Geology Reference
In-Depth Information
Fig. 7.13 Daily variations of ANN modeled reference evapotranspiration values in comparison to
FAO56 values in Santa Monica Station during the training phase
temperature (T), relative humidity (RH), and wind speed. In this attempt, we did not
apply any input selection procedures.
Instead, data from the
rst 3 years
(1995
1998) were used to train the ANN model. In the case of Santa Monica
Station, inputs such as precipitation, maximum and minimum temperature (
-
C),
wind speed (m/s), and daily maximum and minimum relative humidity (%) were
used as inputs. The optimum hidden node numbers were found to be 10 and 9 for
the MLP neural networks in the Beas and Santa Monica Stations. The study
demonstrated that modeling of daily ET
0
is possible through the use of the ANN
models using the above-mentioned inputs in both the Beas and Santa Monica
Stations. In the case of Beas, data from the first 3 years (1995
-
1997) were used for
training the model and from the next 2 years (1998
°
1999) for validation. On the
other hand, in the case of the Santa Monica Station, we have used 3 years each for
training (2000
-
2005) purposes.
The ANN estimated ET0 values for Beas (year 2005) and Santa Monica Stations
(2000
2002) and validation (2003
-
-
2002) are shown in Table
7.4
, along with estimations from empirical ET
0
equations. The test statistics during training and validation as a whole are not shown
here. The ET
0
estimates of the ANN model during the training and validation
phases in Santa Monica Station are shown in Figs.
7.13
and
7.14
in the form of line
plots. Table
7.4
indicates that ANN modeling indices in both Beas and Santa
Monica are closer to FAO-56 PM values than CIMIS-PM and COPAIS but lower
than the ASCE-PM method.
-
7.4.2.3 Monthly and Yearly ET
0
Comparison
One of the major issues associated with HYREX data from the Brue catchment is its
lack of lysimeter information to check the reliability of the presented PM models
and Copais Approach. In Fig.
7.15
the variation of monthly
figures (summed daily
