地理科学进展
地理科學進展
지이과학진전
PROGRESS IN GEOGRAPHY
2010年
3期
279-286
,共8页
杨琳%朱阿兴%秦承志%李宝林%裴韬%邱维理%徐志刚
楊琳%硃阿興%秦承誌%李寶林%裴韜%邱維理%徐誌剛
양림%주아흥%진승지%리보림%배도%구유리%서지강
采样%典型点%模糊聚类%环境因子组合%土壤属性制图%嫩江
採樣%典型點%模糊聚類%環境因子組閤%土壤屬性製圖%嫩江
채양%전형점%모호취류%배경인자조합%토양속성제도%눈강
sampling%typical points%fuzzy clustering%environmental combination%soil property mapping%Nenjiang
鉴于经典采样和空间采样存在的局限性,提出了一种旨在寻找典型点的目的性采样设计方法.该方法通过分析与目标地理要素空间分布具有协同变化关系的环境因子,提取地理要素空间变化的典型模式,进而在典型模式上布设样点,即可获得典型点,从而减少所需样本量.以位于黑龙江鹤山农场的2个研究区为例,分别选择土壤厚度和表层有机质2个土壤属性,通过对土壤属性空间变化的4个协同环境因子进行模糊c均值聚类,获得对应土壤属性空间变化模式的环境因子组合;根据其模糊隶属度结果设计典型点并进行采样,最后结合典型点的属性值与环境因子组合模糊隶属度结果,采用加权平均模型得到土壤属性空间分布图,反映了土壤属性随地形变化的连续性分布.基于独立野外验证点,选择了4个评价指标对所得属性图进行定量评价.结果表明:2个研究区验证点集的预测值和观测值一致性指数均较高,可见本研究提出的方法是一种有效的样点布设方法.研究还对在每一环境组合类设计不同数量典型点所得土壤属性制图结果的差异进行了讨论,认为典型点增多并不一定能提高土壤属性空间推测的精度.
鑒于經典採樣和空間採樣存在的跼限性,提齣瞭一種旨在尋找典型點的目的性採樣設計方法.該方法通過分析與目標地理要素空間分佈具有協同變化關繫的環境因子,提取地理要素空間變化的典型模式,進而在典型模式上佈設樣點,即可穫得典型點,從而減少所需樣本量.以位于黑龍江鶴山農場的2箇研究區為例,分彆選擇土壤厚度和錶層有機質2箇土壤屬性,通過對土壤屬性空間變化的4箇協同環境因子進行模糊c均值聚類,穫得對應土壤屬性空間變化模式的環境因子組閤;根據其模糊隸屬度結果設計典型點併進行採樣,最後結閤典型點的屬性值與環境因子組閤模糊隸屬度結果,採用加權平均模型得到土壤屬性空間分佈圖,反映瞭土壤屬性隨地形變化的連續性分佈.基于獨立野外驗證點,選擇瞭4箇評價指標對所得屬性圖進行定量評價.結果錶明:2箇研究區驗證點集的預測值和觀測值一緻性指數均較高,可見本研究提齣的方法是一種有效的樣點佈設方法.研究還對在每一環境組閤類設計不同數量典型點所得土壤屬性製圖結果的差異進行瞭討論,認為典型點增多併不一定能提高土壤屬性空間推測的精度.
감우경전채양화공간채양존재적국한성,제출료일충지재심조전형점적목적성채양설계방법.해방법통과분석여목표지리요소공간분포구유협동변화관계적배경인자,제취지리요소공간변화적전형모식,진이재전형모식상포설양점,즉가획득전형점,종이감소소수양본량.이위우흑룡강학산농장적2개연구구위례,분별선택토양후도화표층유궤질2개토양속성,통과대토양속성공간변화적4개협동배경인자진행모호c균치취류,획득대응토양속성공간변화모식적배경인자조합;근거기모호대속도결과설계전형점병진행채양,최후결합전형점적속성치여배경인자조합모호대속도결과,채용가권평균모형득도토양속성공간분포도,반영료토양속성수지형변화적련속성분포.기우독립야외험증점,선택료4개평개지표대소득속성도진행정량평개.결과표명:2개연구구험증점집적예측치화관측치일치성지수균교고,가견본연구제출적방법시일충유효적양점포설방법.연구환대재매일배경조합류설계불동수량전형점소득토양속성제도결과적차이진행료토론,인위전형점증다병불일정능제고토양속성공간추측적정도.
In consideration of limitations of classical sampling and spatial sampling, this paper proposed a purposive sampling method based on typical points. This method employed environmental factors which co-vary the target geographical element to find typical patterns of the geographical element over space. Field samples were then designed based on the locations of these typical patterns. It is believed that these field samples capture the overall pattern of spatial variation of the target variable well. With this approach, the number of field samples needed to characterize the spatial pattern of the target variable was reduced. The approach was applied in two study areas in Heshan Farm in Heilongjiang Province of China. Two soil properties were chosen, i.e. A-horizon organic matter and soil thickness (each property for one study area). Environmental combinations were generated by a fuzzy c-means clustering on four local environmental data layers and typical points were then designed with help of fuzzy membership maps of environmental combinations. Spatial variation of the two soil properties was inferred using a linear weighted average model with typical points' soil property values and fuzzy membership maps of environmental combinations. The mapping results reflected continuous changing of soil properties with terrain changing. Four indices were set up for evaluation of mapping results by using independently validation points. The evaluation results showed that, the agreement coefficients between observed values and predicted values of validation points were high for both of the study areas. We then concluded that the proposed sampling design approach was effective. Analysis was also conducted on difference between soil property mapping results by using different amount of typical points when different number of typical points representing each environmental combination. It was showed that using more points would not necessarily result in a more accurate map.