Coordinates for pairs of neighbouring MODIS pixels containing dense and sparse land cover in New South Wales, Australia

This data set relates to detecting changes in areas of native forest in New South Wales, Australia. Under the Australian government Emissions Reduction Fund (ERF) initiative, land owners are able to generate Australian carbon credit units (ACCUs) in exchange for preventing deforestation of native forest for which a clearing permit has previously been issued. To be eligible the native forest must have at least 20\% canopy coverage with tree height greater than two meters. At the commencement of a project an extensive audit is undertaken by an accredited third party and estimates made of the appropriate number of credits to be allocated throughout the project. A number of subsequent audits are also required to ensure compliance throughout the project lifetime. Projects have a permanence period of either 25 or 100 years. During this time period no clearing, with the exception of minimal thinning (< 5 %), is permitted.

At the time of writing there are approximately 400 vegetation projects underway in Australia. A system of automated change alarms based on remotely sensed time series has the potential to significantly reduce the auditing workload and target it to locations where the forest cover has changed and the  project may require reassessment. Continuous monitoring is also important to guarantee the integrity of the carbon credit units in circulation.The region considered occupies the western plains of the state of New South Wales. This is an arid region with the majority of land cover characterized by small shrubs. The forested regions protected under the ERF are majority free standing Eucalyptus trees. The western plains also experiences very low cloud cover making it another ideal location for passive remove sensing.

This data set was created with the intention that it will also be the subject of future studies. Due to the comparative scarcity of actual land cover change in this region the data set was created for the purpose of synthesizing realistic change time series. D=997 MODIS pixels were identified within regions currently assigned to a project which is currently earning carbon credits in exchange for avoided deforestation. These time series span the years of 2008 to 2018. For each pixel a corresponding pair was identified using high resolution imagery provided by Google Earth, DigitalGlobe. The pair was selected such that it is closely located and contains a similar type of land cover but at a lower density. Using this approach it is possible to generate a synthetic change time series by blending between the two while reducing any possible changes in geology, flora, aspect or climate that might occur over a large distance. Blending between high and low density aims to simulate gradual deforestation over the area of the pixel.