The layout of county area cultivated land quality monitoring samples in Huanghua City based on spatial clustering stratified sampling
YANGJiangyan1,, YINShouqiang2, ZHANGLi1, MENMingxin1,, CHENYing1 1. Institute of Land and Resources, Agricultural University of Hebei, Baoding 071000, China2. China University of Mining, Beijing 100083, China 通讯作者:通讯作者: 门明新,E-mail:menminn@sina.com 收稿日期:2018-07-14 修回日期:2018-11-8 网络出版日期:2019-02-25 版权声明:2019《资源科学》编辑部《资源科学》编辑部 基金资助:国土资源部两套指标体系试点项目(20170411)河北省典型样带土地资源与生态环境监测评价项目(20170428) 作者简介: -->作者简介: 杨江燕,女,安徽安庆人,硕士生,研究方向为土地资源管理。E-mail:15931895996@163.com
关键词:耕地质量等别;监测样点布设;抽样误差;抽样效率;空间聚类分层抽样;黄骅市 Abstract Cultivated land quality monitoring is an important strategy to grasp the changes in cultivated land quality and productivity for the country. Arranging cultivated land quality monitoring samples in a reasonable way can greatly improve the efficiency of cultivated land quality monitoring. This study took the typical area of the coastal plain, Huanghua City as an example, generated 20 alternative monitoring area scenarios with the method of spatial clustering. The initial monitoring area plan was selected by comprehensive comparison of variance, sampling error, sampling efficiency, and sampling elastic coefficient, and by partial optimization, the project of the cultivated land quality monitoring area was finally generated. Finally, based on the stratified sampling area, the stratified sampling method was used to lay out the monitoring records of cultivated land quality.The results showed that under the requirement of 1% sampling error, the alternative cultivated land quality monitoring area with a partition number of 65 did exhibit a comprehensive sample capacity of 77 and a relatively high sampling efficiency, which was selected as the initial monitoring area. The difference in spatial position and cultivated land quality between cultivated land units in the same monitoring area decreased rapidly first, and finally remained stable, with the number of cultivated land quality monitoring areas increases from 5 to 100. By laying the same number of monitoring samples, the sampling error in terms of topsoil texture, section configuration, salinization, organic matter content, drainage conditions, irrigation conditions, and national natural index were 0.37, 1.02, 1.39, 0.91, 0.31, 1.53, and 1.27 respectively. Based on spatial clustering stratified sampling proposed in this paper, the observations were lower than traditional stratified sampling, simple random sampling, grid stratified sampling, and had higher sampling efficiency. The results are intended to provide effective guidance for the deployment of relevant work and research on the quality monitoring samples of cultivated land.
Keywords:cultivated land quality;monitoring sample layout;sampling error;sampling efficiency;spatial clustering stratified sampling;Huanghua City -->0 PDF (6070KB)元数据多维度评价相关文章收藏文章 本文引用格式导出EndNoteRisBibtex收藏本文--> 杨江燕, 殷守强, 张利, 门明新, 陈影. 基于空间聚类分层抽样的黄骅市县域耕地质量等别监测样点布设[J]. 资源科学, 2019, 41(2): 257-267 https://doi.org/10.18402/resci.2019.02.05 YANGJiangyan, YINShouqiang, ZHANGLi, MENMingxin, CHENYing. The layout of county area cultivated land quality monitoring samples in Huanghua City based on spatial clustering stratified sampling[J]. RESOURCES SCIENCE, 2019, 41(2): 257-267 https://doi.org/10.18402/resci.2019.02.05
从图1可以看出,各因素间的区内、区间方差有各自的变化特征。其中,盐渍化程度、剖面构型和土壤有机质随着分区数的不断增大,区内方差逐渐减少、区间方差逐渐增大,但是当分区数增大达到一定程度后,区内、区间方差趋于稳定。表明这3项因素随着监测区数目的不断增大,开始区内变异程度逐渐减少、区间变异程度逐渐增大,分区效果越来越好,但是当分区数达到一定程度时,分区效果变化不明显。灌溉保证率在监测区数目小于65时,区内、区间方差存在较大的波动,当监测区数目大于65后趋于稳定。国家耕地质量自然等指数、表土质地、排水条件从分区数为5到分区数为100的范围内,区间、区内方差均较为稳定,表明监测区数目的增加对改善这3项因素的分区效果并不明显。 显示原图|下载原图ZIP|生成PPT 图1不同监测区数目下各因素的的区内方差和区间方差 -->Figure 1Intra-zone and inter-zone variance of each factor in different numbers of monitoring zones -->
通过SPSS 19.0软件得到不同监测区数目下耕地单元质心坐标的区内、区间方差,用来表示不同分区数下耕地单元各项因素的空间分区效果。由图2可知,耕地单元质心坐标的区内方差明显小于区间方差,空间分区效果较好。随着监测区数目的增加,区间方差逐渐增加,区内方差逐渐减少,两者差距逐渐扩大,变化速度由快变慢。表明随着监测区数目的增加,区内单元的空间聚合度和区间单元的空间分离度逐渐增加,即监测区数目越多,空间分区效果越好,并且可以假设当监测区数目增加到一定值时,区内、区间方差趋于平稳,分区效果达到最好。 显示原图|下载原图ZIP|生成PPT 图2耕地单元质心坐标的区内方差和区间方差的变化曲线 -->Figure 2Changes of the intra-zone variance and inter-class variance of space coordinates of cultivated land in different layers -->
通过以上公式(3)—公式(10),可以得到不同备选监测区方案中各抽样误差对应的综合样本量,并用下划线标注出每一列的最小值,作为该抽样误差对应的最小综合样本量,具体见表1。从结果可看出,相同备选方案中,随着抽样误差的增加,综合样本量随之减少,减小速度逐渐变慢,最后逐渐接近备选方案的分区数。当抽样误差在0.1%~1.0%的范围内,抽样误差相同时,最小样本量随分区数的增加整体上逐渐减小,但幅度越来越小;当抽样误差在1.5%~10.0%的范围内,抽样误差一定时,综合样本量随着分区数增加先减小后增加,逐渐与分区数相同。 Table 1 表1 表1黄骅市不同备选监测区方案中各抽样误差对应的综合样本量 Table 1Comprehensive sample sizes corresponding to different sampling errors in different alternative schemes of monitoring zones in Huanghua
分区数/个
0.1%
0.2%
0.3%
0.4%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
3.5%
4.0%
4.5%
5.0%
6.0%
7.0%
8.0%
9.0%
10.0%
5
14 001
11 422
9 312
7 554
6 271
2 596
1 314
777
511
361
267
206
164
133
94
70
55
43
36
10
12 086
9 634
8 051
6 643
5 441
2 195
1 102
651
425
299
222
170
135
110
75
56
44
34
28
15
11 890
9 649
8 142
6 779
5 590
2 282
1 151
680
447
313
230
180
142
116
78
60
47
38
29
20
9 116
5 764
3 881
2 813
2 118
699
332
192
127
88
69
53
49
38
31
25
23
22
21
25
8 960
5 717
3 679
2 491
1 777
530
245
141
93
68
51
41
36
31
29
26
26
25
25
30
5 775
2 981
1 677
1 041
703
194
89
54
42
34
32
30
30
30
30
30
30
30
30
35
5 602
2 964
1 686
1 058
718
199
97
61
47
38
37
35
35
35
35
35
35
35
35
40
4 272
2 065
1 138
702
475
140
73
52
44
42
40
40
40
40
40
40
40
40
40
45
3 670
1 783
972
600
408
124
56
55
49
46
45
45
45
45
45
45
45
45
45
50
2 798
1 319
719
446
305
105
65
55
51
51
50
50
50
50
50
50
50
50
50
55
2 599
1 215
665
416
287
100
66
59
56
55
55
55
55
55
55
55
55
55
55
60
2 354
1 022
543
337
236
90
68
61
60
60
60
60
60
60
60
60
60
60
60
65
1 465
609
327
213
155
77
67
65
65
65
65
65
65
65
65
65
65
65
65
70
2 180
833
425
261
188
83
72
70
70
70
70
70
70
70
70
70
70
70
70
75
1 517
599
323
211
155
83
75
75
75
75
75
75
75
75
75
75
75
75
75
80
1 728
710
383
250
182
91
82
80
80
80
80
80
80
80
80
80
80
80
80
85
1 413
562
309
208
157
92
85
85
85
85
85
85
85
85
85
85
85
85
85
90
1 026
425
246
173
137
94
91
90
90
90
90
90
90
90
90
90
90
90
90
95
1 601
729
412
275
206
112
97
95
95
95
95
95
95
95
95
95
95
95
95
100
1 265
509
289
201
158
107
100
100
100
100
100
100
100
100
100
100
100
100
100
注:每列带下划线加粗数字是在该误差范围下的最小综合样本量,如第一列带下划线加粗的1026表示在各因素的抽样误差为0.1%时的最小综合样本量为1026,由于随着抽样误差的不断增大,与之相对应的综合样本量及其变化均越来越小,因此抽样误差的节点选取0.1%~0.5%以0.1%为间隔,0.5%~5.0%以0.5%为间隔,5.0%~10.0%以1%为间隔。 新窗口打开 基于结果分析可知,分区数、抽样误差和综合样本量三者相互影响,确定最合适的监测区方案,需综合考虑它们之间的变动关系。使用公式(10)、公式(11)可进一步计算出抽样效率和抽样弹性系数随着最小综合样本量的变动曲线。从图3结果可以看出,当综合样本量≤77时,抽样误差弹性系数绝大部分大于1,抽样误差的相对变化率大于样本容量的相对变化率,抽样效率增大;同理,当综合样本量在77~137之间时,抽样效率先增加后减少;综合样本量≥137时,抽样误差弹性系数小于1,抽样效率减小。由分析可知,当抽样误差弹性系数为1时,抽样误差的相对变化率等于样本容量的相对变化率,此时该抽样方案的抽样效率最高,与此相对应,综合样本量在77~137之间,分区数在65~90之间,抽样误差在1.0%和1.5%之间。 显示原图|下载原图ZIP|生成PPT 图3抽样效率和抽样误差弹性系数随着最小综合样本容量变化曲线 -->Figure 3Sampling efficiency and change in elastic coefficient of sampling error corresponding under different minimum comprehensive sample sizes -->
为了增强样点布设的代表性,把各耕地单元分等因素值组合在一起,形成耕地单元的分等因素组合类型。在初始监测区方案中,虽然大部分监测区内组合类型基本相同,且连片分布,但仍有部分监测区内耕地单元的组合类型存在较大差异或空间距离较远的问题,以下针对几种主要不同的情况提出相应的优化方案: (1)部分同一初始监测区内的耕地单元仍处于相距较远的几片区域,区内耕地单元的空间聚集性较差。针对这种情况,把处于同一个初始监测区内相隔较远的耕地单元划为多个监测区; (2)部分初始监测区内的耕地单元尽管在空间上连片分布,但分属不同的分等因素组合类型,且面积均较大。针对这种情况,把处于同一个初始监测区内不同因素组合类型的耕地划分为分等因素组合类型基本相同的多个监测区; (3)将零散分布或面积较小的组合类型耕地监测区与邻近的监测区合并。这虽会在一定程度上增加抽样误差,但改善了耕地质量等别监测区的空间连续性,并极大地减少了耕地质量等别监测区的数量,节约监测成本。 基于以上方法在初始监测区上优化得到了最终的黄骅市监测区方案,运用SPSS 19.0软件计算优化前后监测区方案中各因素的区内方差及其占总方差的比例,从图4结果可以发现,绝大部分因素的区内方差在经过优化后小于优化前,尽管在优化监测区中某些因素的分区效果变差,但变化幅度不大,各因素的区内方差占总方差的比例均在15%以下,表明优化方案比初始方案在确保相同监测区内耕地单元的质量状况一致性和空间分布连续性两个方面更具综合优势,分区效果更好。 显示原图|下载原图ZIP|生成PPT 图4黄骅市不同监测区方案的区内方差及其所占总方差的比例 -->Figure 4Intra-class variance and the proportion accounting for the total variance of different factors in two monitoring zoning schemes in Huanghua -->
3.4 耕地质量等别监测样点布设
3.4.1 与其他监测样点布设方法的比较 文章以优化监测区为分层依据,采用分层抽样的方法抽取监测单元,共布设了81个监测样点。由于在以往的样点布设研究中,经常采用简单随机抽样方法、等别分层抽样方法和网格分层抽样等3种传统抽样方法布设样点,为了与空间聚类分层抽样结果比较,此处同样采用这3种抽样方法进行样点布设,分别计算它们的抽样误差。等别分层抽样和网格分层抽样误差计算方法工时参考公式(3)-(5),简单随机抽样又称纯随机抽样,是从总体中随机抽取某一数量的样本,每个样本被抽中的概率相等。其方差计算公式[8]为: (12) 式中: 为抽样均值;V为 的方差;f为抽样比;n样本容量; 为总体方差。 从计算结果可以看出(表2),当样本容量均为81时,采用空间聚类分层抽样法布设样点的各质量属性抽样误差均明显小于其他3项,使样点在分等因素和耕地自然等指数方面更具有代表性,具有较好的布样效果。网格分层抽样抽样误差较大是因为该方法仅考虑了耕地单元的空间位置,对耕地单元的其他属性考虑较少,导致抽取的监测单元的质量属性代表性较差;国家耕地质量自然等别指数的等别分层抽样误差比其他抽样方法低,虽然抽取的监测单元对耕地自然等指数的代表性较好,但在分区时没有考虑各项分等因素,导致各项分等因素的抽样误差较大;等别分层抽样和简单随机抽样由于缺乏对耕地单元的空间位置属性的考虑,导致抽取的监测单元的空间代表性较差。 Table 2 表2 表2黄骅市不同抽样方法各因素的抽样误差的比较 Table 2Comparison of the sampling error of each factor between different sampling methods in Huanghua
抽样方法
表土质地/%
剖面构型/%
盐渍化/%
有机质含量/%
排水条件/%
灌溉条件/%
耕地质量国家自然等指数/%
空间聚类分层抽样
0.37
1.02
1.39
0.91
0.31
1.53
1.27
等别分层抽样
0.97
2.41
4.28
4.54
1.28
2.04
1.06
简单随机抽样
1.45
3.00
4.23
4.16
1.17
6.01
1.89
网格分层抽样
1.23
2.73
4.69
3.06
0.49
7.69
2.16
新窗口打开 由于文章选取的研究区耕地等别为7、8、9、10、11等,其中面积以10等分布最广。从表3结果中可以看出,空间聚类分层抽样布设的样点分布涵盖了所有的耕地等别,并且整体上数量和相应的等别面积成比例,较好的考虑到了耕地的等别控制因素;简单随机抽样缺乏对8等别耕地的考虑,等别分层抽样和网格分层抽样虽在样点布设上兼顾到了各等别的耕地,但在抽样误差、空间位置以及其他属性等方面的考虑却不够全面。 Table 3 表3 表3黄骅市不同抽样方法布设耕地质量监测样点数量在各耕地等别中的分布 Table 3Distribution of cultivated land quality monitoring sample number by different sampling methods in different cultivated fields in Huanghua
耕地等别
耕地面积/km2
空间聚类分层抽样样点 布设数量/个
等别分层抽样样点 布设数量/个
简单随机抽样样点 布设数量/个
网格分层抽样样点 布设数量/个
10
593.45
55
42
63
50
11
69.94
10
13
12
10
7
21.71
5
5
1
1
8
8.14
1
3
0
4
9
176.41
10
8
5
16
总计
869.64
81
81
81
81
新窗口打开 3.4.2 耕地质量等别监测样点布设结果 按照监测样点选取的标准以及出于监测单元稳定性的考虑,最终在研究区共有7518个满足条件的耕地单元数。在81个耕地质量等别监测区内,只有1个监测区没有满足条件的耕地,需把普通耕地作为备选耕地质量等别监测单元,最终确定7521个耕地单元作为备选耕地质量等别监测单元。借助SPSS 19.0软件,在筛选出来的7521个备选耕地质量等别监测单元中,以耕地质量等别监测区作为分区依据,运用分层抽样方法,在每一区内随机抽取1个耕地单元作为耕地质量等别监测单元。监测样点位于监测单元的质心位置,最终共布设81个监测样点。其中,80个位于基本农田上,1个位于普通耕地上。耕地质量等别监测样点空间分布情况如图5所示。 显示原图|下载原图ZIP|生成PPT 图5黄骅市耕地质量等别监测优化监测区及耕地质量监测样点分布 -->Figure 5Distribution of the cultivated land quality monitoring optimized monitoring zones and points in Huanghua -->
4 结论
本文以滨海平原区的典型地区——河北省黄骅市为例,提出了一种基于空间聚类分层抽样的耕地质量等别监测样点布设方法,并与传统样点布设方法比较了抽样效率。主要结论为: (1)由于综合考虑了耕地各项质量属性和空间位置属性,本文采用的空间聚类算法对耕地质量等别监测区的划定效果较好。随着耕地质量等别监测区数目从5增加到100,相同监测区内耕地单元之间在空间位置和耕地质量等别上的差异程度先快速减少,最后保持平稳。 (2)分区数为65的备选耕地质量等别监测区的抽样效率相对较高,被选定为初始监测区方案,在抽样误差为1%的要求下,综合样本容量为77。与初始方案相比,优化后的耕地质量等别监测区方案在确保相同监测区内耕地质量一致性和空间分布连续性更具综合优势。 (3)本文提出的监测样点布设方法与等别分层抽样、简单随机抽样、网格分层抽样等传统抽样方法相比,具有较高的抽样效率,在布设相同数量的监测样点时,各项耕地质量因素的抽样误差均明显较低。 The authors have declared that no competing interests exist.
[Yang JY, Yue YL, Song HR, et al. Sampling and optimizing methods of cultivated land quality based on spatial simulated annealing algorithm [J]. Transactions of the Chinese Society of Agricultural Engineering, 2015, 31(20): 253-261. ] [本文引用: 1]
[2]
HuffmanE, Eilers RG, PadburyG, et al. Canadian agri-environmental indicators related to land quality: Integrating census and biophysical data to estimate soil cover, wind erosion and soil salinity [J]. Agriculture, Ecosystems and Environment, 2000, 81(2): 113-123. [本文引用: 1]
[3]
Doran JW.Soil health and global sustainability: Translating science into practice [J]. Agriculture, Ecosystems and Environment, 2002, 88(2): 119-127. [本文引用: 1]
[Jiang GH, Zhao TT, Duan ZQ, et al. Cultivated land quality change and its future trend modeling in Beijing mountainous area [J]. Transactions of the Chinese Society of Agricultural Engineering, 2010, 26(10):304-311. ] [本文引用: 1]
[Sun YB, Wu KN, Hu XT, et al. Layout method for monitoring quality level of arable land based on combination of potential index [J]. Transactions of the Chinese Society of Agricultural Engineering, 2013, 29(4): 245-254. ] [本文引用: 2]
[Yin SQ, WangX, He WL, et al. Comparison of precision among different sampling methods in cultivated land quality monitoring: A case study of Huanghua City in Hebei Province, China [J]. Resources Science, 2016, 38(11): 2049-2057. ] [本文引用: 2]
[Han ZW, HuangW, LuoY, et al. Application of simulated annealing method and neural network on optimizing soil sampling schemes based on road distribution [J]. Chinese Journal of Applied Ecology, 2015, 26(3): 891-900. ] [本文引用: 1]
[Yu SQ, Zhang BB, XiangH, et al. Layout method for monitoring sample point of arable land quality level based on combination of factors [J]. Transactions of the Chinese Society of Agricultural Engineering, 2014, 30(24):288-297. ] [本文引用: 1]
[Guo LN, Zhang FR, Ma RH, et al. Methods for setting up farmland grade monitoring points from standard plots in Hebei, Henan and Hubei provinces [J]. Resources Science, 2009, 31(11): 1957-1966. ] [本文引用: 3]
[GuoS, Wei MH, JianQ, et al. Study on the distribution of farmland quality monitoring samples in counties-Taking Changli County, Hebei province as an example [J]. Jiangsu Agricultural Science, 2018, 46(12): 227-232. ] [本文引用: 2]
[Hu XT, Wu KN, Ma JH, et al. Control points for quality of arable land quality monitoring in Daxing District, Beijing [J]. Resources Science, 2012, 34(10): 1891-1897. ] [本文引用: 2]
[Yang JY, TangS, Yun WJ, et al. Sampling method for monitoring classification of cultivated land in county area based on Kriging estimation error [J]. Transactions of the Chinese Society of Agricultural Engineering, 2013, (9): 223-230. ] [本文引用: 1]
[WangQ, Shang YM, FengR, et al. Study on location sampling for monitoring the quality of arable land based on variation functions: Case studies in Zhongjiang County and Daxing District [J]. China Land Sciences, 2012, 26(8):80-86. ] [本文引用: 1]
[Ma JH.Based on variograms and other types study on the layout of cultivated land quality monitoring samples [J]. World Agriculture, 2015, (5):111-116. ] [本文引用: 1]
[Yang JY, Yue YL, Song HR, et al. Sampling distribution method for monitoring quality of arable land in county area based on spatial balanced [J]. Transactions of the Chinese Society of Agricultural Engineering, 2015, 31(24): 274-280. ] [本文引用: 1]
[Zhu JX, Xu GB, Zhang LY.Optimization layout method of monitoring sample points of cultivated land quality based on semi-variance analysis and grade combination [J]. Transactions of the Chinese Society of Agricultural Engineering, 2015, 31(19): 254-261. ] [本文引用: 1]
[Zhang YZ, Kong XB, LiuY, et al. Layout method of monitoring samples of cultivated land at the provincial level based on standard plots in Inner Mongolia [J]. Resources Science, 2016, 38(11): 2037-2048. ] [本文引用: 1]
[Guo LN, Ma HR, Xu DR, et al. Method of setting up farmland grade monitoring belts in Hebei, Henan and Hubei [J]. Resources Science, 2008, 30(8): 1199-1205. ] [本文引用: 1]
[NieY, Wu XZ, YuJ, et al. Primary farmland zoning based on land evaluation and spatial clustering: A case of Hefeng County [J]. China Land Sciences, 2013, (12): 39-45. ] [本文引用: 1]
[WuY, Wu XM, HeY, et al. Demarcation method of prime farmland based on productivity and spatial clustering [J]. Resources and Environment in the Yangtze Basin, 2014, 23(6): 809-815. ] [本文引用: 1]
[Hao XY, Pan YC, Tang XM, et al. Land leveling partitioning of farming area in arid plain based on spatial clustering [J]. Transactions of the Chinese Society of Agricultural Engineering, 2015, 31(5): 301-307. ] [本文引用: 1]
[GaoF, Fan MH, Chen CC, et al. The determination of lunar gravity field by mass-point model and its character analysis [J]. Science of Surveying and Mapping, 2008, 33(1): 45-47. ] [本文引用: 1]
[Che LM, XiaoY, Wang SY, et al. Application of Kmeans clustering analysis in Chinese pronunciation degree [J]. Computer Technology and Development, 2011, 21(2): 223-225. ] [本文引用: 1]
[Liu WD, Zhong WZ, ShiQ.Forecast of China’s total enemy consumption in 2020 based on method of fixed based enemy consumption elasticity coefficient [J]. Resources Science, 2016, 38(4): 658-664. ] [本文引用: 1]