Study On The Climate Yield Of Winter Wheat In Henan Province Based On Artificial Neural Network

Under the influence of natural factors and human activities over the past 30 years, the global average temperature has increased by 0.89?(0.65?1.06?), and the trend of global warming is irreversible. Under the background of climate warming, global social, economic and ecological environment are deeply affected. The response of agriculture to climate change is the most sensitive and vulnerable filed, and to discuss the impact of climate change on agriculture is the key to ensure the stability and sustainability of agriculture. Wheat is an important food crop in China, and Henan province is one of China's major wheat growing areas, due to its unique geographical location, to study the effect of climate change on wheat yield in this region is of great significance. On the one hand, a simple regional demonstration was provided to assess the impact of global climate change on agriculture; On the other hand, it will lay a scientific foundation for the development of scientific and reasonable measures to deal with and adapt to the impact of global climate change.Based on the data of average temperature, precipitation, sunshine duration, daily minimum temperature and daily maximum temperature in Henan province and its surroundings, a total of 32 meteorological stations, and the winter wheat yield data of Henan, Hebei and Shandong province were selected from 1961 to 2014. Combining analysis software such as Excel, SPSS, Eviews and MATLAB, firstly climate change were analyzed by using moving average of five years, M-K test, wavelet analysis, spatial interpolation and linear tendency rate in Henan province since recent 54 years in this paper; Secondly, in order to separate the climatic yield,6 kinds of trend yield separation were used, and the best separation method was selected reasonably; Finally, the critical meteorological factors were selected and inspected by using correlation analysis and multicollinearity; And the artificial neural network model was constructed to simulate and predict spatio-temporal variability of the climatic yield in Henan province.Through the study, we got the following conclusions:(1) The yield per unit area of winter wheat in Henan province is increasing year by year, but the growth rate has slowed down. The order of temperature rising rate in Henan province in recent 54 years were daily minimum temperature, mean temperature and daily maximum temperature, they also exist more than 20 years'domain-wide cycle of total scale signal. Their spatial distribution patterns were shown that southeast was relatively higher than those in northwest, and the regional differences of the daily maximum temperature was narrowing, the daily minimum temperature and average temperature were expanding. Annual precipitation was increasing, while the sunshine duration was significantly reducing, two of them existed separately 25years'and 14-25 years'domain-wide cycle of total scale signal, precipitation distribution in the south was higher than that in the north, and the differences between north and south were still increasing; Sunshine duration distribution in the northeast was higher than that in the southwest, and sunshine duration of northwest had a increasing trend.(2) In the separation of climatic yield, compared with moving average of three years, moving average of five years, second exponential smoothing method, quadratic curve fitting and cubic curve fitting, H-P filter have obvious advantages.(3) The key growing period which impacts climatic yield in Henan province have 4 stages, include the sowing date, the overwinter time, the jointing stage and the heading stage. Precipitation and sunshine duration have higher contribution, but the temperature was relatively small during the growth period.(4) To ensure the accuracy of simulation and prediction, the meteorological factors that affect climatic yield in Henan province were performed multicollinearity test, the result showed that there was no mutual influence between them, it was not required to eliminate collinearity effect on the model. Because of the complex nonlinear relationship between meteorological factors and climatic yield, it determined ultimately that using artificial neural network to simulate climatic yield in Henan province, and the model parameters were set:using 10 hidden layer nodes, setting 2000 as the number of training, using the tansig function as a transfer function.(5) From the time point of view, climatic yield of winter wheat mutation occurred in 1971 from low to high; and it increased significantly in 1980. The climatic yield existed 17-27 years' domain-wide cycle of total scale signal. In the coming decades the average climatic yield of winter wheat in Henan province may be at a relatively high level.From the point of space, the distribution pattern of climatic yield in the middle was higher than that in neighboring regions in 1960s; It was high in southeast, but low in northwest in 1970s; The north was high while the south was low in1980s, and the climatic yield positive area was increasing in this period; The spatial distribution pattern of climate change in 1990s was high in the southeast but low in most areas of the northwest and northeast regions; The spatial distribution pattern increased from the northwest to the southeast in the first 10 years of 21st century; The climatic yield was higher in northeast and lower in southwest in nearly four years. The largest changes of climatic yield occurred in middle-east of Henan province, followed by the southern region, but the Nanyang basin has been in a low climatic yield.In the future, climatic yield showed a significant growth trend in middle-east area of Henan, and a decreasing trend in the mountain area of northern Henan, but no significant; The remaining area was not obvious trend; Hydrothermal conditions were the key to influence the climatic yield of winter wheat.