Introduction. Introduction. Introduction. Introduction. Introduction

Transcription

1 Identifying habitat change and conservation threats with satellite imagery Extinction crisis Volker Radeloff Department of Forest Ecology and Management Extinction crisis Extinction crisis Conservationists need to know Which species are declining How fast declines are Where declines occur, and Why species are declining Biodiversity hotspots 1

2 Biodiversity hotspots Biodiversity hotspots The challenge is There are thousands of species Many species are hard to monitor Extinctions occur all over the globe Reasons for extinctions differ widely and there isn t much time One solution is to monitor habitat change with satellite imagery First step: identify what constitutes good habitat for a given species or species group Second step: map habitat change using a satellite image time series Objectives to change detection How to get started Remote Sensing Information is acquired ( sensed ) without direct contact ( remotely ) 2

3 History Earliest aerial photographs taken around 1860 Troll (1939) uses aerial photographs for Landschaftsökologie (landscape ecology) 1960 s first photographs from space taken by astronauts 1972 first earth observation satellite: Landsat commercial satellites, 1x1 m resolution Astronaut photos A selection of the 375,000 photos that astronauts took during various missions is available at: Optical satellite sensors measure the reflectance of sunlight in different wavelengths Different materials have different reflectance curves 3

4 Spectral Reflectance Patterns High Dry bare soil Vegetation Water Spectral Reflectance Low λ Wavelength 4

5 Spectral Resolution: Thematic Mapper Visible Near-IR Mid-IR Thermal-IR Wavelength (µm) Multi (30 m) Thermal (120 m) 6 Image Classification Image Classification Band 2 Digital Numbers 200 Vegetation Water Band 1 Digital Numbers 5

6 Band 2 Digital Numbers 200 Vegetation Water Band 1 Digital Numbers Water Vegetation Land cover classification Forest versus non-forest classifications reach > 90% accuracy Deciduous very coniferous forest classifications reach > 80% accuracy Single tree species classifications often as low as 60% accuracy Different optical satellites have different resolutions Spatial resolution (pixel size) Temporal resolution (how long until an area is captured again) Spectral resolution (how many wavelengths bands) Radiometric (how many different levels of reflectance values can be separated in each wavelength band) to remote sensing Spatial resolution 6

7 Spectral resolution Temporal resolution Commonly used optical satellites NOAA Advanced Very High Resolution Radiometer (AVHRR), U.S. Government Landsat Thematic Mapper (TM), U.S. Government SPOT, French semi-private company IKONOS, U.S. commercial satellite NOAA- AVHRR Landsat TM SPOT IKONOS Pixel size 1000 m 30 m 10 m 1m Repeat cycle 12 hours 16 days 5-26 days > 3 days Number of bands (visible to mid-ir) (4) to change detection 7

8 8

9 to change detection Change detection Satellite images have been taken repeatedly over the last 25 years All these images are stored in huge data archives This allows monitoring land cover change over large areas to change detection Clear cutting on Vancouver Island to change detection Clear cutting on Vancouver Island to change detection Change detection algorithms The easiest approach is to make two land cover classifications and compare them Accurate monitoring of land cover change is usually based on change detection techniques Special computer algorithms compare the reflectance in the two images to identify change Wildlife habitat appclications 9

10 Highly preferred areas: dark green Moderately preferred areas: light green Un-preferred areas: pale yellow 1989 was a dry La Niña year 1993 was a wet El Niño year Elk habitat is much more restricted in 1989 compared to

11 Gray areas are forested, white areas are nonforested Highly suitable habitat: red; suitable: yellow; marginally suitable: green; and unsuitable in black Change in panda habitat in Wolong Nature Reserve before and after establishment in (A) Highly suitable habitat (B) suitable habitat (C) marginally suitable habitat, and (D) unsuitable habitat. Advantages and disadvantages??? Advantages of satellite imagery for habitat analysis Identifies habitat loss and fragmentation, major drivers of species extinctions Rapid and cheap assessments for large areas Historic data readily available Disadvantages of satellite imagery for habitat analysis Fine-scale habitat features (e.g., snags) are not detected Good habitat may be empty, e.g., where poaching is rampant Analysis requires highly-trained personnel with good computers How to get started 11

12 How to get started How to get started Remote sensing classes on campus Environmental Remote Sensing Center (ERSC) IES 301 to Aerial Photographic Systems IES 302 to Electro-optical optical and Microwave Remote Sensing Systems IES 303 to Digital Image Processing IES 377 to Geographical Information Systems IES 556 Remote Sensing Digital Image Processing IES 758 Image Processing Algorithms for Remote Sensing How to get started Online tutorials NASA Canada Center for Remote Sensing Remote Sensing Core Curriculum c.org/ Please feel free to me at Appendix Landsat Satellite data availability Landsat data for free /index.shtml Landsat data for almost free Landsat data archives and much more