Scale effects of landscape research in Kerqin Sandy Land

There is highest heterogeneous on sandy land landscape due to the faci le changing of topography and land cover in Kerqin Sandy Land. The main causes, which induce the landscape changing, include unsuitable human economic activity, precipitation fluctuation, and wind erosion under huge Quaternary sand sediment condition in this area. Therefore, what are relation between the landscape char acters and research scales in different levels is a principal threshold for the landscape study. According to studied findings, the implications of scale effect can be concluded as below.; First, the scale effect on landscape research in Kerqin Sandy Land was addresse d in this article from regional sampling scale and mathematically analyzing scal e with GIS technique and lacunarity index analysis. Analyses showed that the tre nd and scope of sampling scale effect could be described statistically at both p atch and landscape scales. Relationship between sampling area(y) and patch n umbe r (x 1) was expressed as y=-7.1282x220.31(R=0.9580.708=α 0.01(10) ） , while relationship between the maximum patch area( x2) and sampling area (y) could be presented as y=0.004x2-0.0096x0.3346(R=0.9470.708=α 0.01(10) ） . At landscape scale, the tre nd of landscape elements, such as landscape diversity index and landscape domina nt index, were slightly fluctuated corresponding to the change of sampling scale . The association curve of sampling area and diversity index was a parabola with the diversity index peaked at 1.931 when the sampling area was half of the stud ied area. The averaged value of the diversity indices was 1.863 (±0.075) at all scales, which were divided into eleven levels when the minimum sampling scale w as 28% of the total studied area. The association curve of the sampling area and the dominant index was a reverse parabola and the dominant index was at 0.674 a s the lowest when the sampling area was half of the studied area. The average va lue of the dominant indices was 0.534 (±0.075) in this treatment. Second, findings also pointed out that different sampling area under landscape m atrix was a primary condition for landscape research when the spatial analysis, specifically Arc-info analysis was employed. Otherwise, the consistency of samp l ing rule would be changed. Meanwhile, results demonstrated that the maximum patc h type was changed from mobile sand-dune to cropland when the sampling area was less than 28% of the total studied area. Therefore, it could be possible to scal ing down or up for landscape spatial attributes analysis under properly defined conditions when sampling area was greater than 28% of the total area. Third, the lacunarity analysis showed that sandy land landscape had fractal stru cture in Kerqin Sandy Land when the sampling area was set fixed. The lacunarity index was closely related to the occupying probability of research patches. The higher the probability was, the higher the fractal dimension was. Patch distribu tion was corresponded to the lacunarity change when the patch occupation was set unchanged; Comparatively, the change of actual lacunarity of the sandy land lan dscape was closer to a cluster distribution and deviated from the theoretical ev en distribution pattern.