Geology Reference
In-Depth Information
This chapter describes two different realms in evaporation modeling: [
1
] refer-
ence evapotranspiration (ET
0
) with traditional models from the meteorological data
and [
2
] evaporation modeling with the data-based modeling concepts. In Sect.
7.1
we describe the mathematical details of the models used. Analyses with traditional
reference evapotranspiration models are performed on the data from the Brue
catchment, UK and the Santa Monica Station, USA, and compared with ANN
models developed. In Sect.
7.2
, studies conducted to see how data selection
approaches are responding to the evaporation data from the Chahnimeh reservoirs
region, Iran are described.
7.3 Statistical Indices for Model Comparison
The performances of the developed data-based models and the basic benchmark
models in this chapter have been compared using various standard statistical per-
formance evaluation criteria. The statistical measures considered include the cor-
relation coef
ciency
(E), root mean squared error (RMSE), mean bias error (MBE), and variance of the
distribution of differences with MBE (S
2
). These statistical terms can be de
cient (CORR), mean absolute percentage error (MAPE), ef
ned as
follows:
P
i
¼
1
ð
Q
o
Q
o
Þ
P
i
¼
1
ð
Q
m
Q
m
Þ
:
ð
Correlation Coefficient
CORR
¼
q
ð
7
:
1
Þ
;
2
2
Q
m
Q
m
Þ
Q
o
Q
o
Þ
:
ð
N
X
N
1
ð
Q
m
Q
o
Þ
Mean Absolute Percentage Error, MAPE
¼
100
ð
7
:
2
Þ
Q
o
i
¼
1
E
1
E
2
Efficiency
;
E
¼
ð
7
:
3
Þ
E
1
where
X
N
2
Q
o
Q
o
E
1
¼
ð
Þ
ð
7
:
4
Þ
i
¼
1
and
X
N
2
E
2
¼
ð
Q
0
Q
m
Þ
ð
7
:
5
Þ
i
¼
1
