Geo-statistical Methods For Spatial Interpolation in GIS
Transcription
Geo-statistical Methods For Spatial Interpolation in GIS
International Conference on Space (ICS-2014) Organized by SUPARCO, IST, and ISNET 12-14 Nov 2014, Islamabad Geo-statistical Methods For Spatial Interpolation in GIS Anam Ahsan1, Shahid Parvez2 1B.S (Hon.) student, Dept of Space Science, University of the Punjab, Lahore <anamahsan33@yahoo.com> 2Assist Prof, Dept of Space Science, University of the Punjab, Lahore <shahid.spsc@pu.edu.pk> www.pu.edu.pk Contents Study Objective Study Area Geo-statistical Analyses in ArcGIS • Modeling • Kriging and its Types Spatial Interpolation in New-LocClim Comparison Conclusions References Study Objectives • To understand Kriging and its Types. • To investigate interpolation of Temperature values for unknown places using Kriging. • To examine the comparison of Kriging in ArcGIS and New-LocClim (FAO). Study Area (Temperature Data of Some Cities of Pakistan) Arabian Sea Sample data - Temperature City/Town Abbotabad Bahawalnagar Bahawalpur Chaman Charsadda Chilas Daska Digri Faisalabad Gagai Gupis Hyderabad Islamabad Jhelum Jhang Karampur Kharan Karachi Kasur Lahore Longitude (Deg) Latitude (Deg) 73.21 73.25 71.68 66.45 71.73 74.09 74.35 69.11 73.07 64.70 73.44 68.36 73.06 73.72 72.31 73.69 65.41 67.02 74.44 74.34 34.15 29.99 29.39 30.91 34.15 35.42 32.33 25.15 31.41 29.29 36.22 25.37 33.71 32.93 31.26 30.28 28.58 24.89 31.11 31.54 Source: http://www.accuweaddzther.com High Temperature (˚C) 32 40 40 37 32 32 34 34 34 34 30 34 32 32 34 40 34 34 34 34 Low Temperature (˚C) 20 26 26 20 30 20 27 29 27 29 19 29 30 30 27 26 29 29 27 27 Parameters and Histogram High Temperature Histogram Formula Calculation 14 Mean 34.87 12 Median 34.00 Mode 34.00 Minimum 30.00 Maximum 40.00 34 Frequency Parameter 10 32 8 38 6 40 4 2 30 0 0 Temperature (˚C) Low Temperature Mean Median Formula Calculation 25.53 27.00 12 27 10 Frequency Parameter Histogram 21 8 29 6 Mode 20.00 Minimum 19.00 2 Maximum 30.00 0 31 4 19 1 2 0 0 3 4 5 Temperature (˚C) 6 7 Geo-Statistics Geo-statistics • • Geo-statistics is a branch of statistics focusing on two data set : • Spatial • Spatiotemporal Geo-statistical analyses of data occur in two phase : • Modeling (for semi-variogram or covariance) • Kriging (for surface creation) Modeling • In modeling the semi-variogram or covariance is use to analyze surface properties. Semi-variogram Kriging Interpolation Kriging Interpolation • Kriging is a geo-statistical interpolation. • Kriging uses statistical models that allow a variety of map outputs including predictions, prediction standard errors, probability, etc. • Three types of kriging are normally used: • Simple Kriging • Ordinary Kriging • Universal Kriging method for spatial Simple Kriging Simple Kriging • It assume that the mean of the data set is known. • This assumption is unrealistic in most cases. Threshold Flow Chart (Simple Kriging in Arc-Map) Geo-Statistical Wizard Simple Kriging Next Searching Neighborhood Next Semivarogram/ Covariance Geo-statistical Model Section Next Crossvalidation Finish and Ok Prediction Map - Temperature (Simple Kriging in Arc-map) Contour Values Attribute : High Temp Neighbor : 5 Mean : -0.03012 RMS : 2.021 Model : Exponential Partial sill : 10.86 Minor range : 5.002 Direction : 287.4 Measured and Predicted Values (MS Excel work of Simple Kriging) Measured Predicted Error Std-Error Stdd-Error Norm-Value Source-ID Included 32.00 32.11 0.11 2.08 0.05 -0.28 0 Yes 40.00 37.93 -2.07 2.55 -0.81 -0.83 1 Yes 40.00 34.99 -5.01 3.08 -1.63 -1.68 2 Yes 37.00 35.40 -1.60 3.03 -0.53 -0.72 3 Yes 32.00 33.66 1.66 1.74 0.96 1.42 4 Yes 32.00 32.20 0.20 2.81 0.07 -0.04 5 Yes 34.00 34.00 0.00 1.81 0.00 -0.36 6 Yes 34.00 34.72 0.72 3.07 0.24 0.12 7 Yes 34.00 35.22 1.22 2.74 0.45 0.95 8 Yes 34.00 34.78 0.78 3.13 0.25 0.20 9 Yes 30.00 33.90 3.90 3.03 1.29 1.68 10 Yes 34.00 34.17 0.17 2.62 0.06 -0.20 11 Yes 32.00 32.40 0.40 2.11 0.19 0.04 12 Yes 32.00 33.54 1.54 2.82 0.55 1.23 13 Yes Correlation and Scatter plot Correlation 1 Column 2 0.720794 1 Scatterplot 39.00 Values Column 1 Column 2 Predicted Column 1 Data Point 36.00 Linear (Data Point) 33.00 30.00 27.00 32.00 37.00 Measured Values 42.00 Ordinary Kriging Ordinary Kriging • Data having a constant mean (no trend) - value is not known in advance. • Orange dots show a random error – fluctuate around the unknown mean. Flow Chart (Ordinary Kriging in Arc-Map) Geo-Statistical Wizard Next Searching Neighborhood Ordinary Kriging Next Semivarogram/ Covariance Geo-statistical Model Section Next Crossvalidation Finish and Ok Prediction Map - Temperature (Ordinary Kriging in Arc-Map) Contour Values Attribute : High Temp Neighbor : 5 Mean : -0.03012 RMS : 1.708 Model : Exponential Partial sill : 10.86 Minor range : 5.002 Direction : 287.4 Measured and Predicted Values (MS Excel work of Ordinary Kriging) Measured Predicted Error Std-Error Stdd_Error Norm-Value Source-ID Included 32.00 33.39 1.39 1.93 0.72 1.23 0 Yes 40.00 38.94 -1.06 1.99 -0.53 -0.83 1 Yes 40.00 37.67 -2.33 2.53 -0.92 -1.23 2 Yes 37.00 36.77 -0.23 2.39 -0.10 -0.28 3 Yes 32.00 33.01 1.01 1.41 0.71 1.08 4 Yes 32.00 31.49 -0.51 2.41 -0.21 -0.45 5 Yes 34.00 33.75 -0.25 1.45 -0.18 -0.36 6 Yes 34.00 34.43 0.43 2.08 0.20 0.12 7 Yes 34.00 34.38 0.38 1.93 0.20 0.04 8 Yes 34.00 34.76 0.76 2.44 0.31 0.53 9 Yes 30.00 32.53 2.53 2.37 1.07 1.68 10 Yes 34.00 33.99 -0.01 1.89 -0.01 -0.12 11 Yes 32.00 32.00 0.00 1.82 0.00 -0.04 12 Yes 32.00 33.09 1.09 1.92 0.57 0.83 13 Yes Correlation and Scatterplot Correlation 1 Column 2 0.800989 1 Scatterplot 39.00 Values Column 1 Column 2 Predicted Column 1 Data Point 36.00 Linear (Data Point) 33.00 30.00 27.00 32.00 37.00 Measured Values 42.00 Universal Kriging Universal Kriging • It assumes there is trend in the data, but the terms of the trend function are not known in advance. • The data values (orange dots) are through of as random errors that fluctuate around the unknown tend. Flow Chart (Universal Kriging in Arc-Map) Geo-Statistical Wizard Next Searching Neighborhood Universal Kriging Next Semivarogram/ Covariance Geo-statistical Model Section Next Crossvalidation Finish and Ok Prediction Map – Temperature (Universal Kriging in Arc-Map) Contour Values Attribute : High Temp Neighbor : 5 Mean : -0.03012 RMS : 1.862 Model : Exponential Partial sill : 10.86 Minor range : 5.002 Direction : 287.4 Measured & Predicted Values (Excel work of Universal Kriging) Measured Predicted Error Std-Error Stdd-Error Norm-Value Source-ID Included 32.00 32.24 0.24 2.06 0.12 -0.12 0 Yes 40.00 38.55 -1.45 2.34 -0.62 -0.83 1 Yes 40.00 35.83 -4.17 3.03 -1.38 -1.68 2 Yes 37.00 36.34 -0.66 2.92 -0.23 -0.63 3 Yes 32.00 33.41 1.41 1.60 0.88 1.42 4 Yes 32.00 31.65 -0.35 2.71 -0.13 -0.45 5 Yes 34.00 33.82 -0.18 1.63 -0.11 -0.36 6 Yes 34.00 35.10 1.10 2.95 0.37 0.53 7 Yes 34.00 35.17 1.17 2.52 0.46 0.72 8 Yes 34.00 35.73 1.73 3.14 0.55 1.08 9 Yes 30.00 33.32 3.32 2.88 1.15 1.68 10 Yes 34.00 34.12 0.12 2.44 0.05 -0.28 11 Yes 32.00 32.28 0.28 2.06 0.14 -0.04 12 Yes 32.00 33.21 1.21 2.59 0.47 0.83 13 Yes Correlation and Scatterplot Correlation 1 Column 2 0.760096 1 Scatterplot 39.00 Values Column 1 Column 2 Predicted Column 1 Data Plot 36.00 Linear (Data Plot) 33.00 30.00 27.00 32.00 37.00 Measured Values 42.00 Spatial Interpolation Using New-LocClim (FAO) New-LocClim (FAO) (V 1.10) • New LocClim is a tool for spatial interpolation of agroclimatic data. • New LocClim runs in 3 modes: • Single Point Mode • Workbench Mode • Automatic Mode Sample Data – Temperature Longitude Latitude Average Temperature City/Town 73.21 34.15 26 Abbotabad 73.25 29.99 33 Bahawalnagar 71.68 29.39 33 Bahawalpur 66.45 30.91 28.5 Chaman 71.73 34.15 31 Charsadda 74.09 35.42 26 Chilas 74.35 32.33 30.5 Daska 69.11 25.15 31.5 Digri 73.07 31.41 30.5 Faisalabad 64.7 29.29 31.5 Gagai 73.44 36.22 24.5 Gupis 68.36 25.37 31.5 Hyderabad 73.06 33.71 31 Islamabad 73.72 32.93 31 Jehlam 72.31 31.26 30.5 Jhang Flow Chart New-LocClim (FAO) Workbench Mode Data Source Import From file Setting (method and Grid) Legend Export Spatial Interpolation using New-LocClim Station RMS : 1.945 Grid Size :3 Grid RMS : 1.358 Scale: Pakistan Comparison Comparison of Kriging in ArcGIS Simple Kriging Attribute : High Temp Neighbor :5 Mean : -0.03012 RMS : 2.021 Model : Exponential Partial sill : 10.86 Minor range : 5.002 Direction : 287.4 Ordinary Kriging Universal Kriging Contour Values High Temp 5 -0.03012 1.708 Exponential 10.86 5.002 287.4 High Temp 5 -0.03012 1.862 Exponential 10.86 5.002 287.4 Comparison of Kriging in ArcGIS and New-LocClim 40 ˚ C 35 ˚ C 30 ˚ C RMS : 1.708 RMS : 1.945 Conclusions • The results indicate that the comparison of Geo-statistical analyses in ArcGIS and spatial interpolation in New LocClim of temperature data, Geo-statistical analyses in ArcGIS is better for interpolation. • The correlation of Ordinary Kriging is 0.80 (perfect positive relationship) and RMS is 1.708, which show that Ordinary Kriging is better as compared to the other methods. • We can not say that only Ordinary Kriging is better than others, basically it depends on number of data values. Reference • H.A. Torbert l , E. Kruege, and D. Kurtener (2008) “Soil quality assessment using fuzzy modeling” Received April 11, 2008; accepted August 14, 2008 • Jennifer D. Knoepp , David C. Coleman,D.A. Crossley Jr., James S. Clark (2000) “Biological indices of soil quality: an ecosystem case study of their use Forest Ecology and Management” 138 (2000) 357-368 • CHANG Wen-Yuan DAI Xin-Gang CHEN Hong-Wu (2004) “A CASE STUDY OF GEOSTATISTICAL INTERPOLATION IN METEOROLOGICAL FIELDS CHINESE JOURNAL OF GEOPHYSICS” Vol.47, No.6, 2004, pp: 1104∼1112 • Benjamin J. Mason U.S. EPA (1992). “Preparation of Soil Sampling Protocols: Sampling Techniques and Strategies.” Report No.EPA/600/R-92/128, Environmental Monitoring Systems Laboratory, Office of Research and Development. • https://www.google.com.pk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&cad=rja&uact=8&ved=0CCAQFjAB&url= http%3A%2F%2Fwww.biology.ufl.edu%2FCOURSES%2Fpcb6049%2F2009fall%2Fbolker%2Fpresentations%2FIntrod uction%2520to%2520Geostatistics%2520Presentation_MAGedit.ppt&ei=K1O2U6OoN4eo0wXSjYDoCA&usg=AFQjC NFdsanyrv6L1hLDzN0SSNVd9mRX5A&bvm=bv.70138588,d.d2k. • https://www.google.com.pk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0CBoQFjAA&url= http%3A%2F%2Fwww.uvm.edu%2Fenvnr%2Fgradgis%2Fadvanced%2Fkrig.ppt&ei=K1O2U6OoN4eo0wXSjYDoCA& usg=AFQjCNEsqTe_d2rtJrDlwZ8mdGV0Ib1NTw&bvm=bv.70138588,d.d2k • https://www.google.com.pk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=3&cad=rja&uact=8&ved=0CCcQFjAC&url= http%3A%2F%2Fwww.geog.ucsb.edu%2F~good%2Fcsissworkshop%2Fvespucci_thursday.ppt&ei=K1O2U6OoN4eo0w XSjYDoCA&usg=AFQjCNEHoGtKJOeJIR_roSwD_lF1-ofxQQ&bvm=bv.70138588,d.d2k • https://www.google.com.pk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=7&cad=rja&uact=8&ved=0CD8QFjAG&url= http%3A%2F%2Fae.sharif.edu%2F~aerocad%2FKriging%2520Interpolation.ppt&ei=tFO2U6rzHMKg0QWr9oGYCQ&u sg=AFQjCNErsAcjYMnLzDTnHXhRnvEZzQixAQ&bvm=bv.70138588,d.d2k • http://www.epa.gov/airtrends/specialstudies/dsisurfaces.pdf • http://www.nuim.ie/staff/dpringle/gis/gis09.pdf • http://hcgl.eng.ohio-state.edu/~ceg608/handouts/SpatialInterpolation.pdf • http://msdis.missouri.edu/resources/gis_advanced/pdf/Interpolation.pdf • http://www.esri.com/software/arcgis/extensions/geostatistical Thank you for your kind attention.