| The world is experiencing significant climate change with a main feature (temperature rising). Global average surface temperature has increased by0.74℃in the last100years (1906to2005). The global warming also caused the sea level rising, precipitation changing and the occurrence frequency of extreme weather events increasing. Because agricultural production closely depends on the natural conditions, especially climatic conditions (changes of temperature, precipitation, wind speed and atmospheric carbon dioxide concentration), climate change has a significant impact on it. Taihu Lake region is subtropical monsoon climate, with the changefully advance, retreat and strength of monsoon, and the obvious inter-annual and seasonal variation, resulting in frequently catastrophic climate like cold damage, flood, waterlogging, typhoon, cold wave and other disasters. Based on the meteorological data and shallow soil temperature in Changshu during1960～2009, the characteristics of climate change and their influences on soil temperature in the typical farmland of Tai lake region have been investigated, while climate change trends in the past50years have been understood objectively and correctly in the study area, which have great significance on improving the ecological environment and agricultural production in the region.Climate change has become one of the world’s most serious environmental problems. As one of the environmental ecological characterization factors, soil moisture will inevitably produce a certain response to climate change. As a comprehensive index of the surface hydrological processes, Soil moisture has not only accumulated a large number of surface hydrological processes information, but also reflected the surface hydrological processes-the combined effect of precipitation and evaporation. According to natural generating layer soil moisture samples were collected in the surface soil layer (0～14cm) and plow (14～33cm) where soil moisture dynamics had a closer relationship with and meteorological factors in the study area, and the acquisition time was form winter wheat planting to maturity:October23,2010to June12,2011. Neural network sensitivity analysis was used to determine main meteorological factors that affect soil moisture dynamics in the typical farmland of Tai lake region, then soil moisture prediction method based on artificial neural networks, which provided a scientific data for taking full advantage of the agro-climatic resources, grasping influence condition of the crops by drought conditions, scientifically guiding the water management and disaster prevention and mitigation, and agricultural evaluation and drought damage assessment.Results of this thesis were as follows:(1) The average temperature and accumulated temperature during the growing season of the study area showed an upward trend in the past50years (The climate trend rate of spring, summer, autumn, winter and annual average temperature were0.45℃/10a,0.16℃/10a0.30℃/10a,0.45℃/10a,0.33℃/10a respectively. The climate trend rate of accumulated temperature during the growing season was106.75℃/10a), indicating that temperature raised obviously in the study area in the past50years, which indicated with great background of climate change. By the analysis of climate trend rate, the average maximum and minimum air temperatures also showed a growing trend in the past50years. Annual precipitation and cumulative precipitation during growing season increased gradually, but the trends of each seasonal cumulative precipitation were inconsistent, reducing in spring and autumn (the climate trend rates were-13.51mm/10a42.95mm/10a respectively), increasing in summer and winter (the climate trend rates were-19.16mm/10a,19.03mm/10a respectively). In addition to the winter, the climate trend rate of cumulative evaporation were positive (the climate trend rate of spring, summer, autumn, winter, year and growth period were14.93mm/10a of1.65mm/10a2.19mm/10a of-3.75mm/10a,14.80mm/10a30.12mm/10a respectively), indicating that evaporation increased little in the past50years in generally. The climate trend rates of average air pressure, relative humidity, wind speed, and cumulative sunshine hours were negative, showing that all of they reduced in the past50years.(2) There were six meteorological elements with abrupt climate change in study area during1960～2009, in which abrupt climate change of average wind speed appearing in1978and1980was the most obviously. The occurrence of abrupt climate changes of annual average temperature, annual average maximum temperature and spring average temperatures were in1989, while the occurrence of abrupt climate changes of annual mean minimum temperature and spring mean minimum temperature were in1988and2002, respectively, meaning a climatic jump from a relatively cool period to a relatively warm period. The abrupt climate changes of annal average air pressure and spring average air pressure occurred in1997, while that of summer average air pressure occurred in2004. The abrupt climate changes of annual cumulative sunshine hours and summer cumulative sunshine hours occurred in1979; other climatic elements didn’t exist abrupt climate change.(3) In the all meteorological elements, the climatic extreme year of mean relative humidity mostly appeared at the beginning of the twenty-first century had the highest occurrence rate; while cumulative evaporation had the higher occurrence rate, appearing in each era, The average wind speed had the least abnormal phenomenon, only the average wind speed of summer and autumn occurred climate anomaly in the1960s and1970s.(4) With the increment of the depth of soil layer, the response to mean soil temperature from climate change became increasingly weak, while the response in autumn was the weakest. All the mean soil temperatures at0cm,5cm and20cm had significant correlations with the mean air temperature, mean maximum air temperature and mean minimum air temperature, and all the significant relationships were up to0.01.(5) The strength of function relation between the soil moisture and meteorological factors could be confirmed by sensitivity analysis basing on the BP-ANN. From the local and global sensitivity analysis results, the sensitivity between soil moisture and precipitation was most than others, the average temperature and average surface temperature took second place, while the sensitivity between soil moisture and maximum temperature, minimum temperature, average wind speed and average relative humidity were the least, which can not be considered or less considered in the modeling process.(6) The simulation precision accuracy (PA) of soil moisture dynamics in0-14cm layer was higher than that in14～33cm layer, which was caused by the more strongly response between the surface soil moisture and meteorological factors. Both precision accuracies of training samples and testing samples in two layers were more than0.87by using LS-SVM model, were higher than the precision accuracies of the BP-ANN and RBF-ANN. LS-SVM model had a more clear mathematical meaning, and the application of structural risk minimization principle overcomed the curse of dimensionality and local minimum of the neural network. In practical applications, the learning and training time of LS-SVM were less than that of BP-ANN and RBF-ANN, and the result of the operation was relatively stable, so compared to the BP-ANN RBF-ANN, based on the LS-SVM to build a soil moisture dynamic change model with joining the meteorological factors had stronger superiority.. |
PDF Full Text Request