| 摘要: |
| ［目的］空间化作为一种常用的地学数据处理方法，必然会存在一定的误差，而对空间化结果进行误差修正，可以降低空间化误差。［方法］文章以2005年粮食产量空间化为例，以各地市不同农田类型（水田、水浇地、旱地）面积数据为自变量，以各地市粮食产量统计数据为因变量，进行多元线性回归分析建模。在具体建模时，令常数项为0，将全国分为7个区，各区分别建立回归方程。然后分别利用4种误差修正方法对空间化初步结果进行修正。选取4种误差评价因子，对修正前后的空间化结果的精度进行对比和分析。［结果］①均值法不能被用于修正空间化初步结果； ②比例系数法、权重系数法Ⅰ（不同农田类型同一权重）和权重系数法Ⅱ（不同农田类型不同权重）3种方法都可以被用于修正空间化初步结果； ③利用权重系数法Ⅰ修正后的空间化结果的精度最高，比例系数法次之，权重系数法Ⅱ最差。［结论］误差修正方法对提高空间化精度具有重要影响。该研究虽以粮食产量空间化为例，但所得结论同样适用于其他社会经济统计数据的空间化研究，对以后统计型数据空间化研究具有一定的参考价值和指导作用。 |
| 关键词: 统计数据 空间化 误差修正 误差分析 粮食产量 |
| DOI：10.7621/cjarrp.1005-9121.20170819 |
| 分类号:F326.11 |
| 基金项目:中国科学院战略性先导科技专项“应对气候变化的碳收支认证及相关问题”（XDA05050000） |
|
| COMPARISON OF 4 ERROR CORRECTION METHODS FOR SPATIALIZATION OF GRAIN YIELD |
|
Liao Shunbao1,2, Ji Guangxing3,2, Wang Hui4, Yue Yanlin3,2
|
|
1. Institute of Disaster Prevention, Beijing 101601, China;2.4. College of Environment and Planning, Henan University, Kaifeng 475004, China;3.2. Key Laboratory of Geographic Information Science, East China Normal University, Ministry of Education, Shanghai 200241, China;4.School of Environment and Earth Science, Yunnan University, Kunming 650091, China
|
| Abstract: |
| As a kind of frequently-used geo-data processing method, spatialiation inevitably results in errors during data processing. Spatialization errors can be reduced by correcting spatialization results. In this paper, spatialization models were constructed with different types of farmland areas (paddy field, irrigated land, dry land)at prefectural level using the multiple variable linear regression analysis method. Then 4 error correction methods were used to modify preliminary results of spatialization. The results showed that: (a) the average correction method cannot be used to modify preliminary results of spatialization. (b) The proportional coefficient correction method, the weight coefficient correction method A (different farmland types with same weights) and B (different farmland types with different weights) can be used to modify preliminary results of spatialization. (c) The weight coefficient correction method A was the best method to improve precision of spatialization results. The error correction method had an important effect on the spatial accuracy. It was necessary to select the appropriate error correction method to improve the spatial accuracy. The results can provide valuable reference for spatialization of other types of social and economic statistical data. |
| Key words: statistical data spatialization error correction error analysis grain yield |
