Environmental Engineering Reference
In-Depth Information
Fortunately, the impact of this change was not as severe as that which occurred with
the deployment of ASOS. Since there has been no change to the monitoring heights,
locations, or data collection practices, adjustment factors can be applied to compensate
for the effects of the anemometer change at most ASOS stations. However, there
is additional uncertainty associated with this adjustment, which must be considered
before relying on the pre-IFW data from these stations.
Similar changes in measurement equipment, tower height, and location have
occurred in most countries. Unfortunately, these changes are often not well docu-
mented, which leaves the analyst to investigate the history of each reference station —
a potentially time-consuming task.
Another challenge is changing site conditions around reference stations, which can
create trends in observed wind speeds that have nothing to do with the general wind
climate. Figure 12-4 shows an extreme case: a downward trend in observed speeds
starting around the middle of the twentieth century at the Blue Hill Observatory in
Massachusetts, USA. This trend, which does not appear in other weather records in
the region, is at least partially attributable to the regrowth of previously cleared forest
around Great Blue Hill, as suggested by the photographs.
In the absence of significant trends or discontinuities, the uncertainty in the long-
term mean wind speed derived through MCP should decrease as the length of the
reference station's record increases. This is evident in Equation 12.1: the longer the
reference data period, N R , the better. In most real-world situations, however, the benefit
of going beyond about 10 - 15 years of reference data is limited. Figure 12-5 shows the
uncertainty for a range of values of r 2 (from 0.45 to 0.95) and N R (from 1 to 30 years)
based on the same equation and assuming N T =
4%. The two dashed
curves mark the points where 80% (left-hand curve) and 90% (right-hand curve) of
the maximum possible reduction in uncertainty is achieved. For all reasonable values
of r 2 , 80% of the maximum benefit is reached with less than 10 years of reference
data. For r 2
1 and
σ R = σ T =
.
85, 90% of the benefit is reached with less than 17 years of data.
The presence of trends or discontinuities in the reference data, whether artifacts of
changing site conditions or measurement techniques or real manifestations of climate
change, can have a pernicious effect on the accuracy of MCP. Suppose there is a
linear trend in the reference wind speed. If the trend does not reflect a real change
in the wind climate (perhaps trees are growing around the station or the anemometer
has been slowing down because of wear in the bearings), then the adjusted long-term
mean wind speed will tend to be biased by an amount
0
that depends on the slope of
the trend line and the length of the reference data record:
ε
N
2 s
ε ≈−
(12.3)
where s is the trend slope in percent per year and N is the number of years in the
reference data record. (In this equation, 1 year of overlapping on-site and reference
data and perfect correlation between them are assumed.) Thus, where false trends are
Search WWH ::




Custom Search