Geoscience Reference
In-Depth Information
provided daily extreme temperature statistics that were more accurate than
the LARS-WG.
3.2. Model validation
Results of the SDSM and LARS-WG for the 1976-1990 validation period
at the four selected sites indicated that neither of these models was able to
generate accurately the statistics of daily precipitation. For instance, bias
for the mean of daily precipitation amount (SDII) ranges from
0 . 39 to
+1 . 94 mm for SDSM, and
0 . 32 to +0 . 44 mm for LARS-WG; and for the
90th percentile of daily precipitation ( Prec90p )from
2 . 4to+3 . 12 mm for
SDSM, and from
1 . 93 to +3 . 37 mm for LARS-WG. Table 5 shows the
total bias scores of the two models in the simulation of daily precipita-
tion process. It can be seen that LARS-WG can reproduce more accurate
observed statistics of daily precipitation than the SDSM.
Regarding the simulation of daily temperature extremes, both SDSM
and LARS-WG could provide an accurate description of many daily maxi-
mum and minimum temperature characteristics. For instance, the bias val-
ues for the extreme statistic Tmax90p are from
0 . 9 Cto+0 . 82 Cfor
1 . 55 Cto+1 . 03 C for LARS-WG. Similarly, for the
extreme index Tmin10p the bias values are from
SDSM, and from
0 . 18 Cto+0 . 98 Cfor
0 . 46 Cto+0 . 48 C for LARS-WG. The small bias values
indicate the accuracy of both models in the simulation of the extremes of
daily maximum and minimum temperatures. Similar results were found for
other evaluation statistics. Furthermore, Table 6 shows the total bias scores
for SDSM and LARS-WG. It can be seen that both models could provide
a comparable performance in the simulation of extreme daily temperature
processes.
SDSM and from
Table 5. Total bias scores for daily precipitation
simulation for 1976-1990 validation period.
Evaluation indices
SDSM
LARS-WG
SDII
14
1
Prec90p
9
6
Prcp1
14
2
CDD
6
4
R3days
14
2
Prec-moy
40
8
Prec-std
38
9
Total
135
32
Search WWH ::




Custom Search