17hydrological stations located in the Taizi River basin were selected to investigate the relationship between landscape patterns and runoff using Geographic Information System(GIS).Based on landscape patterns(landscape-level and class-level),landscape pattern index analysis and statistical methods were used to analyze the effects of landscape patterns on runoff.Results showed that the influence of landscape patterns on runoff was significant.At landscape level,the mean fractal dimension index(FRAC_MN),the median contiguity index(CONTIG_MD)and the interspersion and juxtaposition index(IJI)were the main pattern indices effectively affecting the runoff change.At class level,the patch diversity of dry land,the mean fractal dimension index(FRAC_MN)and the cohesion(COHESION)of dry land and residential construction land,the perimeter-area ratio variation coefficient(PARA_CV)and aggregation index(AI)of forest land and dry land,the median contiguity index(CONTIG_MD)and interspersion and juxtaposition index(IJI)of residential construction land were the main landscape pattern indices controlling the variation of runoff.Furthermore,the perimeter-area ratio(PARA_CV)and the aggregation index(AI)of forest land were positively related to the change of runoff.

The spatial pattern of geographical objects has been one of the most important issues in the geomorpho-logical research. Two major themes of it, the organizational structure and characteristic scale range of spatial ob-ject, play significant roles in the geo -surface processing and mechanism modeling. However, these problems have not yet been properly solved up to now. This is an exploring research on the issues proposed above. Feature lines, which mainly include ridge lines and hollow lines of terrain, are controlling factor of the geomorphologic forms. In recent years,with the development of the nonlinear science, fractal theory propose possible methods for depicting the spatial pattern of watershed feature lines. However, different spatial pattern of feature lines may be evaluated for the same fractal dimension value. Thus, it is not sufficient to depict spatial pattern only by using the fractal di-mension indicator. The lacunarity analysis method, which developed on the basis of fractal theory, surmounts the weakness of the fractal indicator. Previous research has confirmed that one lacunarity curve correspond to one kind of spatial pattern uniquely. Besides, the organizational structure and the multi - scale characteristic of spatial ob-jects can be studied by the configuration of the lacuarity curves and 3TLQV curves which is a derivative algorithm of lacuarity respectively. Thus the lacunarity analysis method is a reasonable complementarity for the Digital Terrain Analysis theory. In this paper, The Lacunarity analysis method, which is one of mature analysis methods in land-scape ecology, is presented and discussed, by introducing its principle, arithmetic and feasibility in digital terrain analysis firstly. Using Jiu Yuan - gou drainage in northern Sbaanxi hilly - gully area as the test site and 5m - reso-lution DEMs as original test data, the features lines of the watershed are studied by using of lacunarity and 3TLQV (Three - Term Local Quadrat Variance) methods. The spatial pattern of the drainage features lines is illustrated in the direction of north to south and east to west respectively by the configuration of the lacunarity and 3TLQV analy-sis curves. Results show that they are slightly concave curves of the lacunarity and they have high fitting degree with the straight line. This phenomenon indicates that the configuration of the drainage feature lines is self - organi-zing and self-similar. The similarity between the lacunarity curves of the ridge lines and hollow lines illustrates that the anisotropy character of the drainage landform is not obvious. Because of better consistency of the hollow lines, the lacunarity value of hollow lines is a little higher than that of the ridge lines. Meanwhile, the 3TLQV curves of the feature lines have three peak points in the two directions respectively. This means that there are three characteristic scale ranges in the drainage basin. They are 540m, 2100m and 4845m in EW direction and 1215m, 2810m and 5700m in NS direction. This is an initial experiment for revealing the spatial pattern of the drainage ba-sin in the loess hilly -gully region, and it also provide a reference for choosing the analysis scale in regional land-form study.

As one of the important parts of Digital Terrain Analysis (DTA), spatial correlation analysis of topographic attributes (TAs) is an effective method of analysing the topographical environment. This chapter proposes a spatial correlation model for nine selected TAs, providing an effective method for quantitative DTA research and landform recognition. Forty seven different loess landforms were selected as test areas and their corresponding 5 m grid cell DEM data as test data. With grey correlation analysis, spatial correlations for these TAs were analysed and the TAs’ correlation model built. Furthermore, the variations of the correlation curves are discussed. Results show that (1) TA correlation curves are similar to the spectrum, which provides a means of modelling the natural environment; (2) the correlation curve changes with the topographical area; and (3) the correlation curve reflects the landform and evolution pattern of the sample area.

Slope is a parameter which reflects the rate of inclination that earth' s surface incline relatively to hori- zontal plane.It is also the one of most important quantitative index that describe the earth' s surface.Slope spectrum can reflect the macroscopical hypsography characteristic of surface.For the same type of physiognomy,different slope spectrum correspond to different DEM horizontal resolution.So it is necessary to study DEM' s slope spectrum scaling.Downscaling model of slope spectrum is the first step of landform information scaling.It is of great signifi- cance for revealing the scale dependence of slope spectrum.This study downscale the 25 m DEM' slope spectrum to 5 m' one by the histogram' s matching method in digital image processing,for the DEM grid data is similar to re- mote sensing image.3D Douglas compress method is employed to obtain DEMs at two different resolution levels and eight different threshold value,which express the relief of different surface roughness.After deriving slope spectrum of DEMs of two kinds of resolution,the slope spectrum downscaling model between them could be founded by the histogram's matching method under each different threshold value,which indicates different simplified degree of landform.The relationship between the coefficients of downscaling model and terrain variables is found at each dif- ferent threshold value as well,which lead to the construction of the final downscaling model of 25 m DEM' slope spectrum and 5 m DEM' s.This model is verified in a case study in Jiuyuangou of loess hill-gully area.Then,the slope spectrum derived from this method is compared with the one derived from original 5 m resolution DEM.The result shows that the model could achieve the transformation from 25 m resolution DEM' s slope spectrum to 5 m resolution DEM' s effectively and accurately in loess hill-gully area.The method lays a foundation for more study of slope spectrum downscaling in other types of physiognomy area,so that a more common downscaling model of slope spectrum can be founded finally.

以陕北黄土高原多地貌类型样区为实验样区,采用5m分辨率的DEM为基本信息源,构建不同汇流阈值与所提取沟壑密度量化关系。实验结果显示,汇流阈值X与沟壑密度Y呈定量统计模型。根据所获得的陕北黄土高原不同地区模型系数值,可有效地构建汇流阈值与沟壑密度之间的定量关系。这对于两者之间的相互求解,特别是科学、合理地确定汇流阈值,提供了理论依据。

Digital Terrain Analysis (DTA) is an important way for interpreting and understanding natural landform. Scale, an essential subject for realizing pattern and process in nature, is a fundamental issue in DTA. River basins are the basic natural system of many hydrologic phenomena. Multi-scale analysis of channel network can explore structural characteristic and spatial pattern of drainage basin, make basis on drainage evolution and provide suitable scale for drainage research. This paper investigates the structural characteristic of channel network under multiple scales and finds out accurate critical points of scales. Two kinds of lacunarity algorithms, i.e. gliding box algorithm and 3TLQV are adopted. Several conclusions can be drawn from the experiments. Firstly, there are five scale patterns in WE direction and three scale patterns in NS direction in channel network of Jiuyuangou drainage basin. Each scale pattern indicates a kind of hydrologic process. Secondly, anisotropy is between WE and NS direction in channel network. Thirdly, at each scale examined there's fractal pattern and fractal dimensions in different scales have little difference. Fourthly, an effective way for interpreting spatial pattern under different scales is put forward and it can be used for other network, such as ridgelines, population distribution etc.

提出基于正交条带状分析窗口及DEM数据,实现地形特征线快速提取的方法。通过在黄土丘陵区的实验显示,该方法与现有的方法相比具有提取方法简单、精度高、栅格矢量数据格式间转换较为方便的特点。

Slope spectrum is defined as a statistic model of slope distribution in a certain area. Previous researches mainly focus on morphology depiction of the slope spectrum; its spatial distribution is unknown yet, especially in the Loess Plateau. Theory and methodology of information entropy and statistics are applied for the objective of quantitatively analyzing the slope spectrum and its spatial distribution in the Loess Plateau in North Shaanxi province. Experiment results show that slope spectrum's information entropy (H), skewness of slope spectrum (S) and terrain driving force factor (T-d) can appropriately depict the slope spectrum and its spatial distribution from different points of view. Spatial distribution of the slope spectrum represents spatial distribution of loess landform types, and it is correlatable with spatial distribution of soil erosion intensity in the Loess Plateau. H, Td and gully density, surface incision depth show positive correlation: gully density and surface incision increase as H, Td increase. On the contrary, the S and gully density, surface incision depth show negative correlation. Lastly, spatial relationship between slope spectrum and loess landform types are qualitatively analyzed, and loess landform evolution as well.

The topographic feature is one of the main factors that influence the process of soil erosion and sediment yield of small watershed.It is very necessary to quantitate the topographic feature of small watershed and get the correlative parameters rapidly.According to the process resemble principle and statistic analysis of the topographic feature of small watershed on the Loess Plateau,the small tested watershed is designed.Based on 25 artificial simulated rainfalls,the dynamic development process of the topographic feature of the tested small watershed is studied by means of photogrammetry and GIS technology.DEMs at nine different erosion levels were picked up with a high precision and resolution.The spatial variation of slope and its composition derived from these DEMs are mainly discussed through comparison analysis and experimental validation.The result shows: firstly,in the process of rainfall erosion,the velocity of the increase of the surface mean-slope is increasing during erosion prophase and decreasing during erosion activity stage.The decrease of smoother slope balanced with the increase of steeper slope at erosion critical angle,namely,the proportion of erosion critical angle histogram resultants is keeping almost steady.In addition,the variation of slope composition reflects not only the spatial variation of loess morphology,but also the rule of loess erosion and the developing process as well.

As a key process of surface erosion,the formation of a drainage basin is the results of long-time reciprocity of different factors affecting the surface erosion.It has been a significant task to depict precisely the processes and features of physiognomy development at regional scale.This research probes into the temporal-spatial process of drainage development of Loess Plateau on the basis of a carefully designed experiment.In the experiment,the development of a simulated loess watershed is tested under the condition of manual rainfall.The typical drainage features of Loess Plateau are abstracted and generalized in the process.Through 25 times stimulating rainfall of different intensity and epochs,the drainage has turn from the infinitive smooth shallow watershed into a rough and fragmented terrain.A close-range photogrammetry survey is employed in this experiment,per which a series of high precision and resolution DEM of the drainage is established,which could be applied in investigating the dynamic development features of the drainage.In addition,the mean slope and slope composition of the whole and part of drainage basin are extracted and analyzed.Furthermore,the loess physiognomy form and its development features are discovered in a more macro spatial scale and higher precise temporal scale.1) The stimulation result can effectively reflect the truth if those experimental conditions,i.e. loess soil structure,stimulated rainfall,is adjusted in accord with the true situation;2) The slope of loess drainage varies all through the rainfall erosion process.Regarding the drainage basin at primitive experiment stage to be the infancy period of relief development,an accelerated increasing of the mean slope is presented.While in the adulthood,the increasing range is descending.The durative variances of slope combinations turn suddenly different at the erosion critical angle with the corresponding areas of erosion critical slope basically stable.3) The variances of slope in Loess areas are influenced by the degree of erosion and deposition of loess soil.On the other hand,the composition of different terrain type area contributes a lot to the variation of slope spectrum as well.

Terrain factors,although different in the definition and calculation method,relate each other at different extent.Such relationship can be represented by a correlation index,which reveals the process and stage of terrain development as well.This paper focuses mainly on the correlation between different terrain factors and the mean-slope by means of the Back Propagation model of Neural Network with a latent layer.Furthermore,the regression model and the NN model without a latent layer are compared with the NN model with a latent layer.Fifteen loess gully-hill areas are selected as the experimental area,and the relevant 1∶10 000 scale DEMs(5 m×5 m grid) are applied as the basic data.From the results of the NN model with a latent layer,it is found that roughness and undulation are the most closely correlated with mean-slope.Compared with others,channel density and mean elevation are the least correlated with mean-slope.Experiment results show the NN model with a latent layer is better than the others and it can effectively evaluate the correlation between the terrain factors extracted from DEMs.This method provides a new methodology in the selection of suitable and available terrain factors and the estimation of the relevancy between these factors.

Different terrain factors express the undulating characteristics and spatial variations of the true surface from different aspects. The relationships among them can play a key role in revealing the mechanism and development of the terrain and geomorphologic situation to a great extent. The relationships and their variance discipline between the terrain factors and mean slope are discussed in this paper via the Back Propagation model of Neural Network. Fifteen loess gully-hilly areas are selected as the test areas for experiment, and the relevant 1∶10 000 and 1∶50 000 map scale DEMs of high resolution and high precision are also selected as the basic data. The results show this method can effectively evaluate the relevancy of terrain factors on the mean slope extracted from DEMs at the two scales. It is hoped that this result can be helpful in evaluating the availability of the DEM scale applied, determining the relevancies among multiple topographical factors as well as selecting suitable terrain variables for different applications.