38, 2012
d sensitivity
plication to
ndicated the
| tools and
he effect of
MAUP, this
tem. In the
orm taluks.
the optimal
illages. The
t al. 2011b)
The results
districts of
models for
density
per square
households
| van
households
District
metrics at
, percentage
rcentage of
nent census
lectricity as
illage level.
icts of Pune
erates were
as such as
nd Dehu in
r, Baramati,
These areas
literates per
listrict have
e kilometre.
holds in the
ans. Around
cent of the
ans. Similar
nent census
0% to more
ie district as
ent houses.
ed to have
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B8, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
access to electricity near the urban areas while the district as
a whole exhibited 65% to 70% of rural electrification.
Due to the absence of these metrics at the village level from
the census, it was not possible to validate the results obtained
from the predictions. The results presented in this chapter are
the best estimates of the metrics previously unavailable.
4.1 Comparison of maps from all spatial scales
Census metrics were proposed and predicted over different
spatial scales in this study. Maps of number of households
per square kilometre in all the spatial scales are shown in
figure 3. From the map at the district level (figure 3 (a)) it
was observed that in Pune, as a whole, there were 40 to 50
households per square kilometre. A look at smaller
administrative areas such as taluks and villages, revealed
detailed distribution of the households. The map of number
of households per square kilometre at the taluk level is shown
in figure 3 (b). It was observed that taluks in the north
western, eastern and southern part of Pune district exhibited
the lowest density values of around 20 to 50 households per
square kilometre. These taluks comprised of an area of
around 7000 square kilometre in the district. Five taluks
demonstrated to have moderate density of 80 to 200
households per square kilometre. These taluks occupied an
area of approximately 6000 square kilometre in Pune. Only
1400 square kilometre of the district had high values of 300
to 500 households per square kilometre. These taluks were
located near the centre of the district around the urban areas
of Pune. Since most of the taluks exhibited to have low to
moderate household density, the aggregated district map
showed a low value for the district as a whole.
Number of households per squors hitometre
Figure 3: The effect of scale in the prediction of census metrics
at (a) Districts; (b) Taluks; (c) Villages and (d) one square
kilometre areas.
The taluks were further divided into villages. The map of the
number of households per square kilometre in the villages is
shown in figure 7.5 (c). It was observed that the villages in
the central part of the district around cities exhibited highest
household densities of more than 100 households per square
kilometre. Moderate distribution (50 to 100 households per
square kilometre) was noted in the north eastern and southern
part while the major part of the area had around 15 to 30
households per square kilometre. Therefore it was apparent
from these maps that values of metrics aggregated over small
areas influenced the data of large regions. The analyses of the
results in all the spatial scales helped overcome the
individualistic and ecological fallacies.
5. CONCLUSION
This paper looked into the application of the models for
proposing census metrics otherwise not collected by the
census at the village level for the district of Pune in
Maharashtra. The errors arising from analyses of multi —
scale data such as MAUP and ecological fallacy were
examined and the approach of optimal zoning system was
used to overcome the MAUP effects in predicting the metrics
for the villages of Pune. However, due to the absence of these
data from the census, it was not possible to validate the
results. This chapter showed the potential for models derived
from DMSP-OLS images for mapping and predicting census
metrics for small regional scales. As a result, it is possible to
map the metrics showing levels of development using night
time satellite images collected by DMSP-OLS.
6. REFERNCES
Bhandari, L & Roychowdhury, K 2012, "Night Lights and
Economic Activity in India: À study using DMSP-OLS night
time images', paper presented to Asia Pacific Advanced
Network (APAN), 2011, Delhi.
Blalock, HM 1964, Causal inferences in nonexperimental
research, University of North Carolina Press.
Cao, C & Lam, NSN 1997, 'Understanding the scale and
resolution effects in remote sensing and GIS', in Scale in
Remote Sensing and GIS, Boca Raton, FL: CRC Lewis, pp.
57-72.
Clark, WAV & Avery, KL 1976, "The effects of data
aggregation in statistical analysis', Geographical Analysis,
vol. 8, no. 4, pp. 428-38.
Croft, TA 1978, 'Nighttime images of the earth from space’,
Scientific American, vol. 239, pp. 86-96.
Dark, SJ & Bram, D 2007, "The modifiable areal unit
problem (MAUP) in physical geography', Progress in
Physical Geography, vol. 31, no. 5, pp. 471-9.
Doll, CNH 2008, A CIESIN Thematic Guide to Night-time
Light Remote Sensing and its Applications, Center for
International Earth Science Information Network of
Columbia University, Palisades, NY,
«http://sedac.ciesin.columbia.edu/tg/guide main jsp».
Doll, CNH, Morley, JG & Muller, JP 2004, 'Geographic
Information Issues associated with Socio-Economic
Modelling from Night-time light Remote Sensing Data’,
paper presented to ISPRS Conference, Istanbul,
<http://Www.isprs.org/istanbul2004/comm7/papers/155.pdf>.
Doll, CNH, Muller, J & Morley, J 2006, 'Mapping regional
economic activity from night-time light satellite imagery',
Ecol Econ, vol. 57, pp. 75 - 92.
Doll, CNH, Muller, JP & Elvidge, CD 2000, 'Night-time
Imagery as a Tool for Global Mapping of Socioeconomic
Parameters and Greenhouse Gas Emissions', AMBIO: A
