Methods - how we monitor
We measure indigenous forest, scrub and shrubland and tussock grassland vegetation for the entire Waikato region using the Land Cover spatial database which is based on satellite imagery. We report some of this at the district council level and by bioclimatic zone1.
Waikato Regional Council's monitoring of changes in indigenous vegetation is dependent on the release of updated LCDB layers by Landcare Research NZ. To date these releases have varied between three and nine years apart.
Historic (about 1840) vegetation extent is available as a single estimate of pre-European modification using the RIVI (1840).
The LCDB (recent) data were derived from SPOT satellite imagery collected over four time periods:
- summer 1996/97
- summer 2001/02
- summer 2008/09 and
- summer 2011/12.
The RIV1 (1840) data were derived from:
- early settler accounts and maps
- early aerial photographs
- interpretation based on the current extent of indigenous forests.
We used the:
- Landcover Database (LCDB Version 4.0 and the LCDB change layer) clipped to Waikato region land boundary to estimate the regional extent of indigenous vegetation since 1996. The LCDB minimum mapping unit is 1 ha and the data are suitable for applications down to 1:25,000 scale.
- RIVI (1840) to estimate the extent of indigenous vegetation around 1840. The RIVI (1840) maps patches of vegetation to a nominal size unit of 50 ha for secondary vegetation and 25 ha for primary vegetation. Many patches below these limits were also included where their composition could be determined.
We used FME software to calculate the extent of indigenous terrestrial vegetation within the Waikato region, and by district council area (2012 boundaries).
Gross change in indigenous vegetation extent since European settlement is calculated by comparing the RIVI (1840) data with commensurate vegetation classes in the LCDB. To achieve this, the RIVI (1840) indigenous vegetation classes have been combined into indigenous forest, scrub and shrubland and tussock grassland classes for comparison with the LCDB classes (see guidelines and standards below).
For the current extent of indigenous vegetation the following LCDB classes are used:
- Indigenous forest
- Scrub and shrubland (aggregation of the following five LCDB classes: Broadleaved Indigenous Hardwoods, Fernland, Manuka and / or Kanuka, Matagouri or Grey Scrub and Sub Alpine Shrubland).
- Tussock grassland
The assumption is made that the scrub and shrubland classes included in this indicator are either mostly indigenous in character, or will develop into mostly indigenous vegetation in a relatively short time.
The RIVI (1840) classification of vegetation types is aggregated into seven types comparable with the LCDC land cover classes. The table below shows the aggregated class name derived from the RIVI (1840).
|Regional indigenous vegetation inventory||Aggregated class name|
|Beech forest||Indigenous forest|
|Coastal broadleaved forest||Indigenous forest|
|Conifer forest||Indigenous forest|
|Conifer-broadleaved forest||Indigenous forest|
|Conifer-broadleaved-beech forest||Indigenous forest|
|Kauri-conifer-broadleaved forest||Indigenous forest|
|Mixed conifer- broadleaved /conifer- broadleaved -beech forest||Indigenous forest|
|Mixed kauri/conifer-broadleaved forest||Indigenous forest|
|Mixed kauri/taraire/conifer-broadleaved forest||Indigenous forest|
|Montane conifer-broadleaved forest||Indigenous forest|
|Primary scrub and shrubland above treeline||Indigenous scrub and shrubland|
|Secondary vegetation on alluvial - Coromandel||Indigenous scrub and shrubland|
|Secondary vegetation on alluvial - Lower Waikato||Indigenous scrub and shrubland|
|Secondary vegetation on domes - Taupō Basin||Indigenous scrub and shrubland|
|Secondary vegetation on hillcountry - Coromandel||Indigenous scrub and shrubland|
|Secondary vegetation on hillcountry - Lower Waikato||Indigenous scrub and shrubland|
|Secondary vegetation on plateaux - Taupō Basin||Indigenous tussock grassland|
|Tongariro ringplain||Indigenous tussock grassland|
This indicator is limited to the Waikato region. The Waitomo, Rotorua and Taupo districts all extend beyond the regional boundary. Therefore, Waikato Regional Council's information on this indicator may not represent the entire area of these three districts.
This indicator replaces, rather than updates, the previous indicator for extent of indigenous vegetation published by the Waikato Regional Council in 2001. This updated indicator amends earlier reported data because of:
• local authority boundary adjustments since the previous indicator was calculated
• corrected land cover classifications in the 1996 LCDB layer used in the 2001 indicator.
The 1840 vegetation map (RIVI) was created for the Waikato Region in 1994 based on the regional boundary at that time. In 2010 the region expanded northwards into part of the former Auckland Region, in accordance with the Local Government (Auckland Council) Act 2009, and for those newly acquired areas there is no 1840 vegetation data. This may slightly affect comparison with recent vegetation, however the changes have been so great any effect on the resultant data will be minor.
The LCDB gives a ‘snapshot’ of vegetation when the data were collected and should not be considered the definitive current vegetation cover. The LCDB databases have classification errors estimated at plus or minus 4 per cent.(LCDB v3.0 Accuracy Assessment). Small changes in vegetation cover may fall within the margin of error. Some visual checks were performed on larger areas of apparent loss to confirm the validity of the magnitude of change reported for 1996-2012 in this indicator.
Visual assessment of air photographs from circa 1995 and other images captured around 1996 or early 2000’s suggest that over 40%, and possibly as many as 70% of the polygons reclassified from Indigenous forest in 1996 to another land cover type in later years were incorrectly classified in LCDB1. This included one polygon of 103 ha, classified as Indigenous forest in 1996 that was already in established plantation forest by that time. The actual amount of indigenous forest converted to another land use is thought to be around 10-20% of the amount calculated from the LCDB Change layer. The visual analysis of 1297ha of scrub and shrubland classes reclassified as another class by 2012, and 1241ha of other classes reclassified as scrub and shrubland in 2012, support a net loss of about 500ha of scrub and shrubland, being 75% of the amount calculated using the LCDB Change layer. Despite these potential errors, to provide for consistency nationwide, this indicator reports the current LCDB statistics, being the best data set available to the indicator compilers at the time. Polygons with suspected mis-classifications have been forwarded to the LDCB administrators for consideration in future releases of the database.
The 1840 vegetation was mapped at a scale of 1:250,000, compared with the LCDB scale of 1:25,000 or greater. This means that many smaller areas of any given vegetation type will not be included in the RIVI (1840), but will have been aggregated into the local dominant vegetation type. For instance, the extent of tussock grassland is overestimated, as these areas were known to have patches of scrub and shrubland within them that have not been able to be mapped at that scale. Such generalisation of the vegetation types precludes analysis at finer scales such as bioclimatic zone for the 1840 data and only allows for gross level of change analysis. While there will be errors in comparing the databases, the changes between the two time periods (1840 and relatively recent) have been so great that they give a satisfactory indication of the regional gross change in vegetation since European settlement.
The RIVI and LCDB datasets use different vegetation classification systems. The Land Cover Database for instance has a single generic classification for indigenous forest, while the RIVI has a more detailed classification to forest type (for example, tawa-rimu forest, kahikatea forest). To allow for meaningful comparison between the 1840 and more recent databases, it has been necessary to aggregate vegetation classes to structure level (forest, scrub and shrubland and tussock grassland). This will also reduce inaccuracies from misclassification error, which is more likely to occur within a structural group than between vegetation structures.
Some vegetation classes can occupy terrestrial or wetland systems, e.g. flaxland, manuka, or fernland. For the purpose of this analysis of terrestrial indigenous vegetation, flaxland was treated as being predominately wetland, and excluded from this indicator, while manuka and fernland were treated as predominantly terrestrial and included. Fernland is not structurally a scrub or shrubland class, however its amount is relatively small and being predominantly terrestrial has been included in the scrub and shrubland category.
The LCDB ‘scrub and shrubland’ classes include areas of vegetation that are likely tall and mature enough to be considered forest, however the LCDB does not separate areas of Broadleaved Indigenous Hardwood that are forest from those that are scrub or shrubland. Broadleaved Indigenous Hardwoods are described in the Illustrated Guide to Target Classes for LCDB22 as being predominantly taller than 3 m. This would technically (per Atkinson 1985)3 make these polygons areas of indigenous forest. However the “Table 1: Land Cover correlations between LCDB2 and LCDB3” treats them as scrub and shrubland. To avoid confusion for future developers of this indicator, the Broadleaf Indigenous Hardwoods polygons were aggregated into the scrub and shrubland category.
Likewise, manuka is typically scrub and shrubland classes, while kanuka is typically a forest class, however these vegetation types are mapped as a single class under the scrub and shrubland category in the LCDB series and cannot be separated for this indicator. Therefore it is suggested they be treated as predominantly scrub and shrubland classes, as per the “Table 1: Land Cover correlations between LCDB2 and LCDB3”. As a result, the current extent of scrub and shrubland is likely to be overestimated and the area of indigenous forest underestimated.
Quality control procedures
For information on data quality (lineage, positional accuracy, attribute accuracy, logical consistency and completeness) see the updated metadata for the LCDB available on the Landcover Database website - Land Cover Database (metadata).
For detail on the methods used to reconstruct 1840 vegetation in the Waikato region see Leathwick et al., 1995.4
Waikato Regional Council is developing spatial layers of indigenous vegetation using aerial photography and a slightly simplified version of the LCDB (1 and 4) classifications to improve our ability to monitor change in vegetation extent. The resulting layer (Bioveg) is likely to provide more accurate data in future iterations of this indicator.
Manual analysis of low level colour aerial photos, will provide greater level of accuracy of classification and larger scale imaging techniques, allowing for more accurate boundary placement and area measurement.
- Bioclimatic zones were developed for New Zealand (from international conventions) by Meurk (1984) and refined for the Waikato Region by Leathwick et al. (1995)
- Thompson, S. Gruner, I. and Gapare, N. 2003. New Zealand Landcover Database Version 2. Illustrated guide to target classes. Ministry for the Envrionment. New Zealand Government, Wellington.
- Scrub and shrubland include areas of regenerating vegetation such as manuka and mixed broadleaved hardwood scrub, as well as other low stature vegetation such as fernland and sub-alpine shrubland. Wetland shrubland classes such as flaxland have been excluded from this indicator. 1985 Derivation of the Vegetation Mapping Units for an Ecological survey of Tongariro National Park, North Island, New Zealand. NZ J. Botany, 23:361-378.
- Leathwick, J. Clarkson, B. and Whaley, P. 1995: Vegetation of the Waikato Region: Current and Historic Perspectives. Landcare Research Contract Report LC9596/022. Landcare Research, Hamilton.