Spectral identification of lighting type and character - PubMed Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2010;10(4):3961-88.
doi: 10.3390/s100403961. Epub 2010 Apr 20.

Spectral identification of lighting type and character

Christopher D Elvidge  1 , David M Keith, Benjamin T Tuttle, Kimberly E Baugh
Affiliations

Spectral identification of lighting type and character

Christopher D Elvidge et al. Sensors (Basel). 2010.

Abstract

We investigated the optimal spectral bands for the identification of lighting types and the estimation of four major indices used to measure the efficiency or character of lighting. To accomplish these objectives we collected high-resolution emission spectra (350 to 2,500 nm) for forty-three different lamps, encompassing nine of the major types of lamps used worldwide. The narrow band emission spectra were used to simulate radiances in eight spectral bands including the human eye photoreceptor bands (photopic, scotopic, and "meltopic") plus five spectral bands in the visible and near-infrared modeled on bands flown on the Landsat Thematic Mapper (TM). The high-resolution continuous spectra are superior to the broad band combinations for the identification of lighting type and are the standard for calculation of Luminous Efficacy of Radiation (LER), Correlated Color Temperature (CCT) and Color Rendering Index (CRI). Given the high cost that would be associated with building and flying a hyperspectral sensor with detection limits low enough to observe nighttime lights we conclude that it would be more feasible to fly an instrument with a limited number of broad spectral bands in the visible to near infrared. The best set of broad spectral bands among those tested is blue, green, red and NIR bands modeled on the band set flown on the Landsat Thematic Mapper. This set provides low errors on the identification of lighting types and reasonable estimates of LER and CCT when compared to the other broad band set tested. None of the broad band sets tested could make reasonable estimates of Luminous Efficacy (LE) or CRI. The photopic band proved useful for the estimation of LER. However, the three photoreceptor bands performed poorly in the identification of lighting types when compared to the bands modeled on the Landsat Thematic Mapper. Our conclusion is that it is feasible to identify lighting type and make reasonable estimates of LER and CCT using four or more spectral bands with minimal spectral overlap spanning the 0.4 to 1.0 um region.

Keywords: LED; Nightsat; lighting efficiency; lighting types; nighttime lights; photopic band.

PubMed Disclaimer

Figures

Figure 1.
Figure 1.
Shanghai, China at night.
Figure 2.
Figure 2.
Color camera images of Chicago, Tokyo and Hong Kong acquired at sixty meter resolution from the International Space Station. Note the presence of optically thin clouds in the Chicago image, resulting in a blur of the street pattern.
Figure 3.
Figure 3.
Spectral response functions of the three human photoreceptor bands.
Figure 4.
Figure 4.
Set of four spectral bands based on the Landsat Thematic Mapper (TM). Note that the short wavelength edge of the near infrared TM band (NIR) was shifted to 700 nm. An additional orange spectral band was formed by merging the green (TM 2) and red (TM 3) bands.
Figure 5.
Figure 5.
Emission spectra of lamps burning liquid fuels largely exhibit blackbody curves.
Figure 5.
Figure 5.
Emission spectra of lamps burning liquid fuels largely exhibit blackbody curves.
Figure 6.
Figure 6.
Emission spectra of lanterns burning pressurized fuel.
Figure 7.
Figure 7.
Emission spectra of incandescent lamps.
Figure 8.
Figure 8.
Emission spectra of quartz halogen lamps.
Figure 9.
Figure 9.
Emission spectra of a standard fluorescent tube and an compact fluorescent light (CFL).
Figure 10.
Figure 10.
Spectral variability of fluorescent lamps. Each of the nine spectra were normalized to 1.0 and then an average (blue) and standard deviation (red) were calculated.
Figure 11.
Figure 11.
Emission spectrum of a mercury vapor lamp.
Figure 12.
Figure 12.
Emission spectra of four metal halide lamps.
Figure 12.
Figure 12.
Emission spectra of four metal halide lamps.
Figure 13.
Figure 13.
Spectral variability of metal halide lamps. Each of the four spectra were normalized to 1.0 and then an average (blue) and standard deviation (red) were calculated.
Figure 14.
Figure 14.
Emission spectrum of a high pressure sodium lamp.
Figure 15.
Figure 15.
Spectral variability of metal halide lamps. Each of the three spectra were normalized to 1.0 and then an average (blue) and standard deviation (red) were calculated.
Figure 16.
Figure 16.
Emission spectrum of a low pressure sodium lamp.
Figure 17.
Figure 17.
Spectra of LED lamps. A) Comparison of two white LED streetlights. B) Spectra from four colored LEDs.
Figure 18.
Figure 18.
Error rates for the identification of nine types of lights.
Figure 19.
Figure 19.
Error rates for estimation of luminous efficacy.
Figure 20.
Figure 20.
LE estimated from the 1ON band combination versus the LE calculated from manufacturer reported lumen output and consumption watts.
Figure 21.
Figure 21.
RMSE for LER estimates.
Figure 22.
Figure 22.
LER versus estimated LER from TM1, photopic, TM3 and NIR bands.
Figure 23.
Figure 23.
RMSE values for CCT estimation.
Figure 24.
Figure 24.
The 1ON estimates of CCT’ versus the CCT calculated from the ASD spectra. Linear regression yields an R2 of 0.86.
Figure 25.
Figure 25.
RMSE for CRI estimation. The lowest RMSE came from the TM band 1,2,3 combination.
Figure 26.
Figure 26.
CRI versus estimated CRI’ using TM bands 1, 2 and 3.
Figure 27.
Figure 27.
Mean and standard deviation for luminous efficacies calculated for a wide range of lamps derived from manufacturer data. INC = incandescant, QH = quartz halogen, XE = xenon arc, MV = mercury vapor, LED = light emitting diode, FLU = linear fluorescent, CFL = compact fluorescent, MH = metal halide, HPS = high pressure sodium.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. International Energy Agency . Light’s Labour Lost. Paris, France: 2006.
    1. Mills E. The specter of fuel based lighting. Science. 2005;308:1263–1264. - PubMed
    1. Rich C., Longcore T. Ecological Consequences of Artificial Night Lighting. Island Press; Washington, DC, USA: 2005.
    1. Schubert E.F., Kim J.K. Solid state light sources getting smarter. Science. 2005;308:1274–1278. - PubMed
    1. Croft T.A. Night-time images of the earth from space. Sci. Amer. 1978;239:68–79.

LinkOut - more resources