160 likes | 309 Vues
Geostatistical Analysis of Hydrologic Parameters. Nishesh Mehta Hydrology - CE394K 26 th April 07. Outline of the problem. An interesting study to investigate geospatial correlationship between hydrologic parameters. Industrial Water Use Public Supply Water Use Irrigation Water Use
E N D
Geostatistical Analysis of Hydrologic Parameters Nishesh Mehta Hydrology - CE394K 26th April 07
Outline of the problem An interesting study to investigate geospatial correlationship between hydrologic parameters. • Industrial Water Use • Public Supply Water Use • Irrigation Water Use • Slope (% flatlands) • Geologic Texture (% sand) • Bedrock Permeability • Climate (Precipitation-PET)
Data Sources • Water Use data for the US http://water.usgs.gov/watuse/data/2000/index.html contains: Industrial Water Use Public Supply Water Use Agricultural Water Use • Hydrologic landscape regions of the United States http://water.usgs.gov/GIS/metadata/usgswrd/XML/hlrus.xml Contains:
Semivariograms- How to do it? Geostatistical Analyst • The semivariogram captures the spatial dependence between samples by plotting semivariance against separation distance h= 0.5 * avg[ (value i –value j)2 ]
Preliminary Results • Correlation lengths calculated on county basis • Data used was raw (not treated to have a normal distribution) • Semivariance calculated between each set of counties within the continental US
Synthesis of Analysis • base unit of analysis - Counties – 3077 HUCs – 2158 (grouped on similar hydrologic properties) • Spatial Join – A tool that helps to associate and interpolate values spatially. Ex- convert parameter classified by county basis to HUC basis • Random Sampling – Basis of all statistical processes Enables sampling out of a large number of points
CUAHSI test bed sites as pilot test Sierra Nevada • Random sampling of HUCs from the site comprised of HUC units • Use any parameter from the attribute table
Results • Scaling length Industrial water for the entire Sierra Nevada– 110 kms • Scaling length Industrial water for a random sample of Sierra Nevada- 110 kms
Statistical Significance • Treatment of Data – to attain normality • logIndustrialWaterUse=log10(0.1+ IndustrialWaterUse) • A quick fix method to check results • Moran’s index - A test for spatial autocorrelation Positive spatial autocorrelation indicates spatial clustering
What to take back ?! • The cool randomizing tool ($$ in royalty) • The intellectual framework of how Geospatial correlation may be computed • ArcGIS has powerful geostatistic tools