Geology Reference
In-Depth Information
Table 5.1 Comparison of some basic performance indices of LLR and ANN models in daily solar
radiation estimation
Models and algorithms
Training data (770 Data points)
Validation data
R
2
R
2
RMSE
(W/m
2
and %)
Slope
MBE
W/m
2
RMSE
(W/m
2
and %)
Slope
MBE
W/m
2
LLR
20.72
(19.1)
0.93
0.96
−
0.799
(22.4)
0.90
0.97
0.197
ANN (Conjugate
Gradient)
30.093
(28.5)
35.417
(28.1)
0.88
0.93
−
2.072
0.83
0.93
−
0.162
ANN (BFGS)
32.755
(30.2)
39.236
(29.7)
0.87
0.94
2.060
0.80
0.93
0.255
−
−
ANN
(Levenberg
-
Marquardt)
29.83
(27.6)
30.39
(26.9)
0.89
0.94
−
1.178
0.85
0.95
−
0.222
Fig. 5.22 Solar radiation as observed and estimated using the LLR model for the training data set.
a scatter plot, b line diagram
Fig. 5.23 Solar radiation as observed and estimated using the LLR model for the validation data
set. a scatter plot, b line diagram
model for 770 data points resulted in the minimum overall RMSE value of
20.72 W/m
2
, which is 19.1 %; compare to observed daily solar radiation and the
mean bias error (MBE) observed as
0.799 W/m
2
.
−