RMSE
CO CO
n
k
n
=
′ −
(
)
=
∑
2
2
2
1
(4)
where
n
is the total number of pixels in each of the difference
maps (Plate 1a, b, and c), CO
′
2
is the amount of CO
2
emissions
of each pixel estimated by the nighttime light image data,
and CO
2
is the amount of CO
2
emissions of each pixel in the
Vulcan CO
2
emissions map.
To test whether correlations between the
DN
values of
stable lights imagery and the amount of CO
2
emissions vary
across different geographic scales, we established the correla-
tions again at the state scale (Figures 4, 5, and 6). Three states
(California, Texas, and Florida) were selected because of their
relatively large lit areas in the US. A large lit area provides
the sufficient number of pixels involved in statistical analyses
and consequently leads to relatively small adverse impacts
from random errors
Figure 5. Correlations between DN values of nighttime light imagery and amount of CO
2
emissions for Texas. The DN value of 63 was
excluded from the regression in Figure 5c due to its incompatibility with the other DN values in stable lights image products.
Figure 6. Correlations between DN values of nighttime light imagery and amount of CO
2
emissions for Florida. The DN value of 63 was
excluded from the regression in Figure 6c due to its incompatibility with the other DN values in stable lights image products.
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
December 2015
939
