Geography Reference
In-Depth Information
zones T1 and S1 is 56.84% of the zone S1 population of 532, thus ( 532 / 100 ) ¥ 56.84 = 302
(when rounded to a whole number). Once the populations of each of the areas of
intersection have been obtained they can be summed within the target zones, as shown
in the bottom part of Figure 9.14.
More sophisticated approaches to areal interpolation exist (see Lloyd (2006) for a
summary). h e main focus in this chapter is on the generation of surfaces rather than
zones, and so of more immediate relevance here are approaches such as the pycnophy-
lactic reallocation method of Tobler (1979) or the population surface modelling pro-
cedure of Martin (1989). Both of these approaches allow the transfer of zonal counts
to regular grids. One benei t of such approaches is to enable direct comparison of
values for dif erent time periods even when the original zonal systems used at dif erent
periods are quite dif erent.
Case studies
9.10
h e following two case studies are based on the data introduced in Section 8.7. h ese
case studies demonstrate (1) estimation of the variogram and (2) spatial interpolation
using IDW, TPS, and OK.
9.10.1 Variogram estimation
Figure 9.15 shows an experimental variogram computed from precipitation data.
Recall that the variogram tells us how dif erent observations tend to be a function of
how far apart the observations are. In this case, the semivariance values increase with
increased distance and they level out at a distance of perhaps 75 km. A model can be
i tted to the variogram, as detailed in Section 9.7.1, and the model used to inform
prediction of precipitation amount at locations where there are no measurements
16000
14000
12000
10000
8000
6000
4000
2000
Semivariance
0
0
20000
40000
60000
80000
100000
Lag (m)
Figure 9.15 Experimental variogram computed from precipitation data for
8 May 1986 in Switzerland.
 
Search WWH ::




Custom Search