This paper shows two notes on the representation methods of spatial data. Firstly, with the use of remote sensing technique and data, it proposes one representation of cities in space. Population density grid data shows the distribution of human residents. Focusing on the densely populated regions, it shows the continuous shape of the cities. The shape of the cities can be compared inter temporally across space at different population size. Some examples are shown with Indian data. Secondly, for the representation of geographical units, c.f. administrative area, it is pointed out some measurement errors stemming out from the construction of data. Specifically, centrality of geographical units such as polygons of districts would have certain deviations from the true centre of the districts. With Indian districts data, taking the true centre at the location of district headquarters, we examine the distribution of the measurement errors and correlations with other variables. Results show that there are measurement errors and they correlate with size of the districts.