Long-Term Rainfall Information Forecast by Utilizing Constrained Amount of Observation through Artificial Neural Network Approach

Table 1

ANN parameters and configuration for estimating missing rainfall records.

Parameter	Configuration
Configuration of network; Transfer function; Training algorithm	Input (1); hidden (2) and hidden (2); output (1); The linear transfer function (purelin) at hidden 1 and 2, and output layers; Levenberg–Marquardt algorithm (trainlm)
Learning rate	0.05
Ratio to increase the learning rate	1.05
Target performance in terms of MSE	0.02
The number of optimum epochs to train	10
Performance
MSE	0.053 mm^2d^−2
R	Training: 0.87; test: 0.93; validation: 0.92