东南大学学报(自然科学版)
東南大學學報(自然科學版)
동남대학학보(자연과학판)
JOURNAL OF SOUTHEAST UNIVERSITY
2013年
z2期
265-268
,共4页
戚浩平%杨波%路云阁%杨清华%方洪宾
慼浩平%楊波%路雲閣%楊清華%方洪賓
척호평%양파%로운각%양청화%방홍빈
青藏高原%样条拟合%克里金插值%反距离加权插值%模型插值
青藏高原%樣條擬閤%剋裏金插值%反距離加權插值%模型插值
청장고원%양조의합%극리금삽치%반거리가권삽치%모형삽치
Qinghai-Tibetan Plateau%spline fitting%Kriging interpolation%inverse distance weighted interpolation%model interpolation
利用空间插值技术研究了青藏高原的温度变化,在已有常规插值方法基础上,考虑海拔、经度和纬度等因素,提出了一种“模型拟合+残差修正”的综合插值方法.首先,建立温度与海拔、经度和纬度的回归模型,该模型通过了显著性水平为0.05的t检验;其次,根据模型回归值和站点实测值,得到实测点的残差;然后,选取一种常规插值方法,对残差进行插值;最后,把残差插值结果修正到模型回归图上,得到最终的温度插值图.通过常规插值方法和综合插值方法的对比发现,在所选的14个检验站点中,常规插值方法的中误差都大于3.0℃,而综合插值方法都小于3.0℃,其中“模型拟合+残差修正”的综合插值方法的中误差最小可达到1.26℃,相比常规插值方法的中误差至少提高了2倍,表明综合插值方法对青藏高原地区温度插值有更好的适用性,插值精度更高.
利用空間插值技術研究瞭青藏高原的溫度變化,在已有常規插值方法基礎上,攷慮海拔、經度和緯度等因素,提齣瞭一種“模型擬閤+殘差脩正”的綜閤插值方法.首先,建立溫度與海拔、經度和緯度的迴歸模型,該模型通過瞭顯著性水平為0.05的t檢驗;其次,根據模型迴歸值和站點實測值,得到實測點的殘差;然後,選取一種常規插值方法,對殘差進行插值;最後,把殘差插值結果脩正到模型迴歸圖上,得到最終的溫度插值圖.通過常規插值方法和綜閤插值方法的對比髮現,在所選的14箇檢驗站點中,常規插值方法的中誤差都大于3.0℃,而綜閤插值方法都小于3.0℃,其中“模型擬閤+殘差脩正”的綜閤插值方法的中誤差最小可達到1.26℃,相比常規插值方法的中誤差至少提高瞭2倍,錶明綜閤插值方法對青藏高原地區溫度插值有更好的適用性,插值精度更高.
이용공간삽치기술연구료청장고원적온도변화,재이유상규삽치방법기출상,고필해발、경도화위도등인소,제출료일충“모형의합+잔차수정”적종합삽치방법.수선,건립온도여해발、경도화위도적회귀모형,해모형통과료현저성수평위0.05적t검험;기차,근거모형회귀치화참점실측치,득도실측점적잔차;연후,선취일충상규삽치방법,대잔차진행삽치;최후,파잔차삽치결과수정도모형회귀도상,득도최종적온도삽치도.통과상규삽치방법화종합삽치방법적대비발현,재소선적14개검험참점중,상규삽치방법적중오차도대우3.0℃,이종합삽치방법도소우3.0℃,기중“모형의합+잔차수정”적종합삽치방법적중오차최소가체도1.26℃,상비상규삽치방법적중오차지소제고료2배,표명종합삽치방법대청장고원지구온도삽치유경호적괄용성,삽치정도경고.
The temperature change in the Qinghai-Tibetan Plateau is studied by using spatial interpola-tion technology.The altitude, longitude and latitude factors are taken into account based on the con-ventional interpolation method, and an integrated interpolation method for the temperature, named model fitting and residual correction interpolation method, is proposed.First, the regression model be-tween the temperature and the independent variables including altitude, longitude and latitude is built, and this model has passed t-significant test at 0.05.Then, the residuals calculated by regression value and measured value are itn erpolated by using the conventional interpolation method.Finally, the final interpolation temperature map is obtained by correcting the residuals interpolated results into model re-gression map.The experimental results demonstrate that, compared with the conventional method, the integrated interpolation method can obtain better interpolation results with a smaller root mean square error( RMSE) value.The RMSEs of the integrated interpolation method in the selected 14 test points are less than 3.0℃, while the RMSEs in the conventional method exceed 3.0℃.The smallest RMSE using the proposed method is only 1.26℃, which is 2 times better than that of the conventional meth-od.The results also show the proposed integrated interpolation method has a better applicability and higher accuracy of interpolation of temperature in the Qinghai-Tibetan Plateau.