A Review of Intelligent Practices for Irrigation Prediction (1612.02893v1)

Abstract: Population growth and increasing droughts are creating unprecedented strain on the continued availability of water resources. Since irrigation is a major consumer of fresh water, wastage of resources in this sector could have strong consequences. To address this issue, irrigation water management and prediction techniques need to be employed effectively and should be able to account for the variabilities present in the environment. The different techniques surveyed in this paper can be classified into two categories: computational and statistical. Computational methods deal with scientific correlations between physical parameters whereas statistical methods involve specific prediction algorithms that can be used to automate the process of irrigation water prediction. These algorithms interpret semantic relationships between the various parameters of temperature, pressure, evapotranspiration etc. and store them as numerical precomputed entities specific to the conditions and the area used as the data for the training corpus used to train it. We focus on reviewing the computational methods used to determine Evapotranspiration and its implications. We compare the efficiencies of different data mining and machine learning methods implemented in this area, such as Logistic Regression, Decision Tress Classifier, SysFor, Support Vector Machine(SVM), Fuzzy Logic techniques, Artifical Neural Networks(ANNs) and various hybrids of Genetic Algorithms (GA) applied to irrigation prediction. We also recommend a possible technique for the same based on its superior results in other such time series analysis tasks.

• The paper reviews diverse computational and data mining techniques, highlighting an RNN LSTM model to address temporal dependencies in irrigation prediction.
• It details methods from evapotranspiration measurement using the FAO Penman-Monteith equation to hybrid models combining neural networks and fuzzy logic.
• The study underscores actionable implications for optimizing water resource management and guiding future research with enhanced real-time forecasting.

Overview of Intelligent Practices for Irrigation Prediction

The proliferation of population and the resultant increase in drought occurrences pose significant challenges to the sustainability of water resources, prominently affecting the irrigation sector. As a major consumer of freshwater, effective irrigation management becomes pivotal. This paper reviews intelligent irrigation prediction techniques, classifying them into computational methods and statistical data mining algorithms. It emphasizes the significance of capturing semantic relationships among key parameters like temperature, pressure, and evapotranspiration, tailored to specific locales reflected in the training data.

Key Techniques and Their Efficacies

The paper delineates various methods employed in irrigation prediction, detailing their computational underpinnings and predictive effectiveness. Techniques span from classical models to modern machine learning approaches:

1. Evapotranspiration Measurement: A cornerstone computational method emphasized in the paper involves utilizing the FAO Penman-Monteith equation for evapotranspiration (ET) computations. Regarded for its empirical precision under controlled conditions, ET's real-time application is challenged by environmental variability.
2. Data Mining and Machine Learning Methods: A spectrum of models, including Logistic Regression, Decision Trees, SVM, and ANN, among others, is explored. Each technique is scrutinized for its predictive accuracy and adaptability:
   • Decision Tree Classifiers and SysFor: These hierarchical models provide logical rule-based predictions but lack time-series capability.
   • Support Vector Machines (SVM) and ANNs: While these models excel in regular data prediction, their ineptitude in handling temporal dependencies is highlighted.
   • Fuzzy-based Systems: Models like ANFIS combine neural networks with fuzzy logic, improving upon non-linear and uncertain parameter predictions.
3. Hybrid Models: The paper discusses innovative hybrid models integrating neural networks, fuzzy systems, and genetic algorithms to enhance prediction accuracy. One notable hybrid model demonstrated a marked improvement with a reduction in the error metric.

Proposed RNN LSTM Model

Acknowledging the limitations of existing techniques, the paper advocates for an RNN-based LSTM model. This model is proposed to address inherent temporal dependencies in irrigation prediction, offering a robust transformation to accommodate long sequence memory. While computationally intensive, RNN LSTMs offer enhancement in prediction accuracy and contextual learning through time-sequenced data analysis.

Implications and Future Directions

The implications of this research are manifold, suggesting a directional shift towards deploying sequence-learning architectures for more refined irrigation predictions. This development could greatly influence water resource management policies and inform agricultural planning, particularly amidst growing ecological unpredictabilities. The practicality of such models, however, hinges on the availability of computational resources and extensive training datasets.

The anticipated future trajectory involves not only refining RNN and LSTM applications but also integrating novel data sources such as IoT sensors, enhancing real-time irrigation forecasting capabilities. Further research could leverage these models' adaptability to variolic meteorological and agricultural conditions, potentially transforming irrigation management paradigms globally.