气候资源插值算法在多核环境下的并行计算研究
发布时间：2018-08-11 15:28
【摘要】：气候资源是与农业生产最为密切的资源，并在很大程度上影响主要农作物的生长，直接影响了广大农民的收入和国家的粮食安全。利用地理信息系统实现小网格点的气候资源推算和分析，通过详细描述气候资源在不同地形地貌下的空间分布特征，为区域性的特色农业发展、农业生产合理布局以及未来农业发展规划提供可靠的科学依据。 在精细化农业气候区划中，需要对大量的历史数据、使用多种插值算法进行小网格插值计算，对于海量数据的处理及插值算法的复杂性，需要大量的计算时间，导致系统的反应速度很慢，系统整体性能下降。 克里金(Kriging)插值法，是局部插值法,是一种最优内插法。Kriging插值是建立在变异函数空间分析基础上，，对有限区域内的区域化变量取值进行无偏最优估计的一种方法。与其它插值方法相比，不仅考虑了待插点与邻近己知点的空间相关性，而且能够给出估计误差。用该方法对湖南小网格点气温数据进行海量插值，由于串行Kriging算法比较复杂，且循环较多导致计算量大，且小网格点个数达到107级会导致计算量较大，计算时间较长。在基于传统的单处理器模式，当处理大规模海量数据时，由于计算时间太长，无法满足实时分析的需求。虽然在高性能计算机或者分布式计算机中得以解决，但是在PC中无法满足实时计算的需求。 多核并行计算技术是计算技术发展的重要方向之一。多核计算是使用并行处理技术进行编程，开发并行性、同时执行多个任务，为合理地提高多核处理器性能提供一个理想的平台，也是提升系统性能的关键技术之一，使目前计算机的处理水平有一个质的飞跃。因此，使用多线程编程技术和基于共享存储的OpenMP编程模型，对串行Kriging算法进行改进，不仅改善了该算法效率，而且大量实验数据也表明改进的Kriging算法的性价比很高，满足了对湖南地区小网格实时插值的需求。 采用了多核并行计算技术所构建的信息系统，通过多线程编程技术，充分利用硬件资源，使得并行计算的硬件资源在信息系统开发过程中真正发挥了作用，大大提高了系统的反应速度和整体性能表现。
[Abstract]:Climate resources are the most closely related to agricultural production, and to a large extent affect the growth of major crops, directly affect the income of farmers and national food security. The geographic information system (GIS) is used to calculate and analyze the climate resources of small grid points. By describing the spatial distribution characteristics of climate resources in different landforms and landforms in detail, it is a regional characteristic agriculture development. Reasonable distribution of agricultural production and future agricultural development plan provide reliable scientific basis. In fine agricultural climate regionalization, a large number of historical data are needed, and a variety of interpolation algorithms are used to carry out small grid interpolation calculation. For the processing of massive data and the complexity of interpolation algorithm, it takes a lot of computing time. As a result, the response speed of the system is very slow and the overall performance of the system is reduced. Kriging (Kriging) interpolation method is a local interpolation method. Kriging interpolation is an unbiased optimal method for estimating the values of regionalized variables in a finite region based on the spatial analysis of variogram. Compared with other interpolation methods, not only the spatial correlation between the points to be inserted and the nearest known points is considered, but also the estimation error is given. This method is used to interpolate the temperature data of small grid points in Hunan province. Because of the complexity of serial Kriging algorithm and the large amount of calculation due to more cycles, the number of small grid points reaching 107-level will lead to a large amount of calculation and a longer calculation time. Based on the traditional single-processor mode, when processing massive data, the computation time is too long to meet the needs of real-time analysis. Although it can be solved in high performance computer or distributed computer, it can not meet the demand of real-time computing in PC. Multi-core parallel computing technology is one of the important directions in the development of computing technology. Multi-core computing is to use parallel processing technology to program, develop parallelism and perform multiple tasks at the same time. It provides an ideal platform for reasonably improving the performance of multi-core processors, and is also one of the key technologies to improve system performance. Make the current computer processing level has a qualitative leap. Therefore, using multi-thread programming technology and OpenMP programming model based on shared storage, the serial Kriging algorithm is improved, which not only improves the efficiency of the algorithm, but also shows that the improved Kriging algorithm has a high performance-price ratio. It meets the demand of small grid interpolation in Hunan. The information system based on multi-core parallel computing technology is adopted. Through multi-thread programming technology, the hardware resources of parallel computing are fully utilized, which makes the hardware resources of parallel computing really play a role in the development of information system. The reaction speed and the overall performance of the system are greatly improved.
【学位授予单位】：湖南大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP338.6
本文编号：2177406
[Abstract]:Climate resources are the most closely related to agricultural production, and to a large extent affect the growth of major crops, directly affect the income of farmers and national food security. The geographic information system (GIS) is used to calculate and analyze the climate resources of small grid points. By describing the spatial distribution characteristics of climate resources in different landforms and landforms in detail, it is a regional characteristic agriculture development. Reasonable distribution of agricultural production and future agricultural development plan provide reliable scientific basis. In fine agricultural climate regionalization, a large number of historical data are needed, and a variety of interpolation algorithms are used to carry out small grid interpolation calculation. For the processing of massive data and the complexity of interpolation algorithm, it takes a lot of computing time. As a result, the response speed of the system is very slow and the overall performance of the system is reduced. Kriging (Kriging) interpolation method is a local interpolation method. Kriging interpolation is an unbiased optimal method for estimating the values of regionalized variables in a finite region based on the spatial analysis of variogram. Compared with other interpolation methods, not only the spatial correlation between the points to be inserted and the nearest known points is considered, but also the estimation error is given. This method is used to interpolate the temperature data of small grid points in Hunan province. Because of the complexity of serial Kriging algorithm and the large amount of calculation due to more cycles, the number of small grid points reaching 107-level will lead to a large amount of calculation and a longer calculation time. Based on the traditional single-processor mode, when processing massive data, the computation time is too long to meet the needs of real-time analysis. Although it can be solved in high performance computer or distributed computer, it can not meet the demand of real-time computing in PC. Multi-core parallel computing technology is one of the important directions in the development of computing technology. Multi-core computing is to use parallel processing technology to program, develop parallelism and perform multiple tasks at the same time. It provides an ideal platform for reasonably improving the performance of multi-core processors, and is also one of the key technologies to improve system performance. Make the current computer processing level has a qualitative leap. Therefore, using multi-thread programming technology and OpenMP programming model based on shared storage, the serial Kriging algorithm is improved, which not only improves the efficiency of the algorithm, but also shows that the improved Kriging algorithm has a high performance-price ratio. It meets the demand of small grid interpolation in Hunan. The information system based on multi-core parallel computing technology is adopted. Through multi-thread programming technology, the hardware resources of parallel computing are fully utilized, which makes the hardware resources of parallel computing really play a role in the development of information system. The reaction speed and the overall performance of the system are greatly improved.
【学位授予单位】：湖南大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP338.6
【参考文献】
相关期刊论文 前10条
1 曾怀恩;黄声享;;基于Kriging方法的空间数据插值研究[J];测绘工程;2007年05期
2 马轩龙;李春娥;陈全功;;基于GIS的气象要素空间插值方法研究[J];草业科学;2008年11期
3 傅抱璞;地形和海拔高度对降水的影响[J];地理学报;1992年04期
4 封志明,杨艳昭,丁晓强,林忠辉;气象要素空间插值方法优化[J];地理研究;2004年03期
5 刘宇;陈泮勤;张稳;胡非;;一种地面气温的空间插值方法及其误差分析[J];大气科学;2006年01期
6 李新,程国栋,卢玲;空间内插方法比较[J];地球科学进展;2000年03期
7 彭楠峰;;距离反比插值算法与Kriging插值算法的比较[J];大众科技;2008年05期
8 殷剑敏;辜晓青;林春;;寒露风灾害评估的空间分析模型研究[J];气象与减灾研究;2006年03期
9 苏姝,林爱文,刘庆华;普通Kriging法在空间内插中的运用[J];江南大学学报;2004年01期
10 陈鹏;王乘;任波;;并行Kriging地层电性参数分布估计[J];计算机工程与应用;2008年15期
相关硕士学位论文 前1条
1 程志;多核并行计算在视频服务中的研究及应用[D];山东大学;2008年
本文编号：2177406
本文链接：https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2177406.html
