عنوان مقاله [English]
Nowadays, one of the greatest difficulties in agricultural activities is effects of weather factors on agricultural crops. The purpose of this research is to present prognosis of frosting occurrence in almond orchards in Najafabad region. Freezing can damage agriculture crops on the condition that it was prolonging and intense. The frost phenomenon and consequent damages is a serious problem in all the regions of the world. On time and accurate forecasting of occurrence time can reduce damages. Though about frostbite many studies have been done but the majority of this research just has examined this phenomenon from the perspective of the weather and by specifically the productive synoptic patterns. In the agricultural scale, great views have often useless results. These predictions are scale down on spatial and temporal (Downscaling). The purpose of this research is to combine two phonological and meteorological models (WRF) in order to predict the phenomenon of spring frost in almond gardens of the region.
Materials and methods
Case of study of this research is Zayandeh Rud Basin. Zayandeh Rud Basin as the study area is located 50 degrees and 20 minute until 52 degrees and 24 minute eastern longitude and 31 degrees and 12 minute until 33 degrees and 42 minute northern latitude.
Since the concept of cold damage in agriculture, regardless of product development stages (phenology) and only the temperature test, has no practical value. Therefore, it was necessary to estimate the model of almond flowering in studied gardens. The results are combined with weather forecasting model (WRF). According to the phenology long-term statistics of almond trees in the Najafabad region, the extracted flowering dates were calculated based on Julian dates. Then, the GDD matrix table is plotted for the analysis.
The relationship between flowering history is evaluated with the parameters mentioned. The strongest relationship is selected for further evaluations. The flowering is dependent variable and the other above mentioned is independent variable. The most relevant regression equation is determined with high correlation coefficient. According to the WRF model, such analysis in the study area was performed by Nasr Esfahani et al and predicted model temperature was statistically significant. A quick warning can be made of the occurrence of frostbite in the coming days for the flowering date.
Results and discussion
According to the phonological data of Najafabad station, the flowering date for each year based on Julian calendar, was analyzed in the statistical population of 10 years (2006-2017). The average flowering period is eighty days (80). The years 2007, 2011 and 2014 have the highest flowering time. The number of flowering days was twelve days in 2007, ten days in 2011, fifteen days in 2012, eight days in 2014and six days in 2017. Among the existing parameters, the highest correlation (0.945) was revealed between the flowering date and the number of days above the mean. This is a positive relationship and show that the number of days above the average is related to the flowering in almonds. The value of P-value <0.01 is significant at 1% level. After the flowering date and GDD greater than zero (0.938) with a significant level of 0.01, there is the highest correlation between the flowering date and the number of days below the average of over five (0.921) with a significant level of 0.05. Linear multi-variable regression equations were also investigated. This equation was correlated with 0.96 and there is a very strong connection between the dependent and independent variables. The regression coefficients are estimated separately for the two models using the first column of the table. We can use the regression model as Y = 36.605 + 1.761X for the first model. Where x is the number of days above the average and y is the date of flowering. The estimated regression model for the second model is
Where, X1 is the number of days above the average and X2 is the number of days below the average. Then, we have selected the best model and defined the coefficient for both models. It is noted that the modified coefficient of the second model is determined with two independent variables (0.998). This is higher than the first-order correction coefficient (0.88). It can be concluded that the second model is better than the first one.
As we have seen, the linear multivariate regression equation was significant for predicting flowering history at 1% confidence level. Now with the temperature prediction by the WRF model and calculating the number of days above the average and below the average, the flowering date can be obtained and show the warningss of frostbite in the presence of flowering.
Similar to this study, Prabha and Hoogenboom (2008) showed that use of the WRF Intermediate Scale for effective protection management is a good strategy to protect products and reduce frost damage. The results of this study reveal the feasibility and accuracy of the WRF model for radiation and radiant frost warning. Given the daily temperature test outputs of the WRF model in the study area, we can observe very good results in smooth areas. It can be combined with two phenology models and the prediction of temperature by the WRF model, as a quick warning of 48 hours of frostbite occurred in the gardens of the area with sufficient accuracy.