NORMA eResearch @NCI Library

A Neural Network Modelling for Soil Moisture Prediction

Patil, Shubham (2019) A Neural Network Modelling for Soil Moisture Prediction. Masters thesis, Dublin, National College of Ireland.

[img]
Preview
PDF (Master of Science)
Download (1MB) | Preview
[img]
Preview
PDF (Configuration manual)
Download (2MB) | Preview

Abstract

The infield of agricultural, farmers do need to perform various activities throughout the year on land surface as well on subsurface and these activities get affected by soil moisture. That's why it makes soil moisture game-changer in agricultural for farmers to achieve growth in the farming business. So, it is very essential to get trusted and most accurate real-time soil moisture data in order to help crop water demand with managing water resources. This study focused on use of neural network-based models such as LSTM(Long short-term memory) and ANN(Artificial Neural Network) to help farmers by prognosticating 24-Hours before soil moisture with help of this farmer would get awareness of crop water needs and help them to improve water resources management in field of agriculture. This developed model trained on past volumetric soil moisture data and able to forecast the next day that is 24-Hours in the future. The neural network model trained on U.S.Geological Survey(USGS) data independently on two sites. The various types of transformation introduced in the field of soil moisture prediction to get high performance while training neural network models without damaging its true sequential nature. The result of each site of developed models is compared against each other and with MSE(Mean Squared Error) as an evaluation parameter. A value of MSE of LSTM model is 0:001552 for the Dana Meadow site and for the Gin Flat, it is 0:002365. Whereas, ANN model is 0:001316 for the Dana Meadow site and the Gin Flat is 0:002022. Results of LSTM model outperformed ANN in terms prediction of volumetric soil moisture for 10cm and 36cm depth for next day that is 24-Hours in future for both sites by catching all oscillation like true values.

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science

Q Science > QA Mathematics > Computer software
T Technology > T Technology (General) > Information Technology > Computer software

S Agriculture > S Agriculture (General) > Farming Industry
Divisions: School of Computing > Master of Science in Data Analytics
Depositing User: Dan English
Date Deposited: 16 Jun 2020 12:19
Last Modified: 16 Jun 2020 12:19
URI: http://norma.ncirl.ie/id/eprint/4295

Actions (login required)

View Item View Item