Radiative Cooling Resource Maps for the Contiguous United StatesMengying Li, Hannah B. Peterson, and Carlos F. M. Coimbraa)

Department of Mechanical and Aerospace Engineering, Jacobs School of EngineeringCenter of Excellence in Renewable Resource Integration and Center for Energy ResearchUniversity of California San Diego, 9500 Gilman Drive, La Jolla, CA 92093(Dated: 27 June 2019)

Passive cooling devices take advantage of the partially transparent properties of the atmosphere in the long-wave spectral band from 8 to 13 µm (the so-called ‘atmospheric window’) to reject radiation to outer space.Spectrally-designed thermophotonic devices have raised substantial attention recently for their potential toprovide passive and carbon-free alternatives to air conditioning. However, the level of transparency of theatmospheric window depends on the local content of water vapor in the atmosphere and on the optical depthof clouds in the local sky. Thus, the radiative cooling capacity of solar reflectors not only depends on theoptical properties of their surfaces but also on local meteorological conditions. In this work, detailed radia-tive cooling resource maps for the contiguous United States are presented with the goal of determining thebest climates for large-scale deployment of passive radiative cooling technologies. The passive cooling poten-tial is estimated based on ideal optical properties, i.e., zero shortwave absorptance (maximum reflectance)and blackbody longwave emittance. Both annual and season-averaged maps are presented. Daytime andnighttime cooling potential are also computed and compared. The annual average cooling potential over thecontiguous Unite States is 50.5 m−2. The southwestern United States has the highest annual averaged coolingpotential, over 70 W m−2, due to its dry and mostly clear sky meteorological conditions. The southeasternUnited States has the lowest potential, around 30 W m−2, due to frequent humid and/or overcast weatherconditions. In the spring and fall months, the Arizona and New Mexico climates provide the highest passivecooling potential, while in the summer months, Nevada and Utah exhibit higher potentials. Passive radiativecooling is primarily effective in the western United States, while it is mostly ineffective in humid and overcastclimates elsewhere.Keywords: radiative cooling potential; passive cooling; diurnal and seasonal variations; thermophotonics.

I. INTRODUCTION

Cooling loads for the built environment, includingroads, commercial and residential buildings, factories,power plants and data centers, accounts for a significantportion of the global energy consumption. Commercialbuildings consume 40% of total energy in the EuropeanUnion, with space cooling accounting for nearly half ofthe total power load 1. In the United States, about 39%of the total energy consumed was spent on the residentialand commercial sectors in 2017, with nearly 15% of theenergy consumed for air conditioning purposes 2,3. Over-loaded transformer and transmission line failures due topeak load during the daytime are often caused by air con-ditioning demands 4. Currently, most buildings use re-versible heat pumps or powered air conditioning systemswith associated high electricity loads for space cooling 1.A promising technology for reducing the air condition-ing load on the grid, especially for locations where themajority the heat load is from solar radiation, is the useof passive radiative cooling devices and paints. Theseengineered materials are optically selective for reflecting(shortwave) solar radiance while emitting strongly on the(longwave) infrared portion of the electromagnetic spec-trum. Passive radiative cooling requires little to no elec-tricity input5, and can be deployed in a sustainable way

a)Electronic mail: ccoimbra@ucsd.edu

due to modern photonic design of micro-structured plas-tics and selective paints 6.

The Earth’s atmosphere has low absorptivity for elec-tromagnetic waves within the atmospheric window band(longwave, approximately 8 - 13 µm). Thus, terrestrialobjects with high emissivity within the atmospheric win-dow band can be cooled by radiating longwave energy toouter space when directly exposed to the sky. This formof passive cooling is often referred as ‘radiative cooling’,and has been suggested for nighttime applications7–10. Inthe daytime, even a small value of solar absorptivity willoffset the longwave emitting power of the surface, effec-tively heating the surface. To offset the large cooling de-mand during daytime, particularly under direct sunlight,nanophotonic and polymeric photonics devices have beenoptically designed recently to exhibit near-zero solar ab-sorptivity while retaining the highly emissive propertiesin the atmospheric window band3,6,11–13.

In addition to the physical constraints for spectral op-tical properties of surfaces, the heat rejection from pas-sive cooling devices is also limited by local meteorologicalconditions. The presence of water vapor and clouds in-creases the emissivity (and the absoptivity) of the atmo-sphere, with complementary reduction of the transmis-sivity through the atmospheric window, thus reducingthe ability of surfaces to radiate to outer space4,14–16.Previous literature on modern radiative cooling devicesreport cooling power efficiencies only at one or two test-ing sites during specific (and short) periods of time. For

Preprint, accepted by Journal of Renewable and Sustainable Energy (2019), (11), 036501. https://doi.org/10.1063/1.5094510

Preprint, accepted by Journal of Renewable and Sustainable Energy (2019), (11), 036501. https://doi.org/10.1063/1.5094510

example, Zhai et al.6 reports cooling rate ranges from100 to 145 W m−2 during two clear days in Cave Creek,Arizona, which could only be achieved under cloudlessskies with dry atmospheric conditions, based on the anal-ysis performed by Li and Coimbra16. For mostly hu-mid and/or frequently cloudy climates, the heat rejec-tion capacity of passive devices is necessarily much lower.Hence, the need for radiative cooling resource potentialmaps in order to proper assess the effectiveness of thesepassive technologies. As recently pointed out by Vall andCastell1, very little research has been directed at evalu-ating the radiative cooling potential for diverse climatesand region. A summary of studies related to location-specific radiative cooling resources is presented in TableI (including this work for comparison). Literature foundon this topic is limited, and none of the previous workshave used extensive and up-to-date meteorological mea-surements over continental regions.

The main goal of this work is to present radiativecooling resource maps for the contiguous United States(CONUS), and to examine diurnal and seasonal variabil-ity in order to propose optimal locations for deploymentof passive radiative cooling technologies. The followingsection II presents the theoretical foundations for calcu-lating radiative cooling potentials. Section III presentsthe methodology for processing the meteorological dataused in this work. The passive cooling resource maps arepresented in section IV, and conclusions follow in sectionV.

II. RADIATIVE COOLING VIA THE LONGWAVEATMOSPHERIC WINDOW

A. Principles of radiative cooling

For a body with temperature Ts facing open sky (seeFig. 1), the broadband cooling power of the body is

qcool(Ts) =

∫ ν1

0

εs,lw[πIb(ν, Ts)− Jsky(ν, Ta)]dν

−∫ ∞

ν1

αs,swqsun(ν)dν + hc(Ts − Ta),

(1)

where ν1 =2500 cm−1 is the wavenumber cutoff be-tween solar shortwave and atmospheric longwave ir-radiance; εs,lw is the monochromatic longwave emit-tance/absorptance of the body; Ib(ν, Ts) [W m−2 sr−1 cm]is the monochromatic blackbody emissive intensity of thebody, following Planck’s law; Jsky(ν, Ta) [W m−2 cm] isthe monochromatic sky radiosity at ambient temperatureTa (detailed in section II B); αs,sw is the monochromaticsolar shortwave absorptance of the body; qsun(ν) [W m−2

cm] is the spectral solar irradiance on the body and hc

[W m−2 K−1] is the heat transfer coefficient consideringboth convective and conductive heat loss.

Figure 2 plots the spectral sky radiosity and solar ir-radiance for a cloudless atmosphere using previously de-

FIG. 1. Schematic of the energy balance for a passive radia-tive surface under clear sky conditions. With the exceptionof the direct component of the shortwave solar radiation, allother components and surfaces are assumed to exchange ra-diation diffusively.

veloped radiative models 25,26. In the longwave spectrumfrom 8 to 13 µm, the cloudless atmosphere has a smallerradiosity than the blackbody emissive power of terrestrialobjects, making the first term in Eq. (1) positive. There-fore, radiative cooling is mostly via this spectral band,the so-called ‘atmospheric window’.

When using spectrally averaged properties, the coolingpower is

qcool(Ts) = εs,lw[σT4s −εskyσT

4a ]−αs,swGsun+hc(Ts−Ta),

(2)where σ = 5.67×10−8 W m K−4 is the Stefan-Boltzmannconstant and εsky is the effective emissivity of the sky,which depends on local atmospheric conditions.

To achieve radiative cooling power during both thedaytime and nighttime, the radiative coolers need to behighly reflective in the shortwave solar spectrum whilehighly emissive in the longwave spectrum. For example,Zhai et al. 6 developed a material that has a spectrally av-eraged solar absorptance of 0.04 and longwave emittanceof 0.93. For maximum radiative cooling power, the ra-diative cooler should have reflectance approaching unity(absorptance approaching zero) in the solar spectrum andemittance approaching unity in the longwave spectrum.In addition, convection and conduction heat loss shouldbe suppressed by means of insulation and/or engineeredhoneycomb to suppress the onset of fluid motion 27. Ide-ally, the maximum cooling power (cooling potential) is

qcool,ideal(Ts) = σT 4s − εskyσT

4a . (3)

When the passive cooling surface is kept at tempera-ture Ts = Ta (to further minimize convective and con-ductive losses), the cooling power flux at the surface is

qcool,ideal(Ta) = σT 4a (1− εsky), (4)

with cooling power increasing with decreasing values ofsky emissivity. High water vapor content and the pres-ence of clouds tend to increase the sky emissivity (de-tailed in the next section), thus decreasing the coolingpotential16.

2

Preprint, accepted by Journal of Renewable and Sustainable Energy (2019), (11), 036501. https://doi.org/10.1063/1.5094510

TABLE I. Summary of literature related to location specific radiative cooling resources.

Authors PublishYear Region Reported parameters Data

durationNumber oflocations Frequency Contour

mapAtwater andBall 17 1978 United States effective sky temperature,

sky temperature depression∗ 1971-1972 11 seasonal Y

Exell 18 1978 Thailand sky temperature depression 1969 4 seasonal NPissimanis andNotaridou 19 1981 Athens, Greece effective sky temperature 1969-1970,

1972-1976 1 monthly N

Martin andBerdahl 20 1984 United States

sky temperature depression,percentage of hours with skytemperatures below 16◦C

TMY∗ 211 monthly Y

Schmetz et al 21 1986 Europe downwelling longwave flux 1981-1982 satellite specificdays Y

Argiriou et al 22 1994 Athens, Greece sky temperature depression 1977-1989 1 monthly NMahlia et al 23 2014 Malaysia radiative cooling power 2012 10 monthly NZhang et al 24 2018 United States cooling power NA 4 monthly NPresent work (Li et al.) 2019 United States radiative cooling resource 2017 1681 annual,

seasonal Y∗The sky temperature depression is the temperature difference between sky and surface.∗TMY stands for typical meteorological year.

0.00

0.02

0.04

0.06

0.08

0.10

Spec

tral f

lux

dens

ity [W

cmm

2 ]

cooler radiositysky radiositysolar irradiance

10 100 1000 2500 5000 10000 20000 40000

Wavenumber [cm 1]

0.00

0.05 cooling power

0.40.71.04.08.013.050.0100.01000.0Wavelength [ m]

FIG. 2. Model of spectral solar irradiance and sky radiosity under clear skies when ambient temperature Ta = 294.2 K, relativehumidity ϕ = 70% and solar zenith angle θz = 30◦25,26. The radiosity of an ideal passive cooling device (maximum reflectancefor shortwave solar radiation, and maximum emittance for longwave radiation) is also shown. The cooling power is shown belowwithin the atmospheric window band of the longwave spectrum. For clarity of display, longwave power flux density is scaleddown by a factor of 5.

B. Atmospheric condition effects

A spectrally resolved two-flux radiative model was de-veloped to compute spectral longwave radiative transfer(wavelength greater than 4µm) for the Earth-atmospheresystem under clear skies with different water vapor con-tents 25. As shown in Fig. 2, the clear sky has a trans-parent window from 8 to 13 µm, through which radiativecooling is possible. As discussed above, the transparencyof this window decreases with increased water vapor con-tent 16 and with the presence of clouds. The model allowsus to quantify these effects on the passive cooling poten-tial for ideal surfaces.

Water clouds are modeled using similar methods as

used for aerosols. Each water droplet is assumed tohave a spherical shape, thus the absorption and scat-tering efficiencies of droplets are calculated using Mietheory results25, with the refraction index of water re-trieved from Hale and Querry (1973) 28. The absorptionand scattering coefficients as well as the asymmetry fac-tors of clouds are further calculated by integrating theefficiencies over a model droplet size distribution29. Thesize distribution of droplets in clouds is assumed to followa Gamma distribution30–33

n(r) = r1/σe−3 exp

(− r

reσe

), (5)

where re and σe are the mean effective radius and vari-

3

Preprint, accepted by Journal of Renewable and Sustainable Energy (2019), (11), 036501. https://doi.org/10.1063/1.5094510

0 5 10 15 20Radii of droplets [ m]

0

200

400

600

Num

ber d

ensi

ty(a)

1 5 10 20 50 100 500 2500 20000

Wavenumber [cm 1]

0.00

0.25

0.50

0.75

1.00

1.25

1.50

COD = 1.0

(b)

optical depthasymmetry factorsingle albedo

0.54.010.0500.05000.0Wavelength [ m]

FIG. 3. (a) Size distribution of droplets in the model cloudsfor re = 10µm. (b) Spectral optical depth, single albedo andasymmetry factor of model clouds for unity value of COD.The subscript 497.5 nm is omitted in the text for clarity.

ance, respectively. Here we assume the mean effectiveradius re = 10.0 µm and σe = 0.1, as suggested in Barkeret al. (2003) and Walther et al. (2011) and used in severalgeoestationary satellite products32,33. Miles et al. (2000)suggested that an effective radius of 5.4µm may betterrepresent continental clouds. We use the smaller effectiveradius distribution to assess the uncertainty of the mod-eled cooling potential (see Fig. 5). The spectral disper-sion k often used to represent the spread of droplet size(instead of σe) is defined as k = (1/σe − 2)−1/2. Whenσe = 0.1, k = 0.354, which is a typical value for bothmarine and continental clouds31. The size distributionof water droplets is shown in Fig. 3 (a).

Clouds are considered into layers with a predefinedoptical depth (COD) at 497.5 nm defined as COD =κe,cld∆Hc where κe,cld[cm−1] is the extinction coefficientof clouds and ∆Hc [cm] is the thickness of clouds. Thespectral optical depth, single albedo and asymmetry fac-tors of clouds are shown in Fig. 3 (b) for COD = 1.0.

The extinction coefficient, single albedo and asymme-try factor of each atmospheric layer containing clouds are

κe = κe,gas + κe,aer + κe,cld,

ρ =κs,aer + κs,cld

κe,

eg =eg,aerκs,aer + eg,cldκs,cld

κs,aer + κs,cld,

(6)

where the asymmetry factor eg is the weighted sum forthe mixture of aerosols and clouds using scattering coef-ficients as the weights. The scattering is then treatedas isotropic using the δ-M scaling method34, and themonochromatic fluxes are solved using the proceduresoutlined in Li et al. (2018)25.

For partly cloudy skies, the spectral downwelling long-wave irradiance (sky radiosity) calculated from the modelis expressed as

Jsky(ν, Ta) = Jsky,c(ν, Ta)(1−CF)+Jsky,oc(ν, Ta)CF (7)

where ν is the wavenumber [cm−1], Ta [K] is the ambi-ent air temperature at the screening level (10 m above

the surface), CF is the cloud fraction and the subscripts‘c’ and ‘oc’ represent clear skies and overcast skies, re-spectively. Then the spectral cooling power of an idealcooler at ambient temperature is calculated from Eqs. (1)to (4). Figure 4 shows the spectral longwave coolig powerfor Ts = Ta = 294.2 K and ϕ = 70%. Clearly, the cool-ing power decreases when more clouds are present in theatmosphere because clouds also ‘block’ the atmosphericwindow. Figure 5 shows the broadband cooling poten-tial with respect to both COD and cloud base temper-ature (CBT) when CF = 1.0. The ambient is at 294.2K and ϕ =70%. The cooling power for clear skies atthe nominal conditions described above is 71.9 W m−2

for the AFGL midlatitude summer temperature profile,and it decreases with increasing COD and CBT. Differ-ent temperature profiles (corrected to match the ambientconditions) introduce less than 5 W m−2 deviations tothe cooling power potential. When the effective radiusof clouds is assumed to be 5.4µm, the cooling power isincreased by less than 10 W m−2 when compared withthe cooling power calculated with re = 10µm.

0 250 500 750 1000 1250 1500 1750 2000 2250 2500

Wavenumber, cm 1

0.0

0.1

0.2

0.3

0.4

0.5C

oolin

g flu

x de

nsity

, W c

m m

2

cooler emissionCF= 0CF= 0.3CF= 0.6CF= 0.9

4.05.06.08.010.015.030.0150.0Wavelength, m

FIG. 4. Spectral longwave cooling power with respect to cloudfraction under partly cloudy skies when ambient temperatureTa = 294.2 K and ϕ = 70%. Model clouds are assumed tohave base height of 0.54 km, thickness of 0.44 km, and opticaldepth 5.0 at 497.5 nm.

If detailed atmosphere temperature, pressure and con-centration vertical profiles are specified, the sky radiositycan be calculated precisely using the radiative transfermodel. If only ambient temperature and relative hu-midity are given, empirical expressions of effective skyemissivity can be used. The empirical all-sky effectiveemissivity of the sky for the contiguous United Stateshas been recently determined to be35:

εsky = εsky,c(1− 0.78CF) + 0.38CF0.95ϕ0.17,

εsky,c = 0.618 + 0.056√

Pw,(8)

where εsky,c is the effective sky emissivity under clearskies, ϕ is ambient relative humidity and Pw [Pa] is am-

4

Preprint, accepted by Journal of Renewable and Sustainable Energy (2019), (11), 036501. https://doi.org/10.1063/1.5094510

270 275 280 285 290 295Cloud base temperature [K]

0

10

20

30

40

50

60

70

80C

oolin

g po

tent

ial [

W m

2 ]COD = 0.0

COD = 0.1

COD = 1.0

COD = 10.0

COD = 0.0

COD = 0.1

COD = 1.0

COD = 10.0

COD = 0.0

COD = 0.1

COD = 1.0

COD = 10.0

(a) 10 m effective radius

270 275 280 285 290 295Cloud base temperature [K]

(b) 5.4 m effective radius

AFGL midlatitude summerAFGL US standardAFGL tropical

FIG. 5. Broadband cooling potential for different cloud basetemperatures and cloud optical depths when ambient Ta =294.2 K, ϕ = 70% and cloud fraction is 1.0. Panel (a) showsresults for model clouds effective radius re equal to 10 µm,while panel (b) shows results for re equal to 5.4 µm. Coolingpotentials for different temperature profiles are also plottedfor comparison.

bient water vapor partial pressure. From local measure-ments of ambient temperature, relative humidity (or dewpoint) and cloud fraction, the effective sky emissivity andradiative cooling potentials can be calculated. The meanbias error found by using the above Eq.(8) to estimateDLW is -4.94 W m−2 35, which implies that the uncer-tainty in estimating seasonally and annually averagedcooling potentials is of the order of 5 W m−2.

III. PROCESSING OF METEOROLOGICAL DATA

A. Data retrieval

Local meteorological measurements were retrievedfrom the National Oceanic and Atmospheric Adminis-tration (NOAA) Automated Surface Observing System(ASOS) network via direct FTP access 36. A total of1681 stations across the contiguous United States wereused to collect data for the 2017 year (see Fig. 6 for thespatial distribution of the stations). Most stations con-sidered in this work reported a full year worth of data,with some reporting incomplete or low quality data peri-ods, and those periods were removed from consideration.

Each station records intra-hour measurements of me-teorological variables, including air temperature (Ta),dew point (Td) and cloud fraction (CF). Only measure-ments designated as having passed all quality controlchecks (based on QC codes provided by ASOS) wereused. ASOS data has multiple entries of cloud fractionvalues based on cloud types, and the CF values used inthis work are the average of all entries.

From air temperature Ta and dew point Td, the watervapor partial pressure Pw and relative humidity ϕ are

FIG. 6. Location of the 1681 ASOS stations used in this work

calculated using Magus-like expressions 37

Pw = P0 exp

(cTTd

Td + T0

)ϕ =

Pw

P0 exp(

cTTa

Ta+T0

) (9)

where P0 = 610.94 Pa, cT =17.625 and T0=243.04◦C.Note that in the above two equations, temperatures Ta

and Td have units of ◦C.The radiative cooling potential is then calculated from

Ta, Pw, ϕ and CF using Eqs. (4) and (8) for each en-try. Data is separated into the four seasons accordingto the timestamps of the entries. Timestamps are alsoused to calculate the solar zenith angles of the entries toseparate them into daytime or nighttime periods. Clearsky periods are determined by when the cloud fractionequals zero. With this, overall annual, annual daytimeand nighttime, and seasonal averages of radiative coolingpotential across CONUS are computed and interpolatedfor mapping.

B. Spatial mapping

An inverse distance weighting (IDW) spatial interpo-lation methodology is used to derive variables (air tem-perature, water vapor partial pressure, ratio of clear skyperiod and cooling potential) in order to form a uni-form latitude-longitude grid. The variable values for non-station locations are interpolated using values from allstations 38

Z =

∑Ni=1 w(di)Zi∑Ni=1 w(di)

, (10)

where N is the number of stations; di is the geographicdistance between a station i and the current location; Zi

is the variable value at a station i and w(di) are weightingfunctions that inversely depend on d:

w(d) =1

dp, (11)

5

Preprint, accepted by Journal of Renewable and Sustainable Energy (2019), (11), 036501. https://doi.org/10.1063/1.5094510

where p is a positive power factor set to unity in thiswork.

After the IDW interpolation, 2-dimensional gradientsat each grid point are calculated according to latitude andlongitude values to determine the degree of similarity toadjacent points. Extreme (top 2%) gradients are inter-preted as resulting from potentially faulty data. Theseoutliers are replaced with averages for adjacent points.

IV. RESULTS AND DISCUSSION

Table II presents the spatial range and mean of an-nually and seasonally averaged ambient temperature Ta,water vapor partial pressure Pw, ratio of clear sky pe-riods Rc and radiative cooling potential qcool,ideal overthe contiguous United States (CONUS). The average ra-diative cooling potential for the entire CONUS area is50.5 W m−2 .

The total area of CONUS can be divided into four cen-sus regions: the West, Midwest, South and Northeast 39.The following discussion uses this common geographicdivisions when referring to specific regions.

A. Annual average resource

Figure 7 is a map of annually averaged Ta, Pw, Rc andqcool,ideal. Temperature decreases with increasing lati-tude. The West is the driest as indicated by Pw, whilethe South is more humid. The Southwest has over 50%clear sky periods throughout the year, while the South-east has less than 30%. Since radiative cooling is mosteffective in dry and cloudless conditions, the Southwesthas an annual average radiative cooling potential of over70 W m−2. The other regions have lower than 50 W m−2

cooling potential, with the South having the lowest. Notethat the cooling potential is the maximum possible cool-ing rate under given meteorological conditions. Absorp-tion of solar irradiance, convective and conductive heatloss all reduce the cooling rate of radiative coolers as theyoperate.

B. Diurnal variations

Figures 8 and 9 show the maps of annually averagedvariables during daytime and nighttime, respectively.The water vapor partial pressure and ratio of clear skyperiods show little diurnal variation, while ambient tem-perature is around 6◦C higher during the day. There-fore, the cooling potential is around 5 W m−2 higher dur-ing the day than the night, which matches the coolingdemands for most residential and commercial buildings.According to the longwave irradiance distribution withinthe atmosphere 16, ambient air is primarily heated by theEarth’s surface, which itself is heated by absorbing so-lar irradiance during the day. If the ground is covered

by passive coolers (objects highly reflective in the solarspectrum), the ambient air is heated to a lesser degree,reducing the ambient temperature during the day. Withthis, the deployment of passive coolers decreases diurnalvariations in temperature.

C. Seasonal variations

Figures 10 to 13 are maps of seasonally averagedvariables in the spring (March-April-May), summer(June-July-August), fall (September-October-November)and winter (December-January-February). During thespring, Arizona and New Mexico experience dry and gen-erally cloudless weather, yielding a high cooling poten-tial of around 80 W m−2. California, Nevada, Utah andColorado have potentials around 60 to 70 W m−2. Theambient temperature for the above mentioned states isaround 10 to 15◦C, such that radiative cooling is ex-pected to fulfill most of the cooling demand from resi-dential sectors. In contrast, the South has an ambienttemperature of around 20◦C and a cooling potential ofaround 40 to 50 W m−2 such that additional methods ofcooling are necessary. During the summer, the West isdry and mostly clear, yielding a cooling potential above70 W m−2. Nevada and Utah have the highest potentialof around 80 W m−2. The South experiences generallyhumid and cloudy summers, with a low cooling poten-tial of around 30 W m−2. Therefore, radiative coolingduring the summer is most promising in the West. Dur-ing the fall, the spatial distribution of radiative coolingpotential is similar to that in the spring, with radiativecooling being effective primarily in the Southwest. Dur-ing the winter, water vapor content in the atmosphere(represented by Pw) is low due to lower air temperature.Consequently, the Southwest has a cooling potential ofaround 60 W m−2, while other regions have potentials ofaround 40 to 50 W m−2. To prevent unnecessary cool-ing and/or the coolers from icing over during the winter,they should be covered.

V. CONCLUSIONS

Radiative coolers are optically designed to reject heatto outer space through the longwave ‘atmospheric win-dow’ spectral band (8 to 13µm). The cooling rates ofradiative coolers are found to be dependent not only ontheir optical properties but also on local meteorologicalconditions. The presence of water vapor and clouds re-duce the ‘transparency’ of the atmospheric window, thusreducing the cooling potential (capability) of the cool-ers. This work presents radiative cooling resource (po-tential) maps for the contiguous United States to assistin the design and deployment of such cooling devices.The cooling potential of a cooler is defined to be its max-imum possible cooling rate, assuming its optical proper-ties are ideal and neglecting convective/conductive heat

6

Preprint, accepted by Journal of Renewable and Sustainable Energy (2019), (11), 036501. https://doi.org/10.1063/1.5094510

TABLE II. Range and mean of annually and seasonally averaged Ta, Pw, Rc and qcool,ideal over the contiguous United States.

Annual Spring(MAM)

Summer(JJA)

Fall(SON)

Winter(DJF)

Annualdaytime

Annualnighttime

Air temperature Ta [◦C]Minimum 4.2 4.3 18.2 5.3 −8.6 8.0 0.4Maximum 23.5 23.2 28.7 24.1 18.7 25.5 21.4Mean 14.3 14.0 23.7 15.0 5.0 17.4 11.2

Water vapor pressure Pw [Pa]Minimum 578.3 555.2 909.1 535.6 300.1 626.0 515.9Maximum 2262.5 1992.5 2889.9 2442.4 1683.5 2342.2 2195.6Mean 1279.4 1143.7 1915.7 1301.9 769.2 1341.2 1213.9

Ratio of clear sky periods Rc

Minimum 0.27 0.24 0.28 0.32 0.21 0.24 0.31Maximum 0.67 0.72 0.77 0.74 0.63 0.65 0.70Mean 0.47 0.44 0.52 0.51 0.42 0.43 0.50

Cooling potential qcool,ideal [W m−2]Minimum 32.2 34.6 19.9 27.8 36.9 31.5 33.0Maximum 80.3 89.7 87.6 83.6 73.9 85.6 75.1Mean 50.5 52.8 48.2 52.1 49.7 52.7 48.3

105

051015202530

(a) T

a [◦ C

]

0

500

1000

1500

2000

2500

3000

(b) P

w [P

a]0.00.10.20.30.40.50.60.70.8

(c) R

c

0102030405060708090

(d) q

cool

,ide

al [W

m2 ]

FIG. 7. Map of annual average values for: (a) air temperature Ta, (b) water vapor partial pressure Pw, (c) ratio of clear skyperiods Rc and (d) radiative cooling potential qcool,ideal.

losses. For real (non-ideal) surfaces, the cooling poweris smaller than the estimates in this work due to non-zero solar absorption, non-blackbody longwave emissivepower and non-zero convective heating. Therefore the es-timates in this work represent upper bounds for passivecooling rates. It is also relevant to note that the radiativecooling resource maps for the contiguous United Statespresented here are generated using extensive meteorolog-ical data collected from 1,681 weather stations over theentire year of 2017. Diurnal and seasonal variations ofcooling potentials are also analyzed below.

Annually, the Southwestern United States is dry and

mostly clear, resulting in an average radiative cooling po-tential of over 70 W m−2, and making it the region whereradiative cooling is most effective. In contrast, the Southhas the lowest cooling potential of around 30 W m−2 dueto due local humid and cloudy meteorological conditions.Convective/conductive heat loss and/or as little as 5%solar irradiance absorption lead to ineffective radiativecooling in the South. Additionally, the water vaporcontent in the atmosphere and clearness of the sky areweaker functions of time of day, as opposed to ambientair temperature, which is always higher during the day.Therefore, cooling potentials are around 5 W m−2 higher

7

Preprint, accepted by Journal of Renewable and Sustainable Energy (2019), (11), 036501. https://doi.org/10.1063/1.5094510

105

051015202530

(a) T

a [◦ C

]

0

500

1000

1500

2000

2500

3000

(b) P

w [P

a]

0.00.10.20.30.40.50.60.70.8

(c) R

c0102030405060708090

(d) q

cool

,ide

al [W

m2 ]

FIG. 8. Annual average values during daytime. From top left, clockwise: (a) air temperature Ta, (b) water vapor partialpressure Pw, (c) ratio of clear sky periods Rc and (d) radiative cooling potential qcool,ideal.

105

051015202530

(a) T

a [◦ C

]

0

500

1000

1500

2000

2500

3000

(b) P

w [P

a]

0.00.10.20.30.40.50.60.70.8

(c) R

c

0102030405060708090

(d) q

cool

,ide

al [W

m2 ]

FIG. 9. Annual average values during nighttime. From top left, clockwise: (a) air temperature Ta, (b) water vapor partialpressure Pw, (c) ratio of clear sky periods Rc and (d) radiative cooling potential qcool,ideal.

during the daytime than during the nighttime.

For the spring (March-April-May) and fall (September-October-November), Arizona and New Mexico exhibitcooling potentials around 80 W m−2, due to dry andcloudless conditions. Surrounding states such as Califor-nia, Nevada, Utah and Colorado have potential rangingfrom 60 to 70 W m−2. Other states exhibit cooling poten-tials about 10 W m−2 lower. During the summer, the en-tire West reaches cooling potential values over 70 W m−2,

again due to high ambient temperatures, low humiditiesand clear skies. The South, in contrast, has a coolingpotential lower than 40 W m−2 due to humid and cloudysummer climate. Most regions of CONUS during the win-ter are generally dry and clear, with cooling potentials ofaround 50 W m−2. To prevent over-cooling and/or icingduring colder weather, cooling devices must be properlystored or covered, which also diminishes their range ofpractical applications.

8

Preprint, accepted by Journal of Renewable and Sustainable Energy (2019), (11), 036501. https://doi.org/10.1063/1.5094510

105

051015202530

(a) T

a [◦ C

]

0

500

1000

1500

2000

2500

3000

(b) P

w [P

a]

0.00.10.20.30.40.50.60.70.8

(c) R

c0102030405060708090

(d) q

cool

,ide

al [W

m2 ]

FIG. 10. From top left, clockwise: (a) air temperature Ta, (b) water vapor partial pressure Pw, (c) ratio of clear sky periodsRc and (d) radiative cooling potential qcool,ideal for the spring months (March-April-May).

105

051015202530

(a) T

a [◦ C

]

0

500

1000

1500

2000

2500

3000

(b) P

w [P

a]

0.00.10.20.30.40.50.60.70.8

(c) R

c

0102030405060708090

(d) q

cool

,ide

al [W

m2 ]

FIG. 11. From top left, clockwise: (a) air temperature Ta, (b) water vapor partial pressure Pw, (c) ratio of clear sky periodsRc and (d) radiative cooling potential qcool,ideal for the summer months (June-July-August).

In summary, passive radiative cooling has much poten-tial in the western region of CONUS, due to the prevalenthot and dry weather in the region. Just as with swamp orevaporative coolers, but due to different physical mech-anisms, the effectiveness of passive radiative cooling de-vices is greatly reduced for hot and humid conditions.Unfortunately, those are the conditions that often de-mand very high power loads for air conditioning.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge partial supportby the U.S. Department of Energy’s Office of EnergyEfficiency and Renewable Energy (EERE) under SolarEnergy Technologies Office (SETO) Agreement NumberEE0008216.

9

Preprint, accepted by Journal of Renewable and Sustainable Energy (2019), (11), 036501. https://doi.org/10.1063/1.5094510

105

051015202530

(a) T

a [◦ C

]

0

500

1000

1500

2000

2500

3000

(b) P

w [P

a]

0.00.10.20.30.40.50.60.70.8

(c) R

c0102030405060708090

(d) q

cool

,ide

al [W

m2 ]

FIG. 12. From top left, clockwise: (a) air temperature Ta, (b) water vapor partial pressure Pw, (c) ratio of clear sky periodsRc and (d) radiative cooling potential qcool,ideal for the fall months (September-October-November).

105

051015202530

(a) T

a [◦ C

]

0

500

1000

1500

2000

2500

3000

(b) P

w [P

a]

0.00.10.20.30.40.50.60.70.8

(c) R

c

0102030405060708090

(d) q

cool

,ide

al [W

m2 ]

FIG. 13. From top left, clockwise: (a) air temperature Ta, (b) water vapor partial pressure Pw, (c) ratio of clear sky periodsRc and (d) radiative cooling potential qcool,ideal for the winter months (December-January-February).

REFERENCES

1S. Vall and A. Castell, “Radiative cooling as low-grade energysource: a literature review,” Renewable and Sustainable EnergyReviews 77, 803–820 (2017).

2U.S. Energy Information Administration, “How much energy isconsumed in U.S. residential and commercial buildings?” https://www.eia.gov/tools/faqs/faq.php?id=86&t=1 (2018).

3A. P. Raman, M. A. Anoma, L. Zhu, E. Rephaeli, and S. Fan,“Passive radiative cooling below ambient air temperature underdirect sunlight,” Nature 515, 540 (2014).

4X. Lu, P. Xu, H. Wang, T. Yang, and J. Hou, “Cooling potentialand applications prospects of passive radiative cooling in build-ings: The current state-of-the-art,” Renewable and SustainableEnergy Reviews 65, 1079–1097 (2016).

5P. Berdahl, M. Martin, and F. Sakkal, “Thermal performanceof radiative cooling panels,” International Journal of Heat andMass Transfer 26, 871–880 (1983).

10

Preprint, accepted by Journal of Renewable and Sustainable Energy (2019), (11), 036501. https://doi.org/10.1063/1.5094510

6Y. Zhai, Y. Ma, S. N. David, D. Zhao, R. Lou, G. Tan, R. Yang,and X. Yin, “Scalable-manufactured randomized glass-polymerhybrid metamaterial for daytime radiative cooling,” Science 355,1062–1066 (2017).

7S. Catalanotti, V. Cuomo, G. Piro, D. Ruggi, V. Silvestrini, andG. Troise, “The radiative cooling of selective surfaces,” Solar En-ergy 17, 83–89 (1975).

8J. André and L. Mahrt, “The nocturnal surface inversion and in-fluence of clear-air radiative cooling,” Journal of the AtmosphericSciences 39, 864–878 (1982).

9U. Eicker and A. Dalibard, “Photovoltaic–thermal collectors fornight radiative cooling of buildings,” Solar Energy 85, 1322–1335(2011).

10M. M. Hossain and M. Gu, “Radiative cooling: principles,progress, and potentials,” Advanced Science 3, 1500360 (2016).

11E. Rephaeli, A. Raman, and S. Fan, “Ultrabroadband photonicstructures to achieve high-performance daytime radiative cool-ing,” Nano Letters 13, 1457–1461 (2013).

12Z. Huang and X. Ruan, “Nanoparticle embedded double-layercoating for daytime radiative cooling,” International Journal ofHeat and Mass Transfer 104, 890–896 (2017).

13M. Zeyghami, D. Y. Goswami, and E. Stefanakos, “A reviewof clear sky radiative cooling developments and applications inrenewable power systems and passive building cooling,” SolarEnergy Materials and Solar Cells 178, 115–128 (2018).

14A. Harrison, “Effect of atmospheric humidity on radiation cool-ing,” Solar Energy 26, 243–247 (1981).

15C. Tso, K. Chan, and C. Y. Chao, “A field investigation of passiveradiative cooling under hong kong’s climate,” Renewable Energy106, 52–61 (2017).

16M. Li and C. F. M. Coimbra, “On the effective spectral emis-sivity of clear skies and the radiative cooling potential of se-lectively designed materials,” International Journal of Heat andMass Transfer 135, 1053 – 1062 (2019).

17M. A. Atwater and J. T. Ball, “Computation of ir sky temper-ature and comparison with surface temperature,” Solar Energy21, 211–216 (1978).

18R. Exell, “The atmospheric radiation climate of thailand,” SolarEnergy 21, 73–79 (1978).

19D. Pissimanis and V. Notaridou, “The atmospheric radiation inathens during the summer,” Solar Energy 26, 525–528 (1981).

20M. Martin and P. Berdahl, “Characteristics of infrared sky radi-ation in the United States,” Solar energy 33, 321–336 (1984).

21P. Schmetz, J. Schmetz, and E. Raschke, “Estimation of daytimedownward longwave radiation at the surface from satellite andgrid point data,” Theoretical and Applied Climatology 37, 136–149 (1986).

22A. Argiriou, M. Santamouris, and D. Assimakopoulos, “Assess-ment of the radiative cooling potential of a collector using hourlyweather data,” Energy 19, 879–888 (1994).

23M. Hanif, T. Mahlia, A. Zare, T. Saksahdan, and H. Metselaar,“Potential energy savings by radiative cooling system for a build-

ing in tropical climate,” Renewable and Sustainable Energy Re-views 32, 642–650 (2014).

24K. Zhang, D. Zhao, X. Yin, R. Yang, and G. Tan, “Energy savingand economic analysis of a new hybrid radiative cooling systemfor single-family houses in the usa,” Applied Energy 224, 371–381 (2018).

25M. Li, Z. Liao, and C. F. M. Coimbra, “Spectral model for clearsky atmospheric longwave radiation,” Journal of QuantitativeSpectroscopy and Radiative Transfer 209, 196–211 (2018).

26M. Li, Z. Liao, and C. F. M. Coimbra, “Monte Carlo method forspectral solar irradiance on inclined surfaces,” Journal of Renew-able and Sustainable Energy In preparation (2019).

27J. A. Duffie and W. A. Beckman, Solar Engineering of ThermalProcesses (John Wiley & Sons, 2013).

28G. M. Hale and M. R. Querry, “Optical constants of water inthe 200-nm to 200-µm wavelength region,” Applied Optics 12,555–563 (1973).

29C. F. Bohren and D. R. Huffman, Absorption and Scattering ofLight by Small Particles (John Wiley & Sons, 2008).

30J. E. Hansen and L. D. Travis, “Light scattering in planetaryatmospheres,” Space Science Reviews 16, 527–610 (1974).

31N. L. Miles, J. Verlinde, and E. E. Clothiaux, “Cloud dropletsize distributions in low-level stratiform clouds,” Journal of theAtmospheric Sciences 57, 295–311 (2000).

32H. Barker, G. Stephens, P. Partain, J. Bergman, B. Bonnel,K. Campana, E. Clothiaux, S. Clough, S. Cusack, J. Delamere,et al., “Assessing 1d atmospheric solar radiative transfer mod-els: Interpretation and handling of unresolved clouds,” Journalof Climate 16, 2676–2699 (2003).

33A. Walther, W. Straka, and A. Heidinger, “ABI algorithm theo-retical basis document for daytime cloud optical and microphys-ical properties (DCOMP), version 2.0. NOAA/NESDIS,” Centerfor Satellite Applications and Research (2011).

34M. Lee, “Scaling anisotropic scattering in radiation heat transferfor a planar medium,” Urbana 2, 61801 (1982).

35M. Li, Y. Jiang, and C. F. M. Coimbra, “On the determination ofatmospheric longwave irradiance under all-sky conditions,” SolarEnergy 144, 40–48 (2017).

36National Oceanic and Atmospheric Adminis-tration, “Automated surface observing system(ASOS),” https://www.ncdc.noaa.gov/data-access/land-based-station-data/land-based-datasets/automated-surface-observing-system-asos.

37O. A. Alduchov and R. E. Eskridge, “Improved Magnus formapproximation of saturation vapor pressure,” Journal of AppliedMeteorology 35, 601–609 (1996).

38M. Tomczak, “Spatial interpolation and its uncertainty usingautomated anisotropic inverse distance weighting (idw)-cross-validation/jackknife approach,” Journal of Geographic Informa-tion and Decision Analysis 2, 18–30 (1998).

39United States Census Bureau, “Geographic terms and concepts -census divisions and census regions,” https://www.census.gov/geo/reference/gtc/gtc_census_divreg.html.

11