Estimating change in areas of indigenous vegetation cover in New Zealand from the New Zealand Land Cover Database (LCDB)

Abstract

Four iterations of the New Zealand Land Cover Database have been produced from satellite imagery for nominal dates of 1996/97, 2001/02, 2008/09 and 2012/13. These data may be used to estimate changes in area for land cover classes of interest. However, these estimates are subject to uncertainty, which can be significant, particularly when change in area is small. Changes in indigenous vegetation classes are of interest for a number of applications, including monitoring threatened environments. Here we show how the combination of exhaustive sampling of change polygons with random 'truth' sampling can be used to estimate the uncertainty of area change. We demonstrate the method on five important indigenous covers: indigenous forest, broadleaved indigenous hardwoods, manuka and/or kanuka, tall tussock grassland, and subalpine shrubland. For these classes, we estimate their area in 2008/09 and the change of area between 2001/2002 and 2008/09. Areas were estimated to within plus or minus 5%. Change in areas were estimated to within plus or minus 10% of change for classes with a large change in area, and to within plus or minus 30% for the classes with a small change in area. We anticipate similar uncertainties for estimated changes in area between the other dates and for other classes. The number of random 'truth' samples required for this assessment was very high, in excess of 30 000. Many more samples would be required to further lower the uncertainties.

Geographic versus environmental space: Patterns of potential native and exotic woody propagule pressure on New Zealand's indigenous grasslands

“…) and small changes may not be recorded correctly (Dymond et al . ). Some transitions from grasslands to shrublands might have happened but they were not obvious enough to be captured by the remote sensing methodology applied in the current LCDB datasets.…”

“…LCDB's information is derived from satellite imagery using manual delineation and automatic detection to improve accuracy (Dymond et al . ). Even though this dataset has limitations for land cover change detection (Weeks et al .…”

“…; Dymond et al . ), we consider it as adequate for our propagule pressure analyses at different spatial scales as we are not utilising any between‐year change data in our study. The vector‐formatted LCDB data were rasterised to 1 km grid resolution using the module ‘Polygon to Raster’ with the ‘cell‐centre’ assignment option in ArcGIS 10.1.…”

“…Our analyses are based on the most recent (2012/2013) New Zealand Land Cover Database (LCDB-v4.1 2015) as land cover dataset. LCDB's information is derived from satellite imagery using manual delineation and automatic detection to improve accuracy (Dymond et al 2017). Even though this dataset has limitations for land cover change detection Dymond et al 2017), we consider it as adequate for our propagule pressure analyses at different spatial scales as we are not utilising any between-year change data in our study.…”

“…LCDB's information is derived from satellite imagery using manual delineation and automatic detection to improve accuracy (Dymond et al 2017). Even though this dataset has limitations for land cover change detection Dymond et al 2017), we consider it as adequate for our propagule pressure analyses at different spatial scales as we are not utilising any between-year change data in our study. The vector-formatted LCDB data were rasterised to 1 km grid resolution using the module 'Polygon to Raster' with the 'cell-centre' assignment option in ArcGIS 10.1.…”