentage coverage over the global land area are tabu- month. Figure 1 depicts the schematic estimation pro-
lated in Table 1. The wooded and nonwooded grassland cess of missing observation data. If at a given grid, there
covers about 23% of global land, whereas nonvegetated are one, two or three unavailable LST values, they are
desert covers about 11% of the global land area. It is to be computed by differences between the available values of
noted that the diurnal cycle patterns of LST significantly that date and the monthly average of the same time of
influenced by the land-cover types (Scheitlin and Dixon the day. For instance, only two daily Aqua data points
2010; Norouzi et al. 2012; Prakash et al. 2019). (taken at 0130 and 1330 local time) are available in
Fig. 1. The differences of those two points and the cor-
responding monthly average of the same time of the day
3. Computation of diurnal cycle of LST
are first calculated followed by taking the average of
To characterize the diurnal cycle of global LST under their values. In the next step, the average value calcu-
clear-sky conditions, the MODIS-derived LST data were lated in the latter phase is added to the monthly average
1282 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58

From 2003 to 2017, approximately 31% of global land

area has missing data in the MODIS-derived LST prod-
ucts. Even after generation of synthesized data, there are
still few missing points (about 1.1%) having all-time cloud
contamination for a month that cannot be replaced by
monthly average. The results from the cases that all ob-
servations in a day are missing (such as Fig. 2d) are not
included in studying LST diurnal variations in following
steps. The 15-yr missing data ratio of two contrasting
months (January–July) is 1.1, which states that the number
of missing grids is generally larger during the northern
winter season than the northern summer season.

4. Results and discussion

FIG. 1. Schematic for unavailable data estimation process based on a. Evaluation of interpolated diurnal cycle of LST
the difference of monthly mean LST Tm and observed LST To at each of
To assess the accuracy of the proposed method, a pre-
daily observation time; D1 5 Tm – To (0130 local time) and D2 5 Tm – To
(1330 local time) for a day with only two observations available. liminary evaluation between MODIS-derived LST diurnal
cycle and ground-based observations has been performed
of the unavailable daily data. On the other hand, if there for the year 2017. Hourly LST observations from 18 se-
are missing values at all four times each day, the corre- lected USCRN stations across the North America have
sponding pixel and day is not included in our analysis. been considered. Table 2 presents the location, elevation,
Finally, an interpolation method needs to be applied and land-cover type of the selected USCRN stations. These
with the synthesized data to construct the LST diurnal stations are selected in such a way that they cover a wide
cycle. A spline interpolation method had been success- range of elevations and three distinct vegetation types.
fully utilized to study the diurnal variations of infrared Figures 3a and 3b illustrate the comparison of hourly LST
LST and passive microwave brightness temperatures from the present method and USCRN observations for
(Aires et al. 2004; Norouzi et al. 2012). The spline in- the entire year of 2017 for two stations: Charlottesville,
terpolation method showed smooth diurnal variation of Virginia, and Des Moines, Iowa. Hourly LST from MODIS
LST and there was no need for best-fit to capture mini- is computed from the interpolated half-hourly LST esti-
mum and maximum values between two consecutive mates. Higher correlations of 0.84 and 0.86 and lower root-
observations. Therefore, spline piecewise polynomial in- mean-square errors (RMSE) of 0.25 and 0.15 K between
terpolation has been used in this study to render half- both LSTs for Charlottesville and Des Moines stations,
hourly LST from the MODIS products. Since four respectively, suggest reasonably good performance of the
observations of LST might not be enough as control interpolated LST. The reason for having smaller RMSE
points for obtaining a reliable daily interpolation, the than bias is that to account for systematic biases between
corresponding data for a series of three consecutive days ground-based and satellite measurements, we used ‘‘bias’’
(one day before and one day after) with 12 control points as an indicator. However, to account for only random dif-
were considered for the construction of half-hourly di- ferences, we calculated the RMSE values after the removal
urnal cycle. Figure 2 presents the diurnal variations of of bias estimates. The primary objective of this study is to
LST at four locations having distinct vegetation types interpolate diurnal variations of LST from the satellite es-
using original MODIS products and from interpolated timates. So, we removed the bias in order to compute
half-hourly LST values for February 2017. In each plot of RMSE to just see the random variations in LST diurnally.
the figure, available daily observations from the original The mean diurnal cycles of LST for the two contrasting
MODIS products (red star), the monthly mean LST months of January and July from the present method and
(green triangle), synthesized LST (blue circle), and in- USCRN observations for these two stations are shown in
terpolated LST (pink dot) based on synthesized data are Figs. 3c and 3d. The patterns of diurnal cycle of LST are
shown. Missing data from the MODIS observations can similar for both months from both datasets. However, there
be encountered in four different ways that are also illus- is an underestimation of diurnal cycle magnitude by the
trated in Fig. 2. The general diurnal patterns of LST and present method as compared to the USCRN observations.
their variations with distinct land-cover types (Prakash Figure 4a shows the scatterplot between hourly
et al. 2019) are well-depicted by the interpolated LST. LST estimates from the present method and USCRN
JUNE 2019 SHARIFNEZHADAZIZI ET AL. 1283

FIG. 2. Diurnal variations of interpolated LST for four sample locations with different data availability and for
distinct land cover. The original MODIS-derived LST, synthesized LST, and monthly mean LST are also shown.

observations for all selected stations. There is a good spatial resolutions. The satellite measures area-averaged
agreement (linear correlation of 0.86) between both LST LST, while the USCRN stations provide point observa-
estimates can be seen for the year 2017. The bias and tions representing localized LST. Furthermore, the
RMSE (after bias removal) are appreciably less than 1 K, MODIS sensor records LST above the tree canopies over
revealing that the present model is capable in capturing forest areas (Li et al. 2013; Zhan et al. 2013), while the
the diurnal variations of LST reasonably well for different USCRN stations record data at the surface. The USCRN
land-cover types. The agreement between both LST es- provides LST observations for all weather conditions,
timates is rather weaker around 273 K, corresponding to whereas the interpolated LST has been derived from
freeze–thaw transitions. LST from the present model clear-sky condition LST alone. These factors might be
underestimates during these periods as compared to the intermittently contributing the discrepancies between
USCRN observations. A cold bias in MODIS LST is both LST estimates.
expected during the freezing time periods probably be-
b. Spatial distributions of global land DTR and
cause of the overrepresentation of clear-sky conditions
timing of daily maximum LST
(Muster et al. 2015). Figure 4b presents the diurnal cycle
of correlation and RMSE combined for all selected sta- DTR is defined as the difference between daily max-
tions. The quality of the interpolated LST is not consis- imum and minimum surface air temperature and is
tent throughout the day, and error characteristics vary proven to be a better indicator of climate change and
with time. The differences of LST between MODIS and variability (Braganza et al. 2004; Thorne et al. 2016).
USCRN could be primarily attributed to the facts that The diurnal cycle of LST enables the computation of
these two estimates capture the temperature at different magnitudes and time of occurrence of daily maximum
1284 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58

TABLE 2. Location, elevation, and land-cover types of U.S. Climate Reference Network (USCRN) stations used for evaluation of
MODIS-derived hourly land surface temperatures.

Station No. Station name Lat (8) Lon (8) Elevation (m) LC type
1 Lafayette, Louisiana 30.09 291.87 11 LC02
2 Selma, Alabama 32.46 287.24 63
3 Blackville, South Carolina 33.36 281.33 99
4 Millbrook, New York 41.79 273.74 142
5 Charlottesville, Virginia 38.00 278.47 206
6 Watkinsville, Georgia 33.78 283.39 218
7 Limestone, Maine 46.96 267.88 239
8 Des Moines, Iowa 41.56 293.29 261
9 Crossville, Tennessee 36.01 285.13 578
10 Merced, California 37.24 2120.88 24 LC03
11 Darrington, Washington 48.54 2121.45 110
12 Sebring, Florida 27.15 281.37 46 LC06
13 Chillicothe, Missouri 39.87 293.15 255
14 Manhattan, Kansas 39.10 296.61 357
15 Aberdeen, South Dakota 45.71 299.13 597
16 Wolf Point, Montana 48.31 2105.10 632
17 Monahans, Texas 31.62 2102.81 828
18 Williams, Arizona 35.76 2112.34 1821

and minimum LSTs. The difference in daily maximum 2018). A close association of vegetation type with near-
and minimum LST estimates is termed DTR hereafter surface air temperature diurnal cycle was also reported
in this study. Figures 5a and 5b illustrate the global in earlier studies (Lim et al. 2008; Jackson and Forster
monthly mean DTR derived from the interpolated half- 2010; Scheitlin and Dixon 2010).
hourly LST for the two contrasting months of January Figures 5c and 5d present the spatial distributions of
and July 2017. Larger DTR during the summer than standard deviation in DTR for January and July 2017.
winter is evident in both hemispheres. The largest The tropical vegetated areas such as tropical rain forests
magnitudes of DTR (25–40 K) are evident in primarily show generally small standard deviation of DTR
two land-cover types—nonvegetated deserts, and Bo- (,3 K). A larger standard deviation in DTR of more
real and xeromorphic shrublands with low vegetation than 5 K can be seen over the high-latitude regions that
density. For instance, DTR is consistently large over is more pronounced during the northern winter. High-
the Northern Hemispheric regions of Great Basin of latitude regions eventually receive snowfall during the
Nevada, Colorado Plateau, the Sahara in North Africa, northern winter season, which substantially affects the
Saudi Arabia, and the Gobi Desert in China and DTR variability. A similar seasonal and location-
Mongolia. In the Southern Hemisphere, the Kalahari specific variation in DTR of surface air over land due
Desert in the southern Africa, shrublands of the Andes to combined effect of solar radiation, elevation, cloud
Mountains in South America, and Australian deserts are cover, precipitation, and soil moisture was also reported
three regions with higher DTR. As land surface warms by Jackson and Forster (2010).
and cools rapidly, the absence or lack of vegetation and Figure 6 illustrates the spatial distributions of nor-
moisture across the desert regions make DTR larger malized standard deviation of monthly mean DTR for a
than other regions of Earth. The lowest DTR of less than 15-yr period. It is computed by normalizing the standard
5 K can be found in two main areas: tropical rain forests deviation of monthly mean DTR with respect to the
and high-latitude regions of Russia (Siberia) and 15-yr mean DTR. The global desert regions show ex-
Greenland. In the northern high latitudes, ice and snow ceptionally small variability of DTR (,15%) for the
cover cool the surface air and act as insulators to the study period. However, larger areas of the global land
solar radiation, which maintain rather smaller DTR. It exhibit larger variability in DTR ranging from 20% to
can also be seen that DTR is essentially larger over the 40% for the study period. This larger variability in DTR
high-latitude regions during the northern summer than might be associated with changes in the state of the at-
the winter. In general, the maps of DTR pose a sys- mosphere, albedo, moisture, land emissivity, and land
tematic relation with land-cover type. Land-cover type surface properties (Duan et al. 2014a). The gradual changes
is one of the most important parameters to influence the in vegetation cover are vital for DTR variability. For in-
global LST change (Duveiller et al. 2018; Song et al. stance, the expansion of agricultural land in the tropical
JUNE 2019 SHARIFNEZHADAZIZI ET AL. 1285

FIG. 3. Comparison between MODIS and USCRN hourly LST for 2017. (top) Scatterplots for hourly LST at (a) Charlottesville,
VA, and (b) Des Moines, IA. Correlation coefficient r, bias, and root-mean-square errors (after bias removal) are also provided.
(bottom) Mean diurnal variation for LST at (c) Charlottesville and (d) Des Moines for the two contrasting months of January
and July.

regions enhanced the surface albedo and hence, con- maximum LST for each pixel was computed from
siderably alters the local LST through radiative prop- 48 half-hourly interpolated LST values. The spatial
erties of Earth’s energy balance (Duveiller et al. 2018). distributions of the monthly mean daily maximum LST
The influence of changes in vegetation cover along with hour are presented in Fig. 7 for two contrasting months,
air pollutant on the global LST has also been studied January and July 2017. Comparison of these two maps
recently by Song et al. (2018). The impact of solar ra- reveals a tangible seasonal variation. For the Northern
diation, soil moisture, vegetation cover, and other pa- Hemisphere, with changing seasons from winter to
rameters on DTR was also assessed through the surface summer, timing of maximum hour moves, generally,
energy budget by Jackson and Forster (2010), and it was from around 1330 to around 1230 local time. Similar
noticed that surface net longwave radiation explains characteristics with different time variation range can be
about 95% of the seasonal variation of DTR in the seen in the Southern Hemisphere. Furthermore, it can
tropical regions. be inferred that the time of maximum daily LST de-
The occurrence time of the daily maximum LST is also pends mainly on the latitude. However, higher latitudes,
investigated in this study. The occurrence time of daily especially the Arctic regions, are subjected to some
1286 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58

FIG. 4. Comparison between MODIS and USCRN data in 2017 for all 18 selected stations. (a) Scatterplot for hourly LST. Correlation
coefficient r, bias, and root-mean-square errors are also provided. (b) Diurnal analysis of coefficient of determination R2 and root-mean-
square error using combined data for 18 stations. RMSE values were computed after bias removal.

inconsistencies. The observed inconsistencies can be method may not be able to estimate the timing of
attributed to the snow cover emissivity, which acceler- maximum daily temperature accurately.
ates the reflection of heat energy rather than keeping it
c. Assessment of changes in global land DTR and
on the surface skin. This point can be seen more clearly
timing of daily maximum LST
in Fig. 8, which illustrates the 15-yr average of maximum
LST hour standard deviation. The higher standard de- Review and analysis of the average LST for the years
viation over that region might be because of inaccuracy 2003–17 may reveal the causes of seasonal differences
in diurnal estimation. Since the diurnal temperature between land cover. Thus, 15-yr monthly mean of DTR
changes in those areas are low, spline interpolation and timing of maximum LST for each hemisphere are

FIG. 5. Spatial distributions of (a),(b) monthly mean DTR and (c),(d) its standard deviation for (a),(c) January and (b),(d) July 2017.
JUNE 2019 SHARIFNEZHADAZIZI ET AL. 1287

FIG. 6. Spatial distributions of normalized standard deviation (standard deviation divided by

mean) of monthly DTR for 2003–17.

shown in Fig. 9 as a function of land-cover types. DTR be noted that given the high level of humidity in rain
mean of the Northern Hemisphere also shows a con- forests of tropical regions, it is difficult to detect sub-
sistent seasonal pattern, larger amplitude in summer stantial differences between winter and summer. Duan
than winter for all land-cover types except for tropical/ et al. (2018) noticed rather larger LST discrepancy in
subtropical evergreen broadleaved forest. Interestingly, MODIS products over grassland vegetation cover sites
DTR of the Southern Hemisphere does not line up with due to misclassification of the MODIS land-cover type
the DTR of the Northern Hemisphere for four land- products, resulting to incorrect and unreliable estima-
cover types including tropical/subtropical evergreen tion of the surface emissivity. This study uses a static
broadleaved forest, deciduous forest, deciduous wood- land-cover type and the deciduous forest and deciduous
land, and wooded and nonwooded grasslands. It should woodland regions were subject to deforestation and

FIG. 7. Spatial distributions of time (local time) of maximum LST for (a) January and
(b) July 2017.
1288 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58

FIG. 8. Spatial distributions of standard deviation in time (local time) of maximum LST for the
period of 2003–17.

shift into croplands and grasslands in the recent de- for land and ocean. Land heats up and cools down much
cades. As mentioned earlier, land-cover alteration faster relative to the ocean. Given the fact that North-
changes the reflected solar radiation and latent heat flux ern Hemisphere contains the majority of Earth’s
leading to extreme DTR (Duveiller et al. 2018; Song landmass, a much larger population (roughly 88%), and
et al. 2018). One of the main factors for the notable much more industrialization, pollution variations are
distinction of DTR over the Northern and Southern more considerable compared to the Southern Hemi-
Hemispheres is the distribution of land versus ocean. sphere. The land-cover average of maximum LST hour
About 39% of the Northern Hemisphere is land shows a consistent seasonal pattern for both hemi-
whereas only 19% of the Southern Hemisphere is cov- spheres. The maximum LST hour occurs earlier in
ered by land. Rates of heating and cooling are different summer, for example, July in the Northern Hemisphere

FIG. 9. Mean (a),(b) DTR and (c),(d) time of maximum LST for the (a),(c) Northern and (b),(d) Southern Hemispheres averaged over
different land-cover types for January and July. The full names of each land-cover type are given in Table 1.
JUNE 2019 SHARIFNEZHADAZIZI ET AL. 1289

FIG. 10. Spatial distributions of linear trend (K decade21) in monthly DTR for the period 2003–17.

and January in the Southern Hemisphere for almost all that the assessment was done for a 15-yr period of the
the vegetation classes. MODIS era, which is not adequate for a robust climatic
To examine the changes in mean monthly DTR between trend analysis. It is also worth noting that the obtained
2003 and 2017, linear trend in DTR over the global land is conclusions could be associated with the normalized dif-
presented in Fig. 10. Larger areas show a decline in DTR, ference vegetation index, which needs further investigation.
which reveals that daily minimum temperature increases
at a faster rate than the daily maximum temperature. The
5. Conclusions
Northern Hemisphere showed an overall decrease in DTR
of 0.54 K between 2003 and 2017. However, southern In this paper, a 15-yr data record of LST Collection
Europe and the Middle East exhibit an increase in DTR 6 from the MODIS sensors was studied. A spline in-
during the study period. Deciduous woodland, wooded terpolation method was utilized in order to generate
and nonwooded grassland, and nonvegetated desert land half-hourly LST values over the globe land. An initial
classes show the largest decline. Among them, wooded and evaluation of interpolated LST with hourly ground-based
nonwooded grassland vegetation type has the largest var- observations from the selected 18 USCRN stations over
iations of negative and positive slope values, with the the North America showed a linear correlation of 0.86
negative parts overcoming the positive parts. The trend and a bias and an RMSE of less than 1 K. In general, the
fluctuation in grassland cover can be associated with the present interpolation method showed promising results in
fact that a majority of existing grasslands used to be forests capturing the diurnal variations of LST for different land-
and woodlands. As previously explained, higher radiative cover types. The general patterns of LST diurnal varia-
flux leaving the surface is as a consequence of deforesta- tions were well depicted by the interpolated LST for
tion. However, the balance between shortwave reflected different land-cover types. The interpolated half-hourly
radiation and longwave released radiation can change LST data were used to calculate two important parame-
based on the forest canopy (Duveiller et al. 2018). ters including DTR as well as timing of the maximum
Interestingly, the Southern Hemisphere shows variant LST hour. The largest DTR of 25–40 K was found over
changes in DTR with land-cover classes. A strong increase the desert regions, whereas the smallest DTR of less than
of DTR in deciduous woodland of southern Africa, wooded 5 K was evident over the high-latitude regions of the
and nonwooded grassland and shrubland of northeastern Northern Hemisphere. The global desert regions showed
part of South America, and a strong decrease in DTR smaller variability of monthly mean DTR, and larger
for sclerophyllous woodland and forest were observed. areas of the global land exhibited larger variability in
Wooded and nonwooded grassland and boreal and DTR between 20% and 40% for the 15-yr period. The
xeromorphic shrubland are two land-cover classes having time of maximum LST hour moves to about 1 h earlier
the most variations from highest positive slope value in during summer as compared to during winter in both
southern Africa to highest negative in southern Australia. hemispheres. However, mean DTR showed distinct
Sclerophyllous woodland and forest of Tasmania and east- seasonal variations as a function of land-cover types.
ern Australia, and nonvegetated deserts of central Australia The Northern Hemispheric land showed a decrease in
and Sahara in northern Africa are the two land-cover types monthly mean DTR by about 0.54 K between 2003 and
showing a consistent negative slope. It should be considered 2017. But, the Southern Hemisphere showed variant
1290 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 58

changes in DTR with land-cover classes. However, the Bell, J. E., and Coauthors, 2013: U.S. Climate Reference Network
assessment was done for a 15-yr period and is not ade- soil moisture and temperature observations. J. Hydrometeor.,
14, 977–988, https://doi.org/10.1175/JHM-D-12-0146.1.
quate for a robust climatic trend analysis.
Braganza, K., D. J. Karoly, and J. M. Arblaster, 2004: Diurnal
Furthermore, there are few limitations of this study. temperature range as an index of global climate change during
The interpolated LST has larger uncertainty over the the twentieth century. Geophys. Res. Lett., 31, L13217, https://
high-latitude regions because of low DTR, which needs doi.org/10.1029/2004GL019998.
further investigation. The use of a static land-cover type Diamond, H. J., and Coauthors, 2013: U.S. Climate Reference
Network after one decade of operations: Status and assess-
might also contribute to uncertainty in the results because
ment. Bull. Amer. Meteor. Soc., 94, 485–498, https://doi.org/
of considerable changes in the global land-cover types 10.1175/BAMS-D-12-00170.1.
during the study period. The diurnal cycle was con- Didari, S., H. Norouzi, S. Zand-Parsa, and R. Khanbilvardi, 2017:
structed using cloud-free LST observations in this study, Estimation of daily minimum land surface air temperature
which might lead to uncertainty in the presence of clouds. using MODIS data in southern Iran. Theor. Appl. Climatol.,
There are differences in local solar time for the same pixel 130, 1149–1161, https://doi.org/10.1007/s00704-016-1945-0.
Duan, S.-B., Z.-L. Li, B.-H. Tang, H. Wu, R. Tang, Y. Bi, and
on different revisit days of the MODIS instruments be- G. Zhou, 2014a: Estimation of diurnal cycle of land surface
cause of its intrinsic scanning characteristics. Typically, 3– temperature at high temporal and spatial resolution from
5 K of LST difference could be introduced because of clear-sky MODIS data. Remote Sens., 6, 3247–3262, https://
changes in daytime observations between 1000 and 1200 doi.org/10.3390/rs6043247.
local solar time from the Terra satellite (Duan et al. ——, ——, ——, ——, and ——, 2014b: Generation of a time-
consistent land surface temperature product from MODIS
2014b). A temporal normalization scheme needs addi-
data. Remote Sens. Environ., 140, 339–349, https://doi.org/
tional input parameters such as elevation, vegetation 10.1016/j.rse.2013.09.003.
type, and normalized difference vegetation index corre- ——, ——, and P. Leng, 2017: A framework for the retrieval of all-
sponding to LST observations, and the uncertainty of the weather land surface temperature at a high spatial resolution
normalized LST generally becomes larger than the orig- from polar-orbiting thermal infrared and passive microwave
data. Remote Sens. Environ., 195, 107–117, https://doi.org/
inal LST values because of uncertainties of the input
10.1016/j.rse.2017.04.008.
parameters. Hence, there is a need to develop a suitable ——, ——, H. Wu, P. Leng, M. Gao, and C. Wang, 2018: Radi-
temporal normalization scheme applicable for MODIS ance-based validation of land surface temperature products
global LST products that would essentially improve the derived from Collection 6 MODIS thermal infrared data.
accuracy of the interpolated diurnal cycle. Nonetheless, Int. J. Appl. Earth Obs. Geoinf., 70, 84–92, https://doi.org/
10.1016/j.jag.2018.04.006.
augmentation of a global network of ground-based ob-
Duveiller, G., J. Hooker, and A. Cescatti, 2018: The mark of veg-
servations is vital for the comprehensive evaluation of etation change on Earth’s surface energy balance. Nat. Com-
satellite-derived LST estimates. mun., 9, 679, https://doi.org/10.1038/s41467-017-02810-8.
Ignatov, A., and G. Gutman, 1999: Monthly mean diurnal cycles in
surface temperatures over land for global climate studies.
Acknowledgments. This study was supported by the J. Climate, 12, 1900–1910, https://doi.org/10.1175/1520-0442(1999)
National Oceanic and Atmospheric Administration– 012,1900:MMDCIS.2.0.CO;2.
Cooperative Science Center for Earth System Sciences and Jackson, L. S., and P. M. Forster, 2010: An empirical study of
Remote Sensing Technologies (NOAA–CESSRST) under geographic and seasonal variations in diurnal temperature
range. J. Climate, 23, 3205–3221, https://doi.org/10.1175/
the Cooperative Agreement Grant NA16SEC4810008, by 2010JCLI3215.1.
the NASA under Grant NNH15ZDA001N, and by the Justice, C. O., and Coauthors, 1998: The Moderate Resolution
Center for Remote Sensing and Earth System Sciences Imaging Spectroradiometer (MODIS): Land remote sensing
at the New York City College of Technology. The au- for global change research. IEEE Trans. Geosci. Remote Sens.,
thors thank the editor and anonymous reviewers for their 36, 1228–1249, https://doi.org/10.1109/36.701075.
Kilibarda, M., T. Hengl, G. B. M. Heuvelink, B. Graler,
constructive comments. MODIS land surface tempera-
E. Pebesma, M. P. Tadic, and B. Bajat, 2014: Spatio-temporal
ture datasets obtained from the NASA EOSDID LP interpolation of daily temperatures for global land areas at
DAAC (https://lpdaac.usgs.gov/) and USCRN obser- 1 km resolution. J. Geophys. Res. Atmos., 119, 2294–2313,
vations from the NOAA’s National Climatic Data https://doi.org/10.1002/2013JD020803.
Center (NCDC) are thankfully acknowledged. Li, Z.-L., S.-H. Tang, H. Wu, H. Ren, G. Yan, Z. Wan, I. F. Trigo,
and J. A. Sobrino, 2013: Satellite-derived land surface tem-
perature: Current status and perspectives. Remote Sens. En-
REFERENCES viron., 131, 14–37, https://doi.org/10.1016/j.rse.2012.12.008.
Lim, Y.-K., M. Cai, E. Kalnay, and L. Zhou, 2008: Impact of vegetation
Aires, F., C. Prigent, and W. Rossow, 2004: Temporal interpolation types on surface temperature change. J. Appl. Meteor. Climatol.,
of global surface skin temperature diurnal cycle over land 47, 411–424, https://doi.org/10.1175/2007JAMC1494.1.
under clear and cloudy conditions. J. Geophys. Res., 109, Matthews, E., 1983: Global vegetation and land use: New high-
D04313, https://doi.org/10.1029/2003JD003527. resolution data bases for climate studies. J. Climate Appl.
JUNE 2019 SHARIFNEZHADAZIZI ET AL. 1291

Meteor., 22, 474–487, https://doi.org/10.1175/1520-0450(1983) Ruzmaikin, A., H. H. Aumann, J. Lee, and J. Susskind, 2017: Di-
022,0474:GVALUN.2.0.CO;2. urnal cycle variability of surface temperature inferred from
Mildrexler, D. J., M. Zhao, W. B. Cohen, S. W. Running, X. P. AIRS data. J. Geophys. Res. Atmos., 122, 10 928–10 938,
Song, and M. O. Jones, 2018: Thermal anomalies detect crit- https://doi.org/10.1002/2016JD026265.
ical global land surface changes. J. Appl. Meteor. Climatol., 57, Scheitlin, K. N., and P. G. Dixon, 2010: Diurnal temperature range
391–411, https://doi.org/10.1175/JAMC-D-17-0093.1. variability due to land cover and airmass types in the South-
Muster, S., M. Langer, A. Abnizova, K. L. Young, and J. Boike, 2015: east. J. Appl. Meteor. Climatol., 49, 879–888, https://doi.org/
Spatio-temporal sensitivity of MODIS land surface temperature 10.1175/2009JAMC2322.1.
anomalies indicates high potential for large-scale land cover Shati, F., S. Prakash, H. Norouzi, and R. Blake, 2018: Assessment
change detection in Arctic permafrost landscapes. Remote Sens. of differences between near-surface air and soil temperatures
Environ., 168, 1–12, https://doi.org/10.1016/j.rse.2015.06.017. for reliable detection of high-latitude freeze and thaw states.
Norouzi, H., W. Rossow, M. Temimi, C. Prigent, M. Azarderakhsh, Cold Reg. Sci. Technol., 145, 86–92, https://doi.org/10.1016/
S. Boukabara, and R. Khanbilvardi, 2012: Using microwave j.coldregions.2017.10.007.
brightness temperature diurnal cycle to improve emissivity Song, Z., and Coauthors, 2018: Global land surface temperature
retrievals over land. Remote Sens. Environ., 123, 470–482, influenced by vegetation cover and PM2.5 from 2001 to 2016.
https://doi.org/10.1016/j.rse.2012.04.015. Remote Sens., 10, 2034, https://doi.org/10.3390/rs10122034.
——, M. Temimi, A. AghaKouchak, M. Azarderakhsh, Thorne, P. W., and Coauthors, 2016: Reassessing changes in di-
R. Khanbilvardi, G. Shields, and K. Tesfagiorgis, 2015: In- urnal temperature range: Intercomparison and evaluation of
ferring land surface parameters from the diurnal variability of existing global data set estimates. J. Geophys. Res. Atmos.,
microwave and infrared temperatures. Phys. Chem. Earth, 83– 121, 5138–5158, https://doi.org/10.1002/2015JD024584.
84, 28–35, https://doi.org/10.1016/j.pce.2015.01.007. Tomlinson, C. J., L. Chapman, J. E. Thornes, and C. Baker, 2011:
Prakash, S., H. Norouzi, M. Azarderakhsh, R. Blake, and Remote sensing land surface temperature for meteorology
K. Tesfagiorgis, 2016: Global land surface emissivity estimation and climatology: A review. Meteor. Appl., 18, 296–306, https://
from AMSR2 observations. IEEE Geosci. Remote Sens. Lett., doi.org/10.1002/met.287.
13, 1270–1274, https://doi.org/10.1109/LGRS.2016.2581140. Wan, Z., 2014: New refinements and validation of the collection-
——, ——, ——, ——, and R. Khanbilvardi, 2017: Potential of 6 MODIS land-surface temperature/emissivity product.
satellite-based land emissivity estimates for the detection of Remote Sens. Environ., 140, 36–45, https://doi.org/10.1016/
high-latitude freeze and thaw states. Geophys. Res. Lett., 44, j.rse.2013.08.027.
2336–2342, https://doi.org/10.1002/2017GL072560. Weng, Q., and P. Fu, 2014: Modeling diurnal land temperature
——, ——, ——, ——, C. Prigent, and R. Khanbilvardi, 2018: Es- cycles over Los Angeles using downscaled GOES imagery.
timation of consistent global microwave land surface emis- ISPRS J. Photogramm. Remote Sens., 97, 78–88, https://
sivity from AMSR-E and AMSR2 observations. J. Appl. doi.org/10.1016/j.isprsjprs.2014.08.009.
Meteor. Climatol., 57, 907–919, https://doi.org/10.1175/JAMC- Yu, W., Z. Nan, Z. Wang, H. Chen, T. Wu, and L. Zhao, 2015: An
D-17-0213.1. effective interpolation method for MODIS land surface tem-
——, F. Shati, H. Norouzi, and R. Blake, 2019: Observed differ- perature on the Qinghai–Tibet Plateau. IEEE J. Sel. Top.
ences between near-surface air and skin temperatures using Appl. Earth Obs. Remote Sens., 8, 4539–4550, https://doi.org/
satellite and ground-based data. Theor. Appl. Climatol., 10.1109/JSTARS.2015.2464094.
https://doi.org/10.1007/s00704-018-2623-1, in press. Zhan, W., and Coauthors, 2013: Disaggregation of remotely sensed
Prigent, C., W. B. Rossow, and E. Matthews, 1998: Global maps of land surface temperature: Literature survey, taxonomy, issues,
microwave land emissivities: Potential for land surface character- and caveats. Remote Sens. Environ., 131, 119–139, https://
ization. Radio Sci., 33, 745–751, https://doi.org/10.1029/97RS02460. doi.org/10.1016/j.rse.2012.12.014.
Rossow, W. B., and R. A. Schiffer, 1999: Advances in un- Zhou, J., Y. Chen, X. Zhang, and W. Zhan, 2013: Modelling the
derstanding cloud from ISCCP. Bull. Amer. Meteor. Soc., 80, diurnal variations of urban heat islands with multi-source
2261–2288, https://doi.org/10.1175/1520-0477(1999)080,2261: satellite data. Int. J. Remote Sens., 34, 7568–7588, https://
AIUCFI.2.0.CO;2. doi.org/10.1080/01431161.2013.821576.