Microsoft Word 刘付程_news_.doc

Size: px

Start display at page:

Download "Microsoft Word 刘付程_news_.doc"

板郊锺
5 years ago
Views:

1 基于对数比转换方法的沉积物粒级组分空间预测与底质类型制图刘付程 2, 彭俊 (., ; 2., ) 摘要 : 以废黄河三角洲表层粒度分析数据为基础, 探讨了对数比转换和 krigig 插值相结合方法在沉积物粒级组分空间预测和底质类型制图中的应用结果表明, 基于沉积物粒级组分原始数据的 krigig 预测方法难以保证各组分预测结果的非负和定和要求, 因而预测结果的可信度低 ; 而对数比转换 krigig 方法不但满足非负和定和要求, 而且还有着更优的组分预测结果和较高的底质类型制图精度新方法对于开展定量化的沉积物粒级组分预测和底质类型制图具有参考价值关键词 : 对数比转换 ; 沉积物粒级组分 ; 空间预测 ; 底质制图中图分类号 : P736.2 文献标识码 : A 文章编号 : (205) doi: 0.759/hykx , [],, ( ) [2],, Vorooi krigig [3-4],,,,, [5], 0 [6-7], [6-9], [7],,,,, [0] Aitchiso [8],, [8-9],, krigig,, 材料与方法. 区域概况与数据 28~855 : ; : : ( ); (Z20407) : (97-),,,,, GIS, : , iliufucheg@26.com 22 / 205 / 39 / 6

2 , 855,, 0.25 m/s, NNW-SSE 995~2005, ENE, NE [] ~8, 3,, 5~6 km 3 88 ( ) CoulterLS-00Q, (Φ ~4) (Φ 4~8) (Φ>8) 88, 50, ; 33 ( ) Fig. The spatial distributio of samplig sites.2 对数比转换方法 (additive log-ratio) (cetered log-ratio) (isometric log-ratio) [8, 2], 0 ( 0 ), (modified additive log-ratio) [0], 0,, : : x ij y ij xij + h j = l æ c ö ( xij h j ) + ç çèj = ø / c é ù ú ê exp y h úæ ö = - + c ij j ú h c c úç ç j å j= exp yij h úèç ø å + å j ê j= j= ú ë û () (2), x ij i j ; y ij i j ; η j, 0 ; c.3 空间预测效果评价,, (mea absolute error, MAE) (root mea square error, RMSE) (idex of agreemet, d) [3] : MAE = å Zi-Xi (3) i = RMSE = å ( Z ) 2 i-xi (4) i = é ù d = - Z - X Z + X, 0 d êë i= i= úû, ; X i Z i 2 å( ) ( ) 2 i i å ú i i (5) i ; Z i = Z i X, X i = X i X, X,, 3 00% Marie Scieces / Vol. 39, No. 6 /

3 2 结果与分析 2. 沉积物不同粒级组分含量的描述性统计,, 55.86%,, 3,.26, ;,, 2.2 沉积物粒级组分含量的空间分布预测,, krigig [4] Krigig, [5] GS+7.0, 2 ( 2a, b, c) 表沉积物粒级组分含量的统计特征 Tab. Statistical characteristics of differet grai size compositios of the samplig sedimets (%) Fig. 2 The fitted models for smivariograms of sedimet grai size compositios based o data processed with differet methods 2 ( 2d, e, f)( ), [5],,,, ( 24 / 205 / 39 / 6

4 ),, 2g, h 2,, [5], ArcGIS krigig, 30 m 30 m, ArcGIS,, (2), krigig ( 3a, b, c); krigig, krigig ( 3g, h) 3d, e, f krigig 3,, ; ;, krigig 3, 3a, b, c, krigig, ( 3d, e, f ) krigig ( ),, ( 3g, h ) 2.3 沉积物粒级组分含量的预测结果评价 2, (3)~ (5) MAE RMSE d 2, MAE RMSE d, 3 MAE RMSE krigig < krigig < krigig, d, Fig. 3 3 Predictio maps of sedimet grai size compositios based o differet data processig methods Marie Scieces / Vol. 39, No. 6 /

表 2 不同数据处理方法下的预测结果准确性评价 Tab. 2 Evaluatio of predictio accuracy at validatio samplig statios based o differet data processig methods MAE(%) RMSE(%) d 9.30 7.50 6.90 3.52 0.02 8.39 0.858 0.837 0.83 9.

5 表 2 不同数据处理方法下的预测结果准确性评价 Tab. 2 Evaluatio of predictio accuracy at validatio samplig statios based o differet data processig methods MAE(%) RMSE(%) d krigig, 3, 3 4 3, krigig krigig 4, krigig 00% ; 0%, 3 00%,, 2.4 基于沉积物粒级组分预测结果的底质类型制图 krigig,, Shepard, ArcGIS, ( 5), 7,,, 75%, 5 m ;, %, ; 9%, ;,, 88, 74, 92.6%, 35 39, 97.8% 78.0%, 3 结论 (), krigig,, Fig. 4 4 Spatial distributio of the sum of predicted results of sad, slit ad clay 26 / 205 / 39 / 6

5 Fig. 5 Sedimet type map of the coast of the abadoed Yellow River Delta krigig,,, (2) krigig,, 7,,, 75%, 5 m ;,, 92.6%, : []. [J]., 992, 2(4): -20. [2],. [J]., 2007, 27(5): 889-895. [3],,,.

6 5 Fig. 5 Sedimet type map of the coast of the abadoed Yellow River Delta krigig,,, (2) krigig,, 7,,, 75%, 5 m ;,, 92.6%, : []. [J]., 992, 2(4): -20. [2],. [J]., 2007, 27(5): [3],,,. Vorooi [J]., 20, 34(2): [4],,,. [J]., 20, 30(5): [5] Jerosch K. Geostatistical mappig ad spatial variability of surficial sedimet types o the Beaufort Shelf based o grai size data[j]. Joural of Marie System, 202, doi: 0.06/j.jmarsys [6] Aitchiso J. The statistical aalysis of compositioal data[j].j Royal Stat Soc B, 982, 44(2): [7]. [J]., 996, 4( ): [8] Aitchiso J. The statistical aalysis of compositioal data[m]. Lodo: Chapma ad Hall, 986: [9]. [J]., 998, 23(2): [0] Odeh I O A, Todd A J, Triatafilis J. Spatial predictio of soil particle-size fractios as compositioal data[j]. Soil Sciece, 2003, 68 (7): []. [D]. :, 20. [2] Egozcue J J, Pawlowsky-Glah V, Mateu-Figueras G, et al. Isometric logratio trasformatios for compositioal data aalysis[j]. Mathmatical Geology, 2003, 35(3): [3] Willomtt C J. Some commets o the evaluatio of model performace[j]. Bulleti America Meteorological Society, 982, 63(): [4],,. [J]., 200, 34(7): [5]. [M]. :, 999: Marie Scieces / Vol. 39, No. 6 /

7 A logratio trasformatio based krigig method for spatial predictio of sedimet grai size compositios ad sedimet type mappig LIU Fu-cheg, PENG Ju 2 (. School of Geodesy ad Geomatics Egieerig, Huaihai Istitute of Techology, Liayugag , Chia; 2. School of Urba ad Resources Eviromet, Yacheg Teachers Uiversity, Yacheg , Chia) Received: May, 29, 203 Key words: logratio trasformatio; sedimet grai size compositio; spatial predictio; sedimet type mappig Abstract: The sedimet grai size data are compositioal data ad are characterized by o-egative ad costat sum. I spatial predictio of grai size compositios, oe importat aspect of the predictio quality is whether the predictio results meet the requiremets of o-egative ad costat sum, which is also the essetial coditio for sedimet type recogitio ad mappig. I this paper, the grai size data obtaied from the abadoed Yellow River Estuary are take as a example to discuss the applicatio of combied logratio trasform ad ordiary krigig i spatial predictio of grai size compositios ad sedimet type mappig. Results show that the krigig iterpolatio results by directly usig the grai size data are ureliable owig to their dissatisfactio of o-egative ad costat sum, while the predictio results obtaied usig the logratio-trasformed grai size data ad krigig iterpolatio method ot oly meet those two requiremets, but also have a better predictio accuracy of grai size compositios ad a relative high sedimet type mappig precisio. So the ew method of combied logratio trasform ad ordiary krigig has referece value for quatitative sedimet mappig. ( 本文编辑 : 刘珊珊 ) 28 / 205 / 39 / 6

ttian

ttian á è è é à ú á óè á ú ù ù úú á é é á à ì è í ò á ù à è è ó ù ò é é ó íú ì à ù ù ì ì ò á ó á é ú ú è à à à ù é ú é ì ì à í ú ú ú à à á í é é í è é é ú éè ù á á ù á ó ú à ì ú á à ó è á úú á á ú à á è