Remote Sensing and Gis for Ecologists: Using Open Source Software - Brossura

Brossura

ISBN 10: 1784270229 ISBN 13: 9781784270223

Casa editrice: Pelagic Pub Ltd, 2016

Riassunto

This is a book about how ecologists can integrate remote sensing and GIS in their daily work. It will allow ecologists to get started with the application of remote sensing and to understand its potential and limitations. Using practical examples, the book covers all necessary steps from planning field campaigns to deriving ecologically relevant information through remote sensing and modelling of species distributions.

All practical examples in this book rely on OpenSource software and freely available data sets. Quantum GIS (QGIS) is introduced for basic GIS data handling, and in-depth spatial analytics and statistics are conducted with the software packages R and GRASS.

Readers will learn how to apply remote sensing within ecological research projects, how to approach spatial data sampling and how to interpret remote sensing derived products. The authors discuss a wide range of statistical analyses with regard to satellite data as well as specialised topics such as time-series analysis. Extended scripts on how to create professional looking maps and graphics are also provided.

This book is a valuable resource for students and scientists in the fields of conservation and ecology interested in learning how to get started in applying remote sensing in ecological research and conservation planning.

Su questo libro

Informazioni sull?autore

Martin Wegmann has a PhD in remote sensing focusing on time-series analysis on land cover change and fragmentation in Africa. He is an assistant professor at the Global Change Ecology Msc program at the University of Würzburg, Germany and runs courses in remote sensing analysis for biodiversity and conservation.

Benjamin Leutner is a research assistant at the department of remote sensing at the University of Würzburg. He has extensive experience in geo-spatial analysis of remote sensing data using Open Source software.

Stefan Dech is director of the German Remote Sensing Data Center (DFD) since 1998, and current spokesman of the Earth Observation Center (EOC) at the German Aerospace Center (DLR). Since 2001 he has held the Chair for Remote Sensing at the Institute of Geography and Geology of the University of Würzburg.

Estratto. © Ristampato con autorizzazione. Tutti i diritti riservati.

Remote Sensing and GIS for Ecologists

Using Open Source Software

By Martin Wegmann, Benjamin Leutner, Stefan Dech

Pelagic Publishing

Copyright © 2016 Martin Wegmann, Benjamin Leutner and Stefan Dech
All rights reserved.

ISBN: 978-1-78427-022-3

Contents

List of Contributors, vii, 
Foreword – Woody Turner, ix, 
Preface – Martin Wegmann, Benjamin Leutner and Stefan Dech, x, 
List of Acronyms, xvi, 
Introduction Nathalie Pettorelli, 1, 
1 Spatial Data and Software Benjamin Leutner, Ned Horning, Duccio Rocchini and Martin Wegmann, 11, 
2 Introduction to Remote Sensing and GIS Martin Wegmann and Benjamin Leutner, 22, 
3 Where to Obtain Spatial Data? Martin Wegmann and Benjamin Leutner, 40, 
4 Spatial Data Analysis for Ecologists: First Steps Benjamin Leutner, Martin Wegmann, Mirjana Bevanda and Ned Horning, 60, 
5 Pre-Processing Remote Sensing Data Benjamin Leutner and Martin Wegmann, 114, 
6 Field Data for Remote Sensing Data Analysis Christian Wohlfart, Mirjana Bevanda, Ned Horning, Benjamin Leutner and Martin Wegmann, 136, 
7 From Spectral to Ecological Information Duccio Rocchini, Benjamin Leutner and Martin Wegmann, 150, 
8 Land Cover or Image Classification Approaches Ned Horning, Benjamin Leutner and Martin Wegmann, 166, 
9 Land Cover Change or Change Detection Ned Horning, Benjamin Leutner and Martin Wegmann, 197, 
10 Continuous Land Cover Information Ned Horning, Benjamin Leutner, Mirjana Bevanda and Martin Wegmann, 209, 
11 Time Series Analysis Jan Verbesselt, Fabian Loew, Christian Wohlfart and Martin Wegmann, 223, 
12 Spatial Land Cover Pattern Analysis Duccio Rocchini, Martin Wegmann, Benjamin Leutner and Mirjana Bevanda, 244, 
13 Modelling Species Distributions Björn Reineking, Benjamin Leutner and Martin Wegmann, 258, 
14 Introduction to the added value of Animal Movement Analysis and Remote Sensing Mirjana Bevanda, Kamran Safi, Martin Wegmann and Benjamin Leutner, 295, 
Outlook and Acknowledgements, 317, 
Index, 319, 

CHAPTER 1

Spatial Data and Software

Benjamin Leutner, Ned Horning, Duccio Rocchini and Martin Wegmann

1.1 What is Geospatial Data?

Geospatial data are information that can be pinpointed to spatially explicit locations on Earth. Most of the data you sample in ecology are of geospatial nature, regardless of whether you recorded the spatial coordinates during data collection or not. In principle, a geospatial data element consists of two parts: (1) spatial coordinates in a defined coordinate system, such as latitude and longitude; and (2) one or more values, such as a label, a physical measurement or a species observation associated with this location (Figure 1.1).

The analysis of geospatial data has been at the core of ecology since its very beginning. Answering the fundamental ecological question "What do we find where and why?" requires an explicit investigation into spatial relationships, processes and patterns.

This book will get you started with the basic geospatial analysis techniques you need to approach these questions. What is the distance to the border of the PAs? Which points are located close to a road? What degree of tree cover and how much land cover change exist in my study area? Combining the geospatial data of different sources, such as field observations and remote sensing data, will go a long way in finding answers to these questions.

1.2 Tools

Within this book we will mainly focus on two different packages of GIS-capable software: R and QGIS. Both programs are platform independent and free, including in commercial environments, due to their OS licensing under the GNU General Public License (GPL). However, for some analyses, other software packages such as GRASS (Geographic Resources Analysis Support System), SAGA (System for Automated Geoscientific Analyses) or Orfeo ToolBox (OTB) will be more appropriate, in which case we will point you to the corresponding program and function.

QGIS is a program similar to proprietary GIS packages and offers a GUI with a variety of GIS functionalities. R, on the other hand is purely based on a command-line interface (CLI) that forces you to program your own analysis in a script. While this can be less intuitive at first, it pays out quickly, at the latest when you have to redo your analysis. If you have not programmed or worked with the command line so far, you will hopefully enjoy using it after reading this book. You will have to work continuously on your coding skills and fight the frustrating bits and pieces. However, you will feel greatly rewarded having also managed to solve problems and finish off your script.

We will cover a variety of functions that will guide you from data import, generation and inspection in QGIS to more advanced data modification, analysis and visualization in R. R will be used extensively due to its statistical and spatial data manipulation capabilities.

1.2.1 QGIS

QGIS is a very user-friendly GIS available for free from http://qgis.org. It provides an appealing GUI and is the most intuitive software package used in this book – the ideal GIS to get you started with spatial analysis (Figure 1.2). The QGIS GUI is especially helpful for interactive data analysis, such as data display, digitization or map generation.

QGIS, however, is more than a simple GUI. It has fully fledged support for high-end geospatial data processing with all the functionality you would expect from a modern GIS. While you can achieve everything in the GUI, QGIS also provides a graphical model builder to generate program routines for users not familiar with a scripting language. More importantly, it also includes a scripting console for the Python programming language. Furthermore, QGIS provides interfaces to R, GRASS, SAGA and other software packages so you can use their additional functionality within QGIS.

In addition, QGIS comes with a plug-in architecture for which users have contributed a variety of extensions. The set of available extensions or plug-ins is rapidly evolving as is QGIS itself. QGIS comes with a plug-in manager that makes it easy to browse and install plug-ins from a central repository (Figure 1.3).

Some useful and recommended plug-ins include:

• GPStools: importing and modifying GPS data;

• fTools: vector analysis and management;

• OpenLayers Plugin: display of Google Earth, OpenStreetMap, Bing Maps, and so on;

• Georeferencer GDAL: georeference raster using GDAL;

• Processing: interfaces to GRASS, OTB, R, SAGA.

QGIS has a very high development pace: new releases or candidates are announced every couple of months. Three QGIS versions, 2.2, 2.3 and 2.4, were released in 2014 alone. Odd-numbered versions are development versions while even numbers indicate stable release versions.

QGIS provides an extensive online archive of help pages and tutorials. Go to Help > Help Contents or hit "F1" in QGIS and your browser will be directed to the user guide for your current version of QGIS.

SAGA, GRASS and OTD in QGIS

QGIS is increasingly capable of calling functions available in different packages. This increases the functionality of QGIS tremendously while maintaining a simple and intuitive user interface (Figure 1.4). It is a very good starting point to use SAGA, GRASS or OTB through QGIS, however, you might want to work directly with these packages at a later stage, which would allow more automated analysis.

1.2.2 R

The program R, according to its own definition "a language and environment for statistical computing and graphics", is available from http://www.r-project.org/. Over the last decade, R has become one of the most frequently used tools for both classical and cutting-edge statistical analysis across many research disciplines. Although R started off as a statistical language, it has evolved into a multipurpose programming environment with comprehensive GIS functionality. R offers an incredible number of extensions called packages (http://cran.r-project.org/web/packages/), which are contributed by its community. These packages cover all aspects of data manipulation, statistics, data mining or plotting. On the R website you will also find curated and commented package collections, so-called task views (cran.r-project.org/web/views), for specific disciplines and applications, which are a good starting point to get an overview of available packages.

R does not come with a GUI like QGIS. Since R is based on the command line, all you need is a text editor to write down and save your commands in script files. If you are a beginner, R will set you off on a steep learning curve, which can be challenging at first. R consists of highly complex scientific software packages that cannot be conquered within a few days. Especially if you have never worked using the command line before you may be intimidated by the thought of writing program code instead of clicking on options and tabs in a GUI environment. Why is it a good idea to use command-line scripting anyway?

• First, when you are done with the analysis you have a complete protocol of what you did. Have you ever tried to remember all the steps you took half a year ago using a GUI when you actually did your analysis?

• Second, because of that script you can rerun the full analysis again with a single click without having to click through a whole range of GUI options. Have you ever discovered an error in your input data forcing you to repeat the whole analysis?

• Third, it is easy to share your script with other researchers – an important step towards reproducible research. Have you ever tried to explain that complex GUI workflow to your colleague: "First you click on Vector then on Analysis Tools then on ...".

R is used as a high-level programming language, which means that you do not have to go down to the intricate details of compiled language programming, such as C++ or Fortran, although you can link to such code easily. Rather, it provides an abstracted set of functions that are often easy to read and understand even for non-programmers.

Several programs, so-called integrated development environments (IDEs), provide convenient editors specialized in writing and executing R code; we highly recommended that you make use of one of these. Currently, the most popular and most beginner-friendly IDE for R is RStudio, which you can download for free from http://www.rstudio.com. Another free option is Eclipse (http://www.eclipse.org), to be used with the Eclipse-based StatET plug-in (http://www.walware.de/goto/statet).

If you start up RStudio, it will launch an instance of R (Figure 1.5). The upper left panel shows the script editor where you write your code, with the command line below and on the right several tabs that list your files, plots, R help packages and the package manager. Code from the script editor can be sent to R using the run button or by pressing CTRL+ENTER. Only the current line or highlighted parts are sent to R and executed. Commenting and uncommenting, that is, preventing a command or text from being executed, is done using the CTRL-SHIFT-C shortcut (or prefacing the code manually with #). Hitting F1 while your cursor is on a command will open the help page for this command.

We assume that you have already been exposed at least to the basics of general R programming, because doing so is outside of the scope of this book. If you have not, please take a few hours and familiarize yourself with the basic concepts and methods. A good starting point is the introductory text R for Beginners (http://cran.r-project.org/doc/ contrib/Paradis-rdebuts_en.pdf). While using R, make sure to read the documentation when you use a new function (see ?newfunction or help("newfunction")). Most functions come with a set of default settings that you should be aware of.

While writing your scripts, try to write code that can be read and understood by humans, with excessive commenting and proper indentation. Make it a habit to note down comments about what steps are currently performed, even if you think it is obvious.

Your future self will be thankful when you have to go back to a script in a year's time and try to understand what you were doing back then. A good script should look like this:

# script short description
# #############
# Purpose: Import GPS data and convert it to shp format.
# Author: your name
# Date: month, year
# R version and packages: x.xx
# ##############
# ##############
# Import format: csv, tab delimited
# Output format: shp
# ##############

# Load required packages
library(sp)

# Import the csv with x and y coordinates
gps_in <- read.table("/path/to/mytrack.csv")

# #### data integrity check
# check first entries
head(gps_in)

# overview statistics
summary(gps_in)

# plot data
plot(gps_in)

# data analysis

# ### data conversion for ...
...

Turning R into a GIS

R starts with a minimal set of commands embedded in default packages. To expand the functionality of R you need to install additional packages using the command install. packages(). When prompted to pick a repository, simply select one in your area. After the installation you have to load the required package at the beginning of each session using the library() command, otherwise R cannot use its functionality. For the raster package, the whole process would look like this:

install.packages ("raster") # execute only once to install a package
library ("raster") # execute every time you start a new R session

If you ever come across an error stating Error: could not find function "function_name", you have either misspelled the function or simply forgotten to load the package providing this function. Your package library can be updated using update.packages().

The main packages we rely on to turn R into a fully fledged GIS are:

• raster: The raster package provides representations for spatial raster data (see Chapter 2) with support for large rasters that exceed the size of your computer's memory. Moreover, raster provides powerful GIS functions both for raster and vector data (see Chapter 2).

• rgdal: The rgdal package provides functions for import and export of spatial data in a variety of formats, for example, shapefile, Keyhole Markup Language (KML) or GeoTIFF and projection transformations (see Chapter 3). rgdal interfaces the GDAL and PROJ.4 external libraries.

• RStoolbox: The RStoolbox package provides functions for remote sensing data analysis. This is a very recent package that is being developed in parallel with this book to facilitate the use of R in remote sensing data analysis. This package provides various common remote sensing functions such as Landsat data processing or land cover classification.

• sp: The sp package is automatically loaded by raster and provides fundamental spatial object classes like SpatialPoints, SpatialPolygons or SpatialPixels, as well as projection definition and handling (see Chapter 3), and some data management and querying. The spatial classes of the sp package are intended as the underlying representation of spatial objects in R.

• rgeos: The rgeos package provides functions for complex GIS operations like buffering or intersecting vector data, and interfaces the external Geometry Engine – Open Source (GEOS) library.

To get an overview of even more spatial packages, check the "Spatial" task view for packages dealing with spatial data. There is a specialized help list at R-sig-Geo, where you can post questions concerning spatial data analysis in R (https://stat.ethz.ch/ mailman/listinfo/r-sig-geo).

1.2.3 Other OS GIS Software

The majority of tasks shown in this book can easily be achieved with QGIS and R. However, certain functions might not (yet) be implemented, the performance is not sufficient, or other software packages might provide different or more sophisticated methods. Of course, many other software packages exist that are useful for ecologists working in a spatial context such as:

• SAGA GIS (http://www.saga-gis.org/en/index.html): A comprehensive standalone GIS with a modular plug-in structure, SAGA provides a large suite of functions for terrain analysis, hydrological modelling and geostatistics. SAGA functions can be accessed from QGIS.

• gvSIG Desktop (http://www.gvsig.com/en/products/gvsig-desktop/downloads): This is a comprehensive stand-alone GIS developed for administrations, which is well suited for interactive data exploration and analysis.

• uDig (http://udig.refractions.net): A stand-alone, user-friendly GIS that focuses on online data sources.

• OTB (https://www.orfeo-toolbox.org): Remote sensing image processing software (stand-alone or application programming interface (API)) from the French space agency (Centre national d'études spatiales (CNES)). Orfeo modules are available from within QGIS.

• Opticks (http://opticks.org/confluence/display/opticks/Welcome+To+Opticks): Stand-alone remote sensing image processing software for multi- and hyperspectral imagery.

• GRASS GIS (http://grass.osgeo.org): GRASS or GRASS GIS (Neteler et al., 2012) is a hybrid software programme combining GIS and remote sensing functionality. It is one of the biggest and most powerful raster manipulation OS systems available. GRASS is based on a sophisticated database system, which is well adapted to multiuser environments.

(Continues...)

Excerpted from Remote Sensing and GIS for Ecologists by Martin Wegmann, Benjamin Leutner, Stefan Dech. Copyright © 2016 Martin Wegmann, Benjamin Leutner and Stefan Dech. Excerpted by permission of Pelagic Publishing.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.

Dettagli dell?edizione

• Editore: Pelagic Pub Ltd
• Data di pubblicazione: 2016
• Lingua: Inglese
• ISBN 10: 1784270229
• ISBN 13: 9781784270223
• Rilegatura: Copertina flessibile
• Numero di pagine: 324
• Redattore: Wegmann Martin, Leutner Benjamin, Dech Stefan
• Contatto del produttore: non disponibile
• Persona responsabile: non disponibile