231
Remote Sensing and Other Spatial Data
For the 21 and 30 January and 6 and 7 February 2010, relatively cloud free Landsat
Enhanced Thematic Mapper (ETM+) data were downloaded from the Glovis (http://
glovis.usgs.gov/) and Land Processes Distributed Active Archive Centre (https://
lpdaac.usgs.gov/get_data/data_pool). These data were converted to surface reflectance (%) using the 6S atmospheric correction software. Radiometric normalization
between scenes was also undertaken using the approach outlined by Homer et al.
(1997). All images were geo-referenced by first establishing ground control points
between locations identified from existing maps and the most cloud free Landsat
sensor data, warping these images and then generating a mosaic covering the study
area. Control points were then established subsequently between the remaining
Landsat sensor data and this reference mosaic. The processing of the Landsat sensor
data was undertaken using the open source software RSGISLib (www.rsgislib.org;
Bunting et al. 2013).
The history of burning with the IBMTCL over the period 2000–2012 was documented by referencing the MODIS product MOD 14 (Thermal Anomalies – Fires
and Biomass Burning). This product includes fire occurrence (day/night) and location, the logical criteria used for the fire selection and an energy calculation for each
fire. 8-day and monthly day-and-night composite fire occurrence (full resolution)
and gridded 10-km and 0.5° summaries (including counts) per fire class (daily/8day/monthly) are also available. The Level 2 products include various fire related
parameters including the occurrence of day and night time thermal anomalies,
flagged and grouped into different temperature classes associated with different levels of emitted energy from the fire. These parameters are retrieved daily at 1 km
spatial resolution.
An ASTER Global Digital Elevation Model (GDEM) was used (http://asterweb.
jpl.nasa.gov/gdem-wist.asp) to establish the topography of the two national parks,
with this released at 30 m resolution and available for 1 × 1 tiles (geographic projection). Other data layers were also available to support understanding the movement
of large cats, including continuous distance from roads and the locations of antipoaching camps and settlements. The park boundary, roads and rivers were digitized from topographic sheets and Google Earth images, although the boundary
location was revised using the Landsat sensor data available for 2010.
Methods of Data Analysis
Analysis of Camera Trap Data
Each day (24 h) was defined as one sampling occasion (Otis et al. 1978) and this
was repeated over a period of 57 days (Otis et al. 1978; Karanth 1995). Individual
tigers were identified with reference to their documented stripe pattern, with this
Coupling Field Sampling with Earth Observation Increases Understanding of Tiger…
Remote Sensing and Other Spatial Data
For the 21 and 30 January and 6 and 7 February 2010, relatively cloud free Landsat
Enhanced Thematic Mapper (ETM+) data were downloaded from the Glovis (http://
glovis.usgs.gov/) and Land Processes Distributed Active Archive Centre (https://
lpdaac.usgs.gov/get_data/data_pool). These data were converted to surface reflectance (%) using the 6S atmospheric correction software. Radiometric normalization
between scenes was also undertaken using the approach outlined by Homer et al.
(1997). All images were geo-referenced by first establishing ground control points
between locations identified from existing maps and the most cloud free Landsat
sensor data, warping these images and then generating a mosaic covering the study
area. Control points were then established subsequently between the remaining
Landsat sensor data and this reference mosaic. The processing of the Landsat sensor
data was undertaken using the open source software RSGISLib (www.rsgislib.org;
Bunting et al. 2013).
The history of burning with the IBMTCL over the period 2000–2012 was documented by referencing the MODIS product MOD 14 (Thermal Anomalies – Fires
and Biomass Burning). This product includes fire occurrence (day/night) and location, the logical criteria used for the fire selection and an energy calculation for each
fire. 8-day and monthly day-and-night composite fire occurrence (full resolution)
and gridded 10-km and 0.5° summaries (including counts) per fire class (daily/8day/monthly) are also available. The Level 2 products include various fire related
parameters including the occurrence of day and night time thermal anomalies,
flagged and grouped into different temperature classes associated with different levels of emitted energy from the fire. These parameters are retrieved daily at 1 km
spatial resolution.
An ASTER Global Digital Elevation Model (GDEM) was used (http://asterweb.
jpl.nasa.gov/gdem-wist.asp) to establish the topography of the two national parks,
with this released at 30 m resolution and available for 1 × 1 tiles (geographic projection). Other data layers were also available to support understanding the movement
of large cats, including continuous distance from roads and the locations of antipoaching camps and settlements. The park boundary, roads and rivers were digitized from topographic sheets and Google Earth images, although the boundary
location was revised using the Landsat sensor data available for 2010.
Methods of Data Analysis
Analysis of Camera Trap Data
Each day (24 h) was defined as one sampling occasion (Otis et al. 1978) and this
was repeated over a period of 57 days (Otis et al. 1978; Karanth 1995). Individual
tigers were identified with reference to their documented stripe pattern, with this
Coupling Field Sampling with Earth Observation Increases Understanding of Tiger…