Geology Reference
In-Depth Information
7.5 Data-Based Evaporation Modeling: Data Selection
Approaches
Because of a lack of continuously measured evaporation/evapotranspiration data at
the Brue catchment and Santa Monica region, we have performed the analysis using
the data from the Chahnimeh reservoirs region of Iran. The data set consisted of
11 years (1995
2006) of daily records of air temperature (T), wind speed (W),
saturation vapor pressure de
-
cit (Ed), relative humidity (RH), and pan evaporation
(E). In this modeling,
first we apply the data selection approaches described in the
Chap.
3
, including GT, entropy theory, AIC, and BIC. Later we perform data
modeling of pan evaporation using different data models.
The following sections give the detailed description on how the novel approaches
such as the Gamma Test and entropy theory are used in evaporation modeling to
select effective parameters. The study also deals with the use of other approaches such
as AIC and BIC in input quality checks and input length selection for modeling.
7.5.1 Gamma Test for Input Selection in Evaporation
Modeling
The detailed decryption of the GT and its general applications is explained in
Chap.
3
. In this study, different combinations of input data were explored to assess
their in
uence on evaporation modeling. There were 2
n
1 meaningful combi-
nations of inputs (in this case n = 4 as the total available inputs are daily records of
air temperature (T), wind speed (W), saturation vapor pressure de
−
cit (Ed), and
relative humidity (RH)). Out of this 2
n
1, the best one can be determined by
observing the Gamma value, which indicates a measure of the best MSE attainable
using any modeling methods for unseen input data. Thus, we performed M-tests in
different dimensions, varying the number of inputs to the model. The analysis
results are shown in Table
7.5
, which clearly presents the response of the data
model to different combination of inputs data sets. In Table
7.5
, the minimum value
−
Table 7.5 Gamma test results on evaporation modeling data [
21
]
Parameters
Different combinations
W, T, RH, Ed
T, RH, Ed
W, RH, Ed
W, T, Ed
W, T, RH
Gamma (
ʓ
)
0.02184
0.05295
0.02160
0.02175
0.02207
Gradient (A)
0.05946
0.30649
0.27377
0.09502
0.06902
−
Standard error
0.00049
0.00145
0.00058
0.00025
0.00046
V-ratio
0.08736
0.21182
0.08641
0.08703
0.08830
Near neighbors
10
10
10
10
10
M
4019
4019
4019
4019
4019
Mask
1111
0111
1011
1101
1110
