Introduction
Agriculture is the main axis of economy and development of rural areas of Azna County, which has an important role in the performance of the village and the rural environment. Regarding to natural environment and climatic factors, performance evaluation of agricultural products is one of the important pillars of sustainability of food supply and rural economy. Climatic hazards such as frost cause damage to agricultural products and the economy of rural areas. Freezing and lowering the temperature in mountainous areas is very important. Because precipitation in these agricultural areas plays a key role in the region's economy.
Azna is one of the counties of Lorestan Province, whose rural areas, like most villages in the country, have an economy based on agricultural activities. In the villages of this city, in addition to the cultivation of conventional crops such as wheat, the cultivation of crops such as beans is common, as this city is the center of bean production in the province and the country. The climate of this region, which is partly affected by the geographical location of Azna County, has positive and negative effects on agricultural products such as beans. The occurrence of spring frosts has caused damage to this crop in recent years. In this regard, this study seeks to investigate the relationship between spring frost and bean crop yield and its effect on rural economy. Another goal of modeling is frost alternations and its relationship with crop yields.
Materials and Methods
Pearson correlation test was used to investigate the relationship between spring frost and bean cultivation performance. Two models of K-NN and artificial neural network were used to model the behavior of frost and its effect on frost in bean crop. For this purpose, 15-year statistics (2004-2008) of Azna meteorological station were used, which includes information on minimum temperatures and sub-zero temperatures and tolerance thresholds in different stages of bean plant phenology.
The nearest neighbor K-NN method is one of the most popular nonparametric regression methods. In this method, the prediction value distribution function is obtained using the nonparametric distribution of the kernel function. This model is formulated in such a way that whenever conditions similar to the historical conditions observed in the present occurrence, the probable conditions in the future will be similar to the conditions that occurred on that date.
Neural networks are actually mathematical models for the rapid and accurate processing of information that are able to communicate between the inputs and outputs of a physical system, connected by a network of nodes. In other words, the artificial neural network is a computational mechanism that is able to provide a series of new information by capturing and calculating information.
Discussion and conclusion
This research consists of two main stages. In the first step, the relationship between bean harvest rate and the phenomenon of freezing and frost is evaluated, and in the second step, using the K-NN algorithm and neural networks, the effects of freezing on bean cultivation are predicted based on observational data Based on the data extracted from the minimum daily temperature of Azna station., which indicates the long frost season in this region. The mild frosts occur mostly in October (6 cases) and the rest in December (8 cases). The earliest date of frosts is on the 20th of October and the latest date of frosts is on the 18th of December.
The results of Pearson test showed that there is a correlation coefficient (42%) between bean harvests in Azna villages with spring frosts. In other words, the occurrence of spring frosts reduces bean yield by up to 42%. As a result, the level of farmers' incomes decreases and farmers whose source of income is heavily dependent on annual crops experience severe economic fluctuations. Other objectives of the study were to model the temperature drop and frost damage in crops and bean yield in Azna County using two approaches K-NN and artificial neural network. The results of Nash-Sutcliffe and RMSE statistics in the K-NN model are (76.4%) and (0.0785), respectively, while the same statistics in the artificial neural network model are equal to (81.5%) and (0.0688).
Therefore, in the same condition, the artificial neural network model predicted the actual data better, based on computational data. Thus, in agricultural economics development plans, this model provides a merrier estimate of the state of frost and its effect on bean crop yield. Lowering the production, crop frost will affect other indices such as employment rate, income shortage, savings and investment by reducing production. Because cultivated area of this crop is higher than other crops, frost in beans has caused great economic losses.
Type of Study:
Research |
Subject:
Special Received: 2021/02/28 | Accepted: 2021/02/28