Composite top of atmosphere reflectance image using Landsat 7 ETM+ and Landsat 8 OLI data acquired between November – April each year. Individual Landsat scenes have been masked for cloud and cloud shadow and then combined to produce a gap free image of Tasmania. This image has been produced to facilitate modelling of Foliage Projective Cover and woody change detection work within the Land Cover Program at DPIPWE. This mosaic currently utilises top of atmosphere reflectance imagery with a transformation applied to OLI data. A corresponding dataset, titled Tasmania Landsat Mosaic Date Codes, can be used to identify the source date for each pixel in the scene.

Geometrically and radiometrically corrected imagery from Landsat 7 Enhanced Thematic Mapper Plus (ETM+) and Landsat 8 Operational Land Imager (OLI) was made available to DPIPWE via a partnership with the Queensland Remote Sensing Centre (QRSC). Imagery was obtained in the ‘db8’ processing stage and had top of atmosphere reflectance and bi-directional reflectance distribution corrections applied as per Danaher et al. (2001). Topographic corrections were also applied as outlined in Dymond and Shepherd (1999). Data obtained from OLI was further transformed to replicate the spectral response of ETM+ as per Flood (2014). This transformation provides better accuracy when applying biophysical models which have been parameterised using ETM+ data only (such as Foliage Projective Cover). This data is provided in six band format with the following bands extracted for ETM+; 1,2,3,4,5,7 and the following bands extracted for OLI; 2,3,4,5,6,7 (with the transformations noted in Flood (2014)). A primary or ‘best’ scene (image date) was selected for each Landsat path/row that covers Tasmania. Primary scenes were chosen from those imaged between November and April (inclusive) each year and were selected based on image quality and low amounts of cloud cover. The period November to April was chosen as it presents the highest sun elevations and driest landscape conditions (for discrimination between woody and non-woody vegetation). Automatically generated cloud and cloud shadow masks for each primary scene were then edited to correct for errors of omission and commission. Alternate scenes from the same temporal period were then selected and used to infill areas occluded by cloud and shadow in the primary scene, thus producing a composite nominal scene for each year. References Flood, N. Continuity of Reflectance Data between Landsat-7 ETM+ and Landsat-8 OLI, for Both Top-of-Atmosphere and Surface Reflectance: A Study in the Australian Landscape. Remote Sensing 2014, 6(9), 7952-7970 Danaher, T., Wu, X., Campbell, N. Bi-Directional reflectance distribution function approaches to radiometric calibration of Landsat ETM+ imagery. In Proceedings of the IEEE International Geoscience and Remote Sensing Symposium 2001 (IGARSS ’01), Sydney, Australia, 9–13 July 2001; Volume 6, pp. 2654–2657. Dymond, J., Shepherd, J. Correction of the topographic effect in remote sensing. IEEE Transactions on Geoscience & Remote Sensing 1999, 37, 2618–2619.

Positional Accuracy: All Landsat imagery used in this dataset has been geometrically corrected by the United States Geological Survey (USGS). The USGS indicates that locational error for Landsat 5-8 is typically below a single pixel (30m x 30m) (https://landsat.usgs.gov/geometry). The user should note that King Island typically lies in zone 54 for the Map Grid of Australia projection but has been included in zone 55 (as per mainland Tasmania) for ease of use and distribution of the dataset. Attribute Accuracy: Digital number values represent scaled (8bit) top of atmosphere reflectance. The following gains and offsets were used for bands 1 -6: Gain = [670, 325, 345, 290, 425, 275] Offset = [-30, 0, 0, 5, 5, 5] The following equation can be used to retrieve reflectance values: Reflectance (0-1) = (DN – offset[band])/gain[band] Were DN is the Digital Number value stored in the dataset. Logical Consistency: The dataset has been checked to ensure that all pixels were derived from Landsat imagery between the months November – April only. Completeness: This dataset covers the entire state of Tasmania; including most large islands. The dataset has been clipped using a 3km smoothed coastal buffer. Credit Landsat 7 ETM+ and Landsat 8 OLI images acquired courtesy of U.S. Geological Survey Image processing and correction courtesy of Queensland Remote Sensing Centre - Department of Science, Information Technology and Innovation and the Joint Remote Sensing Research Program.