| 本文已被：浏览 1590次 下载 569次 |
码上扫一扫！ |
|
|
| 基于不同样本分区和模型参数的四川省粮食产量空间化研究 |
|
何鹏1,2,3, 林正雨1,2,3, 景晓卫4, 李晓1,2,3
|
|
1.四川省农业科学院农业信息与农村经济研究所，成都 610066;2.四川省农业科学院区域农业发展研究中心，成都 610066;3.四川省农业科学院大数据中心，成都 610066;4.四川省农业科学院农业信息与农村经济研究所，成都 610066
|
|
| 摘要: |
| ［目的］农业经济统计数据受多种自然条件因素和社会经济条件因素影响，数据空间化难度很大。通过省级尺度粮食产量数据空间化的研究，探索提高农业经济统计数据空间化精度的方法及经验。［方法］以四川省为例，按照不同分区方案（全省不分区、分为5个综合农业分区），选择不同样本数据（县级粮食总产、县级平均粮食产量）为因变量，对应2种耕地类型（水田、旱地）面积数据为自变量，考虑2种模型参数（常数为0和不为0）拟合多元回归模型，对粮食数据进行空间化研究，并选择误差评价因子对空间化结果进行评价。［结果］（1）不管采用县级粮食总产还是县平均粮食产量拟合，常数项设置为非0的，均是分区比不分区的结果精度要高； （2）不管采用县级粮食总产还是县平均粮食产量拟合，常数项设置为0的，均是分区比不分区的结果精度要高； （3）对比不分区和分区这2种方法，以县级粮食总产拟合的结果要比县级平均粮食产量拟合的结果精度要高。［结论］在省级粮食产量空间化时，以分县平均粮食产量为基础，划分土地利用类型、划分农业分区并且常数为0时拟合精度最高。在今后的研究中，有必要结合更多的影响因子进行粮食产量的空间化，以提高其数据重构精度。 |
| 关键词: 粮食 模型 分区方案 多元回归 误差 空间化 |
| DOI：10.7621/cjarrp.1005-9121.20170904 |
| 分类号:F326.11;F224 |
| 基金项目:四川省财政创新能力提升工程“大数据背景下主要农产品数量安全预警分析技术研究”（2016GYSH 004）； 四川省科技支撑计划软科学项目“基于精细区划的农产品气候品质认证评价指标体系研究”（2017ZR0045） |
|
| SPATIAL DISTRIBUTION OF GRAIN YIELD OF SICHUAN BASED ON DIFFERENT SAMPLE SCALES AND PARTITIONING SCHEMES |
|
He Peng1,2,3, Lin Zhengyu1,2,3, Jing Xiaowei4, Li Xiao1,2,3
|
|
1．Agricultural Information and Rural Economy Institute, Sichuan Academy of Agricultural Sciences, Chengdu 610066, China;2．Regional Agricultural Research Center, Sichuan Academy of Agricultural Sciences, Chengdu 610066, China;3．Large Data Center, Sichuan Academy of Agricultural Sciences, Chengdu 610066, China;4.Agricultural Information and Rural Economy Institute, Sichuan Academy of Agricultural Sciences, Chengdu 610066, China
|
| Abstract: |
| he statistical data of agricultural economy are affected by various natural conditions and social and economic conditions. To improve the spatial accuracy of statistical data of agricultural economy, this paper explored the spatialization of grain output data at the provincial level. Taking Sichuan province as an example, according to the different partitioning schemes, it choses a different sample dat as the dependent variable, and the two types of cultivated land, paddy field, dry land area data as independent variables, and constructed a fitting multiple regression mode considering two kinds of model parameters. The results showed that the spatial accuracy of statistical data was higher in partition than that in no partition, no matter whether fitting by the average grain yield in county-level grain production or the average grain yield of county. Compared with the two methods of non-partitioning and partitioning, the fitting accuracy of the total output of the grain at the county level was higher than that of the average grain yield. It proposed that it was necessary to improve the accuracy of data reconstruction by combining more influence factors of the spatial distribution of grain yield. |
| Key words: grain models partitioning scheme multiple variable regression error spatialization |
