Raster Analysis

这是一篇关于Spatial Estimation and modelin的思维导图，主要内容包括：Correlation，Auto Correlation，Point pattern analysis，Hypothesis Testing。

Raster Analysis

• Raster
  • in *gdb folder
  • .tiff, .tif ,.geotiff
  • Versus Vector Image
    • precison depeding on pixel
    • Quick Computation
  • Attribute Table (toolbox: Raster Attribute Table)
    • OBJECTID/OID
      • pixel id
    • VALUE
      • pixel value
    • COUNT
      • # of pixels with same value
• Step 1: Set Raster Environment
  • Choose one raster layer (toolbox: Line/Point/Polygon to Raster)
  • set mask and snap Raster as the given raster layer
• Step 2: Mapping Density (for point Vector Image)
  • Transfer csv to point vector image (toolbox: Display XY Data)
  • Using Density tool
    • Point Density
    • Kernel Density
      • Polulation Field: Give each point a "weight"
• Step 3: Surface Analysis
  • Contours
    • Finding areas with same value
    • output is polyline (vector image)
  • Slope
    • degree of slope
      • arctan (rise/run)
    • percent of slope
      • 100 * tan J
  • Aspect
    • Aspect from each cell of the surface
  • Hillshade
  • ViewShede
    • input: raster layer and observations points
    • output: new raster layer, value are visible/not visible
  • Cut&Fill
    • input: two Raster Layers
    • output: new Raster Layer, positive value means cut, negative value means Fill

Suitability Analysis

• Step 1: Converting Data
  • toolsbox: feature to raster
  • density: Point density/ Kernel density
  • distance: Euclidean Distance...
  • surface analysis: slope, aspect, hillshade, viewshed, Cut/Fill
• Step 2: Reclassify Data
  • Replace input cell value with new cell value
    • Group value or replace it based on new infomation
    • Rescale the value to the common Scale
    • Elimation method with NoData
• Step 3: Weighted Combination
  • toolbox: Raster Calculator
• Step 4: Raster Statistic
  • Cell Statistic: many rasters -> new raster image
  • Focal Statistic: like convolution
  • Zonal Statistic: Polygon + raster -> value

Spatial Estimation and modeling

• Correlation
  • Positive
  • negative
• Auto Correlation
  • Apatial Weights Matrix neihjbourhoods - lags
    • Contiguity
    • Nearestt neighbours
    • Distance
  • Testing
    • Moran's I
      • hypothesis
        • p < rhreshold
          • Moran's I > 0
            • clustering of similar value
          • Moran's I < 0
            • dispersion of similar value
        • otherwise
          • CSR
    • LISA
      • High-High, Low-Low
        • significant cluster
      • High-Low, Low-High
        • Outliers
  • Relulations
    • Everything is related to everythingelse,but near things are more related than distant things
    • Spatial Heterogeneity
    • Spatial Similarity principle
• Point pattern analysis
  • Pattern: Centrography
    • Standard distance Circle
      • Mean center
      • standard distance
    • Standard deviational ellipse
      • Mean center
      • major axis, minor axis, angle
  • nature of distribution: Spatial Clustering
    • Point Patterns
      • random
        • equal prob.
        • independence
        • number of events in any subregion follows Poisson distribution
      • uniform/dispersed
      • clustered
    • Statistics: Average Nearest Neighbour
      • • D_o : observe mean distance
        • D_E: expected mean distance
      • do z-score of D_O and D_E can get p-value
      • ANN > 1
        • Dispersed
      • ANN » 1
        • Random
      • ANN < 1
        • Clustered
    • hypothesis
      • H0
        • compete CSR
      • H1
        • deviate from CSR
• Hypothesis Testing
  • • null hypothesis is prediction of no relationship
  • Comparison of within-group var or between-group var
  • Test we choose
    • continuous-continuous
      • Peasonr
    • continuous-bivariate categorical
      • t-test
    • continuous-multivariate categorical
      • Anova
    • categorical-categorical
      • Chi-square test
  • returns
    • estimate of difference
    • p-value
      • smaller than threshold
        • reject null hypothesis
      • otherwise
        • can not reject