Prediction of daily rainfall using soft computing techniques
Indian Agricultural Research Journals
View Archive InfoField | Value | |
Title |
Prediction of daily rainfall using soft computing techniques
|
|
Creator |
KUMAR, AMIT
SARANGI, ARJAMADUTTA SINGH, D.K. |
|
Subject |
Agriculture
Rainfall Neural networks Prediction ANN model Flood control |
|
Description |
Accurate rainfall forecasting empowers precision irrigation, optimizing crop productivity and resource management for sustainable agriculture. In the present study, the performance of soft computing techniques viz. LSTM and ANN were evaluated for daily rainfall prediction using 5 days preceding meteorological data. Besides this, the effect of static time-series days of the year (1 to 365/366) as input variables were investigated on performance of neural network techniques. Both models were trained, tested and validated with eight metrological variables using 37 years of data besides static time-series as input variables. Data were partitioned into three groups as training (60%), testing (20%), and validation (20%), respectively. It was observed that the LSTM model with only meteorological data performed poorly in predicting rainfall, while, the incorporation of static time series only marginally improved the LSTM model’s performance for peak rainfall prediction, but it overestimated lower values of rainfall. Besides this, the ANN model using only meteorological data and meteorological data with days of the year were excellent performance in prediction both peak and lower values of rainfall. Performance error matrices MAE, MSE and RMSE were varied from 3.07 to 4.75, 4.31 to 142.78 and 6.66 to 11.95 for LSTMs, and 0.11 to 0.15, 0.14 to 0.49 and 0.38 to 0.70 for ANNs, respectively. Moreover, NSE, KGE and PBias were varied from -0.21 to -4.99, -0.81 to 0.31 and 11.93 to 46.16 for LSTMs and 0.99, 0.93 to 0.98 and -5.79 to 3.02 for ANNs, respectively. This indicated that the ANN models outperformed than the LSTM models for daily rainfall prediction. The ANN model show promise as an effective tool for rainfall prediction using meteorological data.
|
|
Publisher |
Journal of Soil and Water Conservation
|
|
Date |
2024-08-23
|
|
Type |
info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion Peer-reviewed Article |
|
Identifier |
https://epubs.icar.org.in/index.php/JSWC/article/view/155476
|
|
Source |
Journal of Soil and Water Conservation; Vol. 23 No. 1 (2024)
2455-7145 0022-457X |
|
Rights |
Copyright (c) 2024 Soil Conservation Society of India, New Delhi
|
|