| The material basis and carrier of agricultural production is the soil,it is a complex and changeable entity.It is affected by many factors such as the soil parent material, climate, method of management and so on.Many studies have been proved that the soil nutrientsand thecrop yield have the obvious characteristics of the spatial variability in time and space. A lot of people have put the amount of soil nutrients is used as a measure of the output and quality of the important indexes in the actual production.It is an important target to predict the variation characteristics of soil nutrient spatial and temporal variability in precision farming precision fertilization.It is the foundation of precision agriculture management to research soil nutrient spatial and temporal variability in planting area. The research is also related to the major technical problems in the sustainable development of agriculture in china.In this study, the typical precision operation area soil(black soil) of Jilin Provincewas choosen the research object by using classical statistics, geostatistics, spatial autocorrelation, time series analysis method and combination evaluation method in the soil.The important factors are alkali hydrolyzable nitrogen, available phosphorus, available potassium content, organic matter, soil moisture, soil temperature and other indicators of different in the time and space change law. The results can provide the scientific method and strategy for precision fertilization operation area in agricultural production management.Through this research, the following results are obtained: 1. Study on spatial variability of soil nutrient based on Statisticsin precision operation area(1)The basic characteristic analysis of soil nutrient by using classic statisticsThe research results show that soil alkali hydrolysable nitrogen, organic matter, soil water, pH value after inspection belong entirely to the normal distribution, and the original data of available phosphorus and available potassium does not belong the normal distribution.The variation coefficient value of soil organic matter and pH is less than 10%.The spatial variability characteristics is weak. The variation coefficient value of Soil pH value is only 2.88%, which indicates that the pH value of the soil is good stability, and is less affected by man-made factors. The variation coefficient value of other parameters was between 11.15%-48.59%.They belong to moderate spatial variability.(2)The spatial trends analysis of soil nutrient based on GeostatisticsThe analysis results show that rear left and right projection trend lines were drawn through the content value of the alkali hydrolyzable nitrogen, the available phosphorus, the available potassium, the organic matter and the pH value of soil. The property characteristics of them were grasped from the east to the West and from south to north. The spatial distribution of soil was described by using the Kriging interpolation method. Execept the variation property of soil water was weak spatial correlation, the spatial correlation of the alkali hydrolyzable nitrogen, the available phosphorus, the available potassium, the organic matter and the pH value was medium.The results showed that the nutrient distribution wasdetermined by the interaction of regional factors and non regional factor. 2.The variation of the soil temperature and humidity based on the time series modelThrough the application of various curve models to forecast the comparative analysis, the results showed that temperature prediction model of three exponential smoothing curve model is relatively good based on hours.Its effect is the best model than others prediction model.The humidity prediction effect of single exponential smoothing model is good based on hours.The temperature prediction model of an exponential smoothing model is good based on days. The two moving average of the humidity can be used to simulate well based on days. The relative errors of them are smaller in hours than in days. It can be seen that temperature is more suitable for this method to predict the specific humidity from the application of curve method for temperature and humidity forecast analysis results. 3. The multiple factors of influcing on maize yield in precise operation area(1)The timing analysis characteristic of maize yieldThe conclusions as follows: if you choose to use curve estimation scheme to predict the yield,the cubic equation predict is the best than other curves in regression analysis.If you choose to the ARIMA model of time series analysis to forecast, the ARIMA(2, 1, 1) of time series model is good results.The production line of different years is normal distribution.The coefficient of variation of yield decreased gradually, after several rounds of precision operation, and the coefficient of variation was less than 20%. The yield lineartrend line is smoothin east-west direction, and the difference is be weakened gradually by precision operation.(2)The distribution characteristics of trace elements boronThe results showed that trace elements boron has spatial autocorrelation according to Moran’s I coefficient evaluation method in the precise operation area.When the sampling distance of soil is increasing, variogram value is increasing too. The point cloudvariation of boron is significantly different in the two directions.There are longer distance of the spatial autocorrelation in the east-west direction than the north-south direction. 4. The application of data mining method in soil fertility evaluation(1)Fertility evaluation results based on principal component analysisAccording to the characteristic of soil nutrient, there are 5 principal components by using principal component analysis.The extraction of the cumulative contribution rate is 87.98%. The rates were respectively 40.93%, 16.45%, 15.15%, 9.16% and 6.29%.The variance contribution of each factor formula was given.(2)Fertility evaluation results based on clustering analysisThe precise operation area No. 5 was the object of study.When the clustering distance is 0.285, the soil is divided into four categories.The characteristics of the first class is that the available phosphorus is very low.The characteristics of thesecond class is that the alkaline hydrolysis nitrogen high type.The characteristics of the third classis that the content value of organic matter and available potassium is high.The characteristics of thefourth classis that available phosphorus fertilization is high.The different types fertilization schemes were given corresponding to the different categorys, and can guide the actual production field in the future. |
PDF Full Text Request