Work begins on SoilML – a soil equivalent of the Geoscience Markup Language (GeoSciML)

A report by Alistair Ritchie.

On August 16 and 17 the National Land Resource Centre hosted a trans-Tasman soil information workshop at the Landcare Research office in Wellington. The workshop followed a one day joint soil and geology information modelling workshop at GNS Science Lower Hutt.

These workshops are typical of a growing community of environmental and information scientists that has been collaborating to develop standardised ways of storing and sharing data. Key to this are well defined information models describing scientific phenomena that can be subsequently adapted to define the structure of the digital datasets. Sharing land and soil information in a standardised way can bring many benefits: it increases the efficiency of data management and exchange by reducing the number of file formats and data structures users must deal with; it allows the integration of data from disparate data sources; and, by focussing on core land and soil concepts rather than one-off user interfaces, it increases the likelihood that the data may be reused for multiple, unforeseen, purposes.

This work is increasingly important as environmental science makes extensive use of computational models that combine data describing a variety of phenomena from a variety of sources. These models may be run on desktop computers or across interconnected computing facilities using and producing very large volumes of data. This increasing scale of operation – often called the ‘data deluge’ – means that data acquisition and exchange is becoming more and more expensive. It is vital that we reduce the time spent exchanging and manipulating data, allowing more time to do the actual analysis.

The resulting datasets, delivered using internet technology, can be used for any number of applications, ranging from environmental research, economic and land use modelling, to the management of farmland.

When building these information models it is helpful to not only address the data from the perspective of an individual discipline, say soil science, but also from the those that describe related systems (geology, hydrology and so on). This improves the ability to integrate and compare data collected by different scientists, but also to reduces the amount of time spent developing the models by sharing models for common concepts.

With this in mind, Landcare Research and CSIRO Land and Water took the opportunity to conduct a joint soil and geology information workshop with members of the International Union of Geological Sciences (IUGS) geology information working group as they met at GNS Science, Lower Hutt. The geologists have spent the last ten years developing the GeoScience Mark-up Language (GeoSciML) – a web-oriented data model for the exchange of geology data and the driver for projects such as the OneGeology digital map of the world geology (http://onegeology.org/), and earth science data infrastructures in Australia, the US, Canada and Europe.

The one day workshop (attended by Landcare Research staff) covered how the two groups used one another’s data and identified a great deal of common ground, considering shared technology and integration options in their information models (not just covering the physical phenomena, but also describing observation and sampling activities, and the use of technical vocabularies). It is clear that the emerging soil information community can learn a lot from the geology communities successes (and mistakes!) while the soil community can teach the geologists a lot about describing quantitative data and the related uncertainty.

Immediately after this workshop, the Landcare Research and CSIRO attendees attended a National Land Resource Centre sponsored soil information modelling meeting. Here work began on an Oceania SoilML, the soil equivalent of GeoSciML, and a related information model for the GlobalSoilMap.net (http://globalsoilmap.net/) project. To ensure that we didn’t reinvent the wheel, the meeting considered models that have been developed in Australia and Europe, particularly a draft International Standards Organisation (ISO) Soil Data Quality standard. While the group found the ISO standard to have merit, especially when blended with the Australian model, a number of issues were identified and these have been submitted to ISO via the Australian standards body. Meanwhile, the GlobalSoilMap.net model – developed by Landcare Research – was agreed to be a viable candidate as an international standard. This model, and a web interface demonstrating the exploration of the data, will be demonstrated to an Oceania GSM meeting in October.

This work will also have direct application to research Landcare Research is presently engaged in. The Oceania SoilML information model will be used to inform the development of a new model for the ideal content of the New Zealand National Soils Database – potentially saving many months of development time.

Both workshops were a great success. They identified scope for collaboration between the geology and soil communities, allowing both to effectively share technical and intellectual resources. Meanwhile, the development of a SoilML can be considered a more tractable exercise, benefiting already from preceding work, and clear points of contact with GeoSciML. The challenge now is to extend the effort beyond Oceania and formally integrate the information modelling work into activities of the International Union of Soil Sciences (IUSS) Soil Information Working Group.

 

 

Tags: , , , , , , , ,
Posted in News | No Comments »

New Zealand’s Landcover Database v3.0 (LCDB3) released

By David Pairman

New Zealand’s Landcover Database v3.0 (LCDB3) was released at the start of July 2012 and is now available on the LRIS download portal.

LCDB v3.0 has been compiled by staff within the Informatics team 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 version 2 of the Landcover Database (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 polygons changed!

Further information on the LCDB can be found on the LCDB website.

Tags: , , , ,
Posted in News | No Comments »

Landcare wins contract for new editions of the New Zealand Land Cover Database (LCDB)

The Informatics team along with colleagues from the Soils and Landscapes team have successfully secured the LCDB3+ contract.

Land Cover Database (LCDB) is a digital map of the surface of New Zealand created using satellite imagery. It contains detailed information on categories of land cover and boundaries, and is a record of land cover changes over time.  Two completed editions (LCDB1 and LCDB2), which show the state of New Zealand’s land cover in 1996-1997 and 2000-2001 respectively, have become critical to central, regional and local government, industry and research institutions. The information is used for land, water and biodiversity management, pest control and monitoring, wildfire threat and risk analysis, and environmental monitoring and reporting.

The creation of new land cover databases and related research has been identified as an important priority by Ministry for the Environment (MfE), Department of Conservation (DOC) and regional councils.  The Government will give $1 million a year over the next four years to the creation the new land cover databases and will involve the use of satellite technology, geospatial mapping techniques and advanced computing systems.

Landcare Research has been commissioned to deliver two further editions of LCDB and, a parallel research programme will develop improved ways of generating land cover information.  LCDB3, the map of land cover as at 2008, should be released in July 2012. LCDB4, which will use imagery acquired over the summers of 2011/12 and 2012/13, is  expected to be released in July 2014.

Tags: , , , , ,
Posted in News | No Comments »

Informatics assists in the Christchurch earthquake recovery

Members of the Informatics team  have been doing their bit following the Christchurch earthquake. While staff based in Lincoln were coming to grips with the aftermath of the earthquake (all are safe although a few have damaged property), those based in Palmerston North have been taking the lead in the processing and orthorectification of satellite imagery. GeoEye imagery was made available to Landcare through the NZDF Geospatial Intelligence Organisation,  processed, rectified and mosaiced and then uploaded to the BeSTGRID data fabric where it could be downloaded by the emergency services and government agencies. We have also made 15m Digital Elevation Model data of Christchurch and surrounding areas available to the the emergency services and government agencies and also through the LRIS Portal.

Two members of the Informatics team have also been assisting on the Christchurch Recovery Map.

Posted in News | No Comments »