Version 4.1 of the Land Cover Database is now available from our download portal at https://lris.scinfo.org.nz/layer/423-lcdb-v41-land-cover-database-version-41-mainland-new-zealand/  for the mainland and near shore islands, and for the Chatham Islands at https://lris.scinfo.org.nz/layer/427-lcdb-v41-land-cover-database-version-41-chatham-islands/.

In addition to the update and inclusion of all time-steps for the Chatham Islands, LCDB v4.1 is an improvement to LCDB v4.0 as a result of three processes:

• Significant contradictions in woody land covers between LCDB v4.0 and 2012 Kyoto Land Use Mapping were investigated and, where necessary, corrections made - these mainly represented indigenous/exotic and forest/scrub confusions.
• Unmapped woody vegetation patches in grassland polygons were detected by spectral methods verified with radar (ALOS PALSAR) analysis, and incorporated in a semi-automated process.
• Error-correction arising from user feedback and limited 'green field mapping' improvements were made during and following the foregoing processes.

Of course, these changes are cumulative from those made in version 4.0 released in June 2014 that, added the 2012 time-step and incorporated a variety of other improvements.

In version 4.1, over 12,500 new polygons (>1 ha) of woody vegetation in grassland were identified using a radar based semi-automated process. These new polygons cover 21,000 hectares. Along with the other improvement processes mentioned above, over 27,000 polygons are either new, have changed class or have been significantly re-mapped (> 20% of their area changed) to improve boundaries. These new or improved polygons represent nearly 646,000 hectares. A direct intersection with v4.0 produces over 56,000 polygons covering 340,000 hectares where the class is now different at one or more time-steps compared to the previous version.

Error Feedback

As always we encourage users to feedback any errors they are aware of. The online data quality feedback (DQF) facility incorporated into the download site is provided for this purpose . It enables users to report specific polygons as requiring attention by submitting a tracked report, or ticket, giving details of the problem and attaching associated evidence such as pictures, screen shots etc. While we encourage the use of the DQF facility we are happy to receive feedback in any form convenient to the user.

Usage Feedback

As any user can download the LCDB we will not always be aware of applications making use of it. We would really appreciate any feedback you are willing to give us on this, both to help us report to the Ministry of Business Innovation and Employment (our funders), but also to help chart the future for the LCDB. Even a couple of lines in an email alerting us to an application would be useful – we can always follow up if we need more detail. Alternatively, feel free to call me or Peter Newsome on 03 321 9763 or 06 353 4975 respectively, to let us know what you do with LCDB.

Associated data

Three sets of summary pivot tables have been generated from LCDB v4.1 to show change between the time steps, i.e., 1996-2001, 2001-2008, and 2008-2012. These tables contain analysis for both the national coverage and by region, including a table for the Chatham Islands. They are best accessed via the links on the LCDB v4.1 download pages shown above.

Future versions

Version 4.0 of the Land Cover Database is now available from our download portal at https://lris.scinfo.org.nz/layer/412-lcdb-v40-land-cover-database-version-40/.   LCDB v4.0 incorporates a new 2012 time-step primarily based on SPOT satellite imagery acquired between November 2011 and February 2013. In addition, this version corrects errors noticed for earlier time periods and refines unsatisfactory line-work representations. Specific mapping activities included: Incorporation of improved wetland mapping from six regional councils Mapping of detected change between 2008/09 and 2012/13 including that associated with harvesting/replanting of production forests Correction of error (in classification or delineation) noticed in the vicinity of edits undertaken above and those notified by users Incorporation of new urban development not detected by the change analysis above but recorded in the LINZ core record system (cadastral) database Of course, these changes are cumulative from those made in version 3.3 released last October that, in addition to fixing errors we were aware of,  incorporated a review of South Island tussock and low producing grasslands and updated croplands mapping to account for interim work from the LUM maps. In version 4.0, over 63,000 polygons have either changed class or have been significantly (> 20% of their area changed) re-mapped to improve boundaries. A direct intersection with v3.3 produces over 260,000 polygons of change. The difference between these numbers  indicates that a lot of minor boundary improvements have been made.   Checking Unlike the mapping done for LCDB v3.0, this time we have not formally passed draft maps back to Regional Councils and other interest groups for checking. Our experience with LCDB v3.0 was that the feedback and number of errors identified was extremely variable between regions. It was obviously difficult to get the appropriate staff in those organisations really focused on this task within the tight timeframes. On our side, creating gaps for this checking made it difficult to efficiently use the mapping staff that we had available.  For v4.0 we felt it was better to press on, using the time to make a better job of the mapping and to incorporate much more formal internal checking of results in the workflow, and then to encourage external feedback on a more relaxed timeframe.  Error Feedback Instead of passing draft maps to key interest groups, an online data quality feedback (DQF) facility has been incorporated into the download site. This enables users to report specific polygons as requiring attention by submitting a tracked report, or ticket, giving details of the problem and attaching associated evidence such as pictures, screen shots etc. We would really encourage all users and particularly key interest groups to make use of this facility. The intention is that we will make a further release (v4.x) of the LCDB within the next year to both incorporate any errors identified and to apply changes identified by new methodologies that were not ready when the version 4.0 workflows were developed. Usage Feedback As any user can download the LCDB we will not always be aware of applications making use of it. We would really appreciate any feedback you are willing to give us on this, both to help us report to the Ministry of Business Innovation and Employment (our funders), but also to help chart the future for the LCDB. Even a couple of lines in an email alerting us to an application would be useful – we can always follow up if we need more detail. Alternatively, feel free to call me or Peter Newsome on 03 321 9763 or 06 353 4975 respectively, to let us know what you do with LCDB. Update 1-July-2014 Three sets of summary pivot tables have been generated from LCDB v4.0  to show change between the time steps, i.e., 1996-2001, 2001-2008, and 2008-2012. These tables contain analysis for both the national coverage and by region. They are best accessed via the links on the LCDB v4.0 download page https://lris.scinfo.org.nz/layer/412-lcdb-v40-land-cover-database-version-40/. There is also a change layer to track both non-temporal and temporal change between LCDB v3.3 and LCDB v4.0 available at https://lris.scinfo.org.nz/layer/413-lcdb-v40-change/.  As all four time steps are present in v4.0, this change layer, and the earlier change layers for v3.3 vs v3.0 and v3.0 vs LCDB-2, are only relevant to users who need to reconcile analysis done on an earlier version with results from a newer version. You should only need the most recent LCDB version for new studies of real temporal change.

LCDB v3.3 has been released and is now available on the LRIS download portal at the address below.

http://lris.scinfo.org.nz/#/layer/401-lcdb-v-33-land-cover-database-version-33/. 

This version contains the same time steps as v3.0, i.e. 1997/97, 2001/02 and 2008/09. It contains a variety of corrections and improvements that were incorporated before embarking on mapping from the 2012 data to produce LCDB v4. Improvements in this version include:

• Recovery of hand-drawn line work where corners were unnecessarily rounded as part of the smoothing of stair-casing artifacts in an earlier processing step (see slide 14 in Digital Earth presentation).
• A review of improvements to low producing grasslands in the South Island
• Updates to croplands from Kyoto Land Use mapping
• Corrections to the dating of land cover change mapped in LCDB v3.0
• Edits arising from feedback

In this version we have included the "EditAuthority” and "EditDate" attributes to indicate the source of the change for both non-temporal and temporal changes mapped. A unique polygon identifier "LCDB_UID" has also been added to aid in the reporting of errors and tracking changes. 

We had hoped to release an online error reporting facility along with v3.3 but as it is still undergoing testing. We decided to not further delay the release of v3.3, and so we will release the error reporting facility separately next month. 

Update 14 November:

The summary pivot tables for change between the 2001 and 2008 time steps based on LCDB v3.3 are now available at http://lris.scinfo.org.nz/file/390-lcdb-v33-cover-class-change-summary/. These now contain tables for national coverage and by region.

There is also a change layer to track both non-temporal and temporal change between LCDB v3.0 and LCDB v3.3 available at http://lris.scinfo.org.nz/#/layer/404-lcdb-v33-change/. As all three time steps are present in v3.3, this change layer, and the earlier change layer for v3.0 vs LCDB-2, are only relevant to users who need to reconcile analysis done on an earlier version with results from a newer version. You should only need the most recent LCDB version for studies of real temporal change. 

LCDB v3.0 has been released and is now available on the LRIS download portal at the address below.

http://lris.scinfo.org.nz/layer/304-lcdb-v30-land-cover-database-version-3/#@nc=&z=6&c=-41.05388%2C172.80014&e=&f=&l=304&mt=OpenStreetMap

I would like to thank Peter, James and all the other Landcare staff involved for the effort they have put in to get this revision out on a very tight timetable. I would also like to thank the many staff in the Regional Councils, Territorial Authorities and DOC who have worked over the draft maps to identify many improvements. Thanks also go to members of our Governance and Technical groups for the valuable advice and assistance they provide.

David Pairman

 LCDB v3.0 has been compiled over the past year from about 160 satellite images of New Zealand that were captured in the summer of 2008/9. With 431,666 polygons, the database classifies all land cover under 33 different classes at three nominal summer periods (1996/97, 2000/01 and 2008/09). 

It is a little difficult to dissect the change statistics into contributing factors as the LCDB v3.0 was produced by a mix of many processes. Some were automated such as the pre-processing to smooth stepping artefacts, and post-processing to align with the Topo50 coastline, while other involved operators interpreting imagery. On top of this some processes were done on regional data sets before final compilation as a national data set for the coastline adjustments. Even after that there was manual editing to fix a few glitches. During the manual mapping stage some areas needed rubber-sheeting where the LCDB2 was found to be misregistered. 

Around 64,000 polygons were manually modified in this revision with about 36% of these being real change between 2000/01 and 2008/09 and the remainder being corrections on previous mapping. About 3000 of the manually modified polygons were as a result of checking or other information by regional councils, territorial authorities and DOC. A further 215,000+ polygons or polygon parts were modified by semi-automated adjustments (excluding smoothing artefacts) to make the dataset more compatible with others. This includes some rubber sheeting and the Topo50 coastline adjustments. However, if you simply take a union of the final product with the LCDB2 (pre- step artefact smoothing), then there are over 13 million change polygons!

Download this change summary spread sheet to get a matrix of LCDB v3.0 class changes between 2001 and 2008 for the whole country in hectares. This will include any areas mapped outside the Topo50 coastline.