Page 229 - Modern Spatiotemporal Geostatistics

210 Modern Spatiotemporal Geostatistics — Chapter 10

Table 10.2. The effect of soft PMio data at a set of estimation
points on August 31, 1995.
Estimation error

Monitoring (a) Without soft data at (b) With soft data at
station no. the estimation points the estimation points
40 0.75 0.12
44 0.73 0.13
46 1.33 0.22
39 4.95 0.67
1 3.20 0.54

h

3
Figure 10.6. Map of the Ae values (in /ng/m ) over North Carolina averaged
during a three-day period. This map demonstrates the improvement
gained by considering soft data at the estimation points.

data into the mapping process. An assessment of this work is presented in the
following example.

EXAMPLE 10.2: In Figure 10.7 we show the numerical work (CPU time) re-
quired by the BME technique on an HP-9000 computer for a typical case
(SANLIB99, 1999). The CPU time is plotted as a function of the number of
soft (interval) data points (for 2, 8, and 32 hard data points). It is evident
from Figure 10.7 that the CPU time remains small (e.g., less than 0.15 sec
for up to about 7 soft data points), which makes BME a numerically efficient
method for spatiotemporal analysis. For larger numbers of soft data points one
should use a Monte Carlo method.