Abstract
Frost forecast is an important issue in climate research because of its economic impact on several industries. In this study, we propose GRAST-Frost, a graph neural network (GNN) with spatio-temporal architecture, which is used to predict minimum temperatures and the incidence of frost. We developed an IoT platform capable of acquiring weather data from an experimental site, and in addition, data were collected from 10 weather stations in close proximity to the aforementioned site. The model considers spatial and temporal relations while processing multiple time series simultaneously. Performing predictions of 6, 12, 24, and 48 h in advance, this model outperforms classical time series forecasting methods, including linear and nonlinear machine learning methods, simple deep learning architectures, and nongraph deep learning models. In addition, we show that our model significantly improves on the current state of the art of frost forecasting methods.
Highlights
Attention for Multi-Sources TimeGenerating accurate weather forecasts from reliable localized data is a key feature of precision agriculture that enables farmers to improve their resources in terms of efficiency, productivity, sustainability, etc
The historical average accounts for weekly seasonality and predicts for a day by using the weighted average of the same day in the past few weeks; Deep learning methods that produce forecasting for each series separately such as feed-forward neural network (FNN) and long short-term memory (LSTM); Autoencoder forecasting method with attention mechanism (AC-att); Graph convolutional network applied to the given graph without spatio-temporal attention mechanism (GCN); Variants of this architecture using convolutions [7] and GRU [26]
All methods are evaluated with three metrics: mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE)
Summary
Attention for Multi-Sources TimeGenerating accurate weather forecasts from reliable localized data is a key feature of precision agriculture that enables farmers to improve their resources in terms of efficiency, productivity, sustainability, etc. It is a tool for countering weather uncertainty by reducing the risks posed by extreme weather that can impact the overall quality of the production. Frost is one such threat that kills plant tissue, causing low production and economic losses since it prevents the normal development of crops. A low temperature can cause the crop to flower early or if there is frost it can cause a considerable reduction in production [1] These negative consequences could be prevented or mitigated with a frost forecast model that provides information to the farmer regarding the probability of a frost event several hours in advance, so the farmer can take action to protect the crops
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call