Difference between revisions of "Spatial Analysis"

Jump to: navigation, search
Line 15: Line 15:
 
==Cloud Computing for Spatial Analysis==
 
==Cloud Computing for Spatial Analysis==
  
 
+
As the size and complexity of geospatial data grows to include resources like [https://www.digitalglobe.com global scale satellite imagery], the size of some data alone, let alone the computing complexity, has become infeasible on personal computers. Specialized firms, which are often kept on a subscription basis for large consumers like the World Bank, provide data and cloud computing services on that data. For example, a researcher may wonder whether traffic in a certain part of a transport corridor has increased in specific ways; by using satellite data and machine learning on a series of satellite images of the same location, the firm can produce very good estimates of the changes in the number of cars on the road over time. This functionality can extend to land use, settlement quality (roof type), and, where imagery are of high enough resolution, even identifying crop usage on individual agricultural plots.
  
 
==Resources==
 
==Resources==

Revision as of 15:47, 9 November 2017

Introduction

Spatial analysis involves utilizing geographical information to create descriptive and informative outputs. These can be maps or other images and/or statistical data such as distances to other points, boundaries, roads or public services. The development world is rapidly expanding its capacity to access and analyze spatial data ranging from market clustering to satellite imagery to agricultural plot mapping in SurveyCTO, the ability to process this data is also improving quickly.

Interactive Spatial Analysis Tools

Historically, spatial analysis tools have been challenging to code in a reproducible way, due to proprietary software systems with little extensibility. ArcGIS is a prime example of this, as the default implementation can only be coded reproducibly using Python. The open-source equivalent QGIS has similar limitations. Both are useful for exploring spatial data interactively, but poor for saving and reproducing output workflows.

Reproducible Coding for Spatial Analysis

By contrast, programs such as R now support an extensive suite of open-source tools for the reproducible manipulation of spatial datasets. These tools are essential for tasks that involve creating new data – especially when specific formulae or measurements are needed or parameters may want to be varied in subsequent runs.

Stata also supports some commands like spmap for spatial analysis, such as displaying maps with data. However it is not well-equipped to calculate new statistics, such as linking points to underlying political geographies or surrounding features, as it handles non-point information poorly in general.

Cloud Computing for Spatial Analysis

As the size and complexity of geospatial data grows to include resources like global scale satellite imagery, the size of some data alone, let alone the computing complexity, has become infeasible on personal computers. Specialized firms, which are often kept on a subscription basis for large consumers like the World Bank, provide data and cloud computing services on that data. For example, a researcher may wonder whether traffic in a certain part of a transport corridor has increased in specific ways; by using satellite data and machine learning on a series of satellite images of the same location, the firm can produce very good estimates of the changes in the number of cars on the road over time. This functionality can extend to land use, settlement quality (roof type), and, where imagery are of high enough resolution, even identifying crop usage on individual agricultural plots.

Resources