Table of Contents for
PostGIS Cookbook - Second Edition

Version ebook / Retour

Cover image for bash Cookbook, 2nd Edition PostGIS Cookbook - Second Edition by Thomas J Kraft Published by Packt Publishing, 2018
  1. PostGIS Cookbook, Second Edition
  2. Title Page
  3. Copyright and Credits
  4. PostGIS Cookbook Second Edition
  5. Packt Upsell
  6. Why subscribe?
  7. PacktPub.com
  8. Contributors
  9. About the authors
  10. Packt is searching for authors like you
  11. Table of Contents
  12. Preface
  13. Who this book is for
  14. What this book covers
  15. To get the most out of this book
  16. Download the example code files
  17. Download the color images
  18. Conventions used
  19. Sections
  20. Getting ready
  21. How to do it…
  22. How it works…
  23. There's more…
  24. See also
  25. Get in touch
  26. Reviews
  27. Moving Data In and Out of PostGIS
  28. Introduction
  29. Importing nonspatial tabular data (CSV) using PostGIS functions
  30. Getting ready
  31. How to do it...
  32. How it works...
  33. Importing nonspatial tabular data (CSV) using GDAL
  34. Getting ready
  35. How to do it...
  36. How it works...
  37. Importing shapefiles with shp2pgsql
  38. How to do it...
  39. How it works...
  40. There's more...
  41. Importing and exporting data with the ogr2ogr GDAL command
  42. How to do it...
  43. How it works...
  44. See also
  45. Handling batch importing and exporting of datasets
  46. Getting ready
  47. How to do it...
  48. How it works...
  49. Exporting data to a shapefile with the pgsql2shp PostGIS command
  50. How to do it...
  51. How it works...
  52. Importing OpenStreetMap data with the osm2pgsql command
  53. Getting ready
  54. How to do it...
  55. How it works...
  56. Importing raster data with the raster2pgsql PostGIS command
  57. Getting ready
  58. How to do it...
  59. How it works...
  60. Importing multiple rasters at a time
  61. Getting ready
  62. How to do it...
  63. How it works...
  64. Exporting rasters with the gdal_translate and gdalwarp GDAL commands
  65. Getting ready
  66. How to do it...
  67. How it works...
  68. See also
  69. Structures That Work
  70. Introduction
  71. Using geospatial views
  72. Getting ready
  73. How to do it...
  74. How it works...
  75. There's more...
  76. See also
  77. Using triggers to populate the geometry column
  78. Getting ready
  79. How to do it...
  80. There's more...
  81. Extending further...
  82. See also
  83. Structuring spatial data with table inheritance
  84. Getting ready
  85. How to do it...
  86. How it works...
  87. See also
  88. Extending inheritance – table partitioning
  89. Getting ready
  90. How to do it...
  91. How it works...
  92. See also
  93. Normalizing imports
  94. Getting ready
  95. How to do it...
  96. How it works...
  97. There's more...
  98. Normalizing internal overlays
  99. Getting ready
  100. How to do it...
  101. How it works...
  102. There's more...
  103. Using polygon overlays for proportional census estimates
  104. Getting ready
  105. How to do it...
  106. How it works...
  107. Working with Vector Data – The Basics
  108. Introduction
  109. Working with GPS data
  110. Getting ready
  111. How to do it...
  112. How it works...
  113. Fixing invalid geometries
  114. Getting ready
  115. How to do it...
  116. How it works...
  117. GIS analysis with spatial joins
  118. Getting ready
  119. How to do it...
  120. How it works...
  121. Simplifying geometries
  122. How to do it...
  123. How it works...
  124. Measuring distances
  125. Getting ready
  126. How to do it...
  127. How it works...
  128. Merging polygons using a common attribute
  129. Getting ready
  130. How to do it...
  131. How it works...
  132. Computing intersections
  133. Getting ready
  134. How to do it...
  135. How it works...
  136. Clipping geometries to deploy data
  137. Getting ready
  138. How to do it...
  139. How it works...
  140. Simplifying geometries with PostGIS topology
  141. Getting ready
  142. How to do it...
  143. How it works...
  144. Working with Vector Data – Advanced Recipes
  145. Introduction
  146. Improving proximity filtering with KNN
  147. Getting ready
  148. How to do it...
  149. How it works...
  150. See also
  151. Improving proximity filtering with KNN – advanced
  152. Getting ready
  153. How to do it...
  154. How it works...
  155. See also
  156. Rotating geometries
  157. Getting ready
  158. How to do it...
  159. How it works...
  160. See also
  161. Improving ST_Polygonize
  162. Getting ready
  163. How to do it...
  164. See also
  165. Translating, scaling, and rotating geometries – advanced
  166. Getting ready
  167. How to do it...
  168. How it works...
  169. See also
  170. Detailed building footprints from LiDAR
  171. Getting ready
  172. How to do it...
  173. How it works...
  174. Creating a fixed number of clusters from a set of points
  175. Getting ready
  176. How to do it...
  177. Calculating Voronoi diagrams
  178. Getting ready
  179. How to do it...
  180. Working with Raster Data
  181. Introduction
  182. Getting and loading rasters
  183. Getting ready
  184. How to do it...
  185. How it works...
  186. Working with basic raster information and analysis
  187. Getting ready
  188. How to do it...
  189. How it works...
  190. Performing simple map-algebra operations
  191. Getting ready
  192. How to do it...
  193. How it works...
  194. Combining geometries with rasters for analysis
  195. Getting ready
  196. How to do it...
  197. How it works...
  198. Converting between rasters and geometries
  199. Getting ready
  200. How to do it...
  201. How it works...
  202. Processing and loading rasters with GDAL VRT
  203. Getting ready
  204. How to do it...
  205. How it works...
  206. Warping and resampling rasters
  207. Getting ready
  208. How to do it...
  209. How it works...
  210. Performing advanced map-algebra operations
  211. Getting ready
  212. How to do it...
  213. How it works...
  214. Executing DEM operations
  215. Getting ready
  216. How to do it...
  217. How it works...
  218. Sharing and visualizing rasters through SQL
  219. Getting ready
  220. How to do it...
  221. How it works...
  222. Working with pgRouting
  223. Introduction
  224. Startup – Dijkstra routing
  225. Getting ready
  226. How to do it...
  227. Loading data from OpenStreetMap and finding the shortest path using A*
  228. Getting ready
  229. How to do it...
  230. How it works...
  231. Calculating the driving distance/service area
  232. Getting ready
  233. How to do it...
  234. See also
  235. Calculating the driving distance with demographics
  236. Getting ready
  237. How to do it...
  238. Extracting the centerlines of polygons
  239. Getting ready
  240. How to do it...
  241. There's more...
  242. Into the Nth Dimension
  243. Introduction
  244. Importing LiDAR data
  245. Getting ready
  246. How to do it...
  247. See also
  248. Performing 3D queries on a LiDAR point cloud
  249. How to do it...
  250. Constructing and serving buildings 2.5D
  251. Getting ready
  252. How to do it...
  253. Using ST_Extrude to extrude building footprints
  254. How to do it...
  255. Creating arbitrary 3D objects for PostGIS
  256. Getting ready
  257. How to do it...
  258. Exporting models as X3D for the web
  259. Getting ready
  260. How to do it...
  261. There's more...
  262. Reconstructing Unmanned Aerial Vehicle (UAV) image footprints with PostGIS 3D
  263. Getting started
  264. How to do it...
  265. UAV photogrammetry in PostGIS – point cloud
  266. Getting ready
  267. How to do it...
  268. UAV photogrammetry in PostGIS – DSM creation
  269. Getting ready
  270. How to do it...
  271. PostGIS Programming
  272. Introduction
  273. Writing PostGIS vector data with Psycopg
  274. Getting ready
  275. How to do it...
  276. How it works...
  277. Writing PostGIS vector data with OGR Python bindings
  278. Getting ready
  279. How to do it...
  280. How it works...
  281. Writing PostGIS functions with PL/Python
  282. Getting ready
  283. How to do it...
  284. How it works...
  285. Geocoding and reverse geocoding using the GeoNames datasets
  286. Getting ready
  287. How to do it...
  288. How it works...
  289. Geocoding using the OSM datasets with trigrams
  290. Getting ready
  291. How to do it...
  292. How it works...
  293. Geocoding with geopy and PL/Python
  294. Getting ready
  295. How to do it...
  296. How it works...
  297. Importing NetCDF datasets with Python and GDAL
  298. Getting ready
  299. How to do it...
  300. How it works...
  301. PostGIS and the Web
  302. Introduction
  303. Creating WMS and WFS services with MapServer
  304. Getting ready
  305. How to do it...
  306. How it works...
  307. See also
  308. Creating WMS and WFS services with GeoServer
  309. Getting ready
  310. How to do it...
  311. How it works...
  312. See also
  313. Creating a WMS Time service with MapServer
  314. Getting ready
  315. How to do it...
  316. How it works...
  317. Consuming WMS services with OpenLayers
  318. Getting ready
  319. How to do it...
  320. How it works..
  321. Consuming WMS services with Leaflet
  322. How to do it...
  323. How it works...
  324. Consuming WFS-T services with OpenLayers
  325. Getting ready
  326. How to do it...
  327. How it works...
  328. Developing web applications with GeoDjango – part 1
  329. Getting ready
  330. How to do it...
  331. How it works...
  332. Developing web applications with GeoDjango – part 2
  333. Getting ready
  334. How to do it...
  335. How it works...
  336. Developing a web GPX viewer with Mapbox
  337. How to do it...
  338. How it works...
  339. Maintenance, Optimization, and Performance Tuning
  340. Introduction
  341. Organizing the database
  342. Getting ready
  343. How to do it...
  344. How it works...
  345. Setting up the correct data privilege mechanism
  346. Getting ready
  347. How to do it...
  348. How it works...
  349. Backing up the database
  350. Getting ready
  351. How to do it...
  352. How it works...
  353. Using indexes
  354. Getting ready
  355. How to do it...
  356. How it works...
  357. Clustering for efficiency
  358. Getting ready
  359. How to do it...
  360. How it works...
  361. Optimizing SQL queries
  362. Getting ready
  363. How to do it...
  364. How it works...
  365. Migrating a PostGIS database to a different server
  366. Getting ready
  367. How to do it...
  368. How it works...
  369. Replicating a PostGIS database with streaming replication
  370. Getting ready
  371. How to do it...
  372. How it works...
  373. Geospatial sharding
  374. Getting ready
  375. How to do it...
  376. How it works...
  377. Paralellizing in PosgtreSQL
  378. Getting ready
  379. How to do it...
  380. How it works...
  381. Using Desktop Clients
  382. Introduction
  383. Adding PostGIS layers – QGIS
  384. Getting ready
  385. How to do it...
  386. How it works...
  387. Using the Database Manager plugin – QGIS
  388. Getting ready
  389. How to do it...
  390. How it works...
  391. Adding PostGIS layers – OpenJUMP GIS
  392. Getting ready
  393. How to do it...
  394. How it works...
  395. Running database queries – OpenJUMP GIS
  396. Getting ready
  397. How to do it...
  398. How it works...
  399. Adding PostGIS layers – gvSIG
  400. Getting ready
  401. How to do it...
  402. How it works...
  403. Adding PostGIS layers – uDig
  404. How to do it...
  405. How it works...
  406. Introduction to Location Privacy Protection Mechanisms
  407. Introduction
  408. Definition of Location Privacy Protection Mechanisms – LPPMs
  409. Classifying LPPMs
  410. Adding noise to protect location data
  411. Getting ready
  412. How to do it...
  413. How it works...
  414. Creating redundancy in geographical query results
  415. Getting ready
  416. How to do it...
  417. How it works...
  418. References
  419. Other Books You May Enjoy
  420. Leave a review - let other readers know what you think

How to do it...

To use ST_MapAlgebra() on more than two bands, we must use the callback function variant. This means we need to create a callback function. Callback functions can be written in any PostgreSQL PL language, such as PL/pgSQL or PL/R. Our callback functions are all written in PL/pgSQL, as this language is always included with a base PostgreSQL installation.

Our callback function uses the following equation to compute the three-band EVI:

The following code implements the MODIS EVI function in SQL:

CREATE OR REPLACE FUNCTION chp05.modis_evi(value double precision[][][], "position" int[][], VARIADIC userargs text[]) 
RETURNS double precision 
AS $$ 
DECLARE 
  L double precision; 
  C1 double precision; 
  C2 double precision; 
  G double precision; 
  _value double precision[3]; 
  _n double precision; 
  _d double precision; 
BEGIN 
  -- userargs provides coefficients 
  L := userargs[1]::double precision; 
  C1 := userargs[2]::double precision; 
  C2 := userargs[3]::double precision; 
  G := userargs[4]::double precision; 
  -- rescale values, optional 
  _value[1] := value[1][1][1] * 0.0001; 
  _value[2] := value[2][1][1] * 0.0001; 
  _value[3] := value[3][1][1] * 0.0001; 
  -- value can't be NULL 
  IF 
    _value[1] IS NULL OR 
    _value[2] IS NULL OR 
    _value[3] IS NULL 
    THEN 
      RETURN NULL; 
  END IF; 
  -- compute numerator and denominator 
  _n := (_value[3] - _value[1]); 
  _d := (_value[3] + (C1 * _value[1]) - (C2 * _value[2]) + L); 
  -- prevent division by zero 
  IF _d::numeric(16, 10) = 0.::numeric(16, 10) THEN 
    RETURN NULL; 
  END IF; 
  RETURN G * (_n / _d); 
END; 
$$ LANGUAGE plpgsql IMMUTABLE; 

If you can't create the function, you probably do not have the necessary privileges in the database.

There are several characteristics required for all of the callback functions. These are as follows:

  • All ST_MapAlgebra() callback functions must have three input parameters, namely, double precision[], integer[], and variadic text[]. The value parameter is a 3D array where the first dimension denotes the raster index, the second dimension the Y axis, and the third dimension the X axis. The position parameter is an array of two dimensions, with the first dimension indicating the raster index, and the second dimension consisting of the X, Y coordinates of the center pixel. The last parameter, userargs, is a 1D array of zero or more elements containing values that a user wants to pass to the callback function. If visualized, the parameters look like the following:
        value = ARRAY[ 1 => 
[ -- raster 1 [pixval, pixval, pixval], -- row of raster 1 [pixval, pixval, pixval], [pixval, pixval, pixval] ], 2 => [ -- raster 2 [pixval, pixval, pixval], -- row of raster 2 [pixval, pixval, pixval], [pixval, pixval, pixval] ], ... N => [ -- raster N [pixval, pixval, pixval], -- row of raster [pixval, pixval, pixval], [pixval, pixval, pixval] ] ]; pos := ARRAY[ 0 => [x-coordinate, y-coordinate], -- center pixel o f output raster 1 => [x-coordinate, y-coordinate], -- center pixel o f raster 1 2 => [x-coordinate, y-coordinate], -- center pixel o f raster 2 ... N => [x-coordinate, y-coordinate], -- center pixel o f raster N ]; userargs := ARRAY[ 'arg1', 'arg2', ... 'argN' ];
  • All ST_MapAlgebra() callback functions must return a double-precision value.

If the callback functions are not correctly structured, the ST_MapAlgebra() function will fail or behave incorrectly.

In the function body, we convert the user arguments to their correct datatypes, rescale the pixel values, check that no pixel values are NULL (arithmetic operations with NULL values always result in NULL), compute the numerator and denominator components of EVI, check that the denominator is not zero (prevent division by zero), and then finish the computation of EVI.

Now we call our callback function, modis_evi(), with ST_MapAlgebra():

SELECT ST_MapAlgebra(rast, ARRAY[1, 3, 4]::int[], -- only use the red, blue a 
nd near infrared bands 'chp05.modis_evi(
double precision[], int[], text[])'::regprocedure,
-- signature for callback function '32BF',
-- output pixel type 'FIRST', NULL, 0, 0, '1.', -- L '6.', -- C1 '7.5', -- C2 '2.5' -- G ) AS rast FROM modis m;

In our call to ST_MapAlgebra(), there are three criteria to take note of, which are as follows:

  • The signature for the modis_evi() callback function. When passing the callback function to ST_MapAlgebra(), it must be written as a string containing the function name and the input-parameter types.
  • The last four function parameters ('1.', '6.', '7.5', '2.5') are user-defined arguments that are passed for processing by the callback function.
  • The order of the band numbers affects the order of the pixel values passed to the callback function.

The following images show the MODIS raster before and after running the EVI operation. The EVI raster has a pale white to dark green colormap applied for highlighting areas of high vegetation:

If you are unable to run the standard EVI operation, or want more practice, we will now compute a two-band EVI. We will use the ST_MapAlgebraFct() function. Please note that ST_MapAlgebraFct() is deprecated in PostGIS 2.1, and may be removed in the future versions.

For the two-band EVI, we will use the following callback function. The two-band EVI equation is computed with the following code:

CREATE OR REPLACE FUNCTION chp05.modis_evi2(value1 double precision, value2 double precision, pos int[], VARIADIC userargs text[]) 
RETURNS double precision 
AS $$ 
DECLARE 
  L double precision; 
  C double precision; 
  G double precision; 
  _value1 double precision; 
  _value2 double precision; 
  _n double precision; 
  _d double precision; 
BEGIN 
  -- userargs provides coefficients 
  L := userargs[1]::double precision; 
  C := userargs[2]::double precision; 
  G := userargs[3]::double precision; 
  -- value can't be NULL 
  IF 
    value1 IS NULL OR 
    value2 IS NULL 
    THEN 
      RETURN NULL; 
  END IF; 
  _value1 := value1 * 0.0001; 
  _value2 := value2 * 0.0001; 
  -- compute numerator and denominator 
  _n := (_value2 - _value1); 
  _d := (L + _value2 + (C * _value1)); 
  -- prevent division by zero 
  IF _d::numeric(16, 10) = 0.::numeric(16, 10) THEN 
    RETURN NULL; 
  END IF; 
  RETURN G * (_n / _d); 
END; 
$$ LANGUAGE plpgsql IMMUTABLE; 

Like ST_MapAlgebra() callback functions, ST_MapAlgebraFct() requires callback functions to be structured in a specific manner. There is a difference between the callback function for ST_MapAlgebraFct() and the prior one for ST_MapAlgebra(). This function has two simple pixel-value parameters instead of an array for all pixel values:

SELECT ST_MapAlgebraFct( 
   rast, 1, -- red band 
   rast, 4, -- NIR band 
   'modis_evi2(double precision, double precision, int[], text[])'::regprocedure,
-- signature for callback function '32BF', -- output pixel type 'FIRST', '1.', -- L '2.4', -- C '2.5' -- G) AS rast FROM chp05.modis m;

Besides the difference in function names, ST_MapAlgebraFct() is called differently than ST_MapAlgebra(). The same raster is passed to ST_MapAlgebraFct() twice. The other difference is that there is one less user-defined argument being passed to the callback function, as the two-band EVI has one less coefficient.