Research Projects and Consultancy
Our research projects and consultancy services cover a wide range of areas, for customers including commercial and research organisations, government bodies and Māori iwi. Below are some examples to give an indication of the type of projects we do. For more information, see the Landcare Research e-Science page on the main website.
Remote sensing research
We continuously improve our remote sensing capabilities and develop new technologies; for example work has been undertaken in the past year to improve the processing of satellite imagery. The model for topographic correction of satellite imagery with either low sun angle or high slope angle has been made more sensitive using high-precision measurement of directional reflectance of vegetation in the laboratory. New algorithms to standardise high-resolution Quickbird imagery have been used to generate a standardised reflectance regional mosaic for a trial area covering the Marlborough Sounds. This will allow the imagery to be used for the monitoring of sustainable land use.
Advanced text matching: Global Working Checklist of Compositae (Daisy family)
The Informatics team at Landcare Research was funded by a GBIF Seed Funding grant to provide the technical and biodiversity expertise necessary to develop an approach for producing this global checklist of plants in the daisy family. The Compositae is the largest group of vascular plants worldwide and as such attracts much research interest amongst the numerous research and herbaria groups involved in the study and curation of daisy specimens. We developed smart taxonomic tools for merging the multiple sets of ‘names’ and ‘taxonomic concepts’ that species from this family can be known by into a single consensus list. Our method relied on implementation of an international taxonomic concepts schema (TCS) to compile data into standard formats and algorithms with tuneable optimisation steps for parsing data, and selecting most likely data matches across multiple input sources. From there we constructed a tool that allowed a skilled operator to assess unresolved matches and rerun the consensus process as data improvements were made.
Impervious surface mapping service
To enable quantification of urban growth and change for the Christchurch City Council and Auckland Regional Council, we have been mapping the greater urban areas of Auckland and Christchurch using a method called impervious surface mapping.
We use multispectral satellite imagery to estimate the percentage of impervious surface cover within urban areas to high levels of accuracy. Our process uses all available ground-truth information and a semi-automated modelling approach. Based upon differences in ground reflectances in the near infrared band, the method can be used on satellite imagery with spatial resolutions from 30m. The technique requires some ground truth from aerial photographs to provide calibration samples for the logistic regression model applied to estimate impervious fractions.
Spatially specific environmental assessments
Spatially specific environmental assessments recommend mitigation measures that can be built into development projects at the design stage, to enable adverse effects to be avoided or minimised. Building on a history of international consulting, and research collaboration with the University of the South Pacific, we have just completed an environmental assessment of the Fiji–Tonga Cable System for the World Bank and Asian Development Bank. This project intends to lay a submarine fibre-optic cable from the western arm of the Southern Cross network, across 800 km of ecologically sensitive, and seismically, volcanically and tectonically active seabed, to Tonga. Our environmental assessment integrates legal, environmental, social and management knowledge within a spatial framework to deliver policies, plans and rules that will guide the detailed project design and implementation.
High resolution DEMs
The efforts of Informatics team members to produce high-resolution digital elevation models (DEMs) to support S-map landform analysis from Japanese ALOS satellite along-track stereo imagery has been extended in four respects. First, a two-stage method for removing column striping artefacts in the imagery now improves the visual quality and the fine-scale accuracy of the stereo height estimates. Second, Land Information New Zealand geodetic marks are introduced after DEM generation to remove a systematic cross-track error in height. Third, a method has been introduced to suppress the high-frequency noise generated by the image compression process in the ALOS-PRISM subsystem; this change has a small but noticeable effect on the fine-scale accuracy of the stereo height estimates. The cumulative effect of these changes is a reduction in the residual standard deviation of the DEM heights to as low as 4.5 m. Finally, we have gathered ground-control information for an additional five scenes, and introduced a radar interferometry DEM to provide coverage of a gap in ALOS-PRISM coverage of the Wanganui/Manawatu area due to cloud. A sample ALOS-PRISM-derived DEM is now available on the new LRIS Portal.
Policy options for managing contamination of water due to nutrient loss from agricultural land uses is a critical policy issue for Environment Canterbury. To develop such policies, they have embarked on an ambitious stakeholder process in the Hurunui catchment, bringing together the community, industry representatives, planners, and scientists with their knowledge and models. Key Informatics personnel have had a lead role in the modelling part of this project, using both AquiferSim and CLUES (a NIWA model).
The AquiferSim tool was used to illustrate the possible significance of further land intensification on water quality in the Canterbury Water Management Strategy document. AquiferSim has also been tested in another case study area (Hurunui) – a considerably more challenging boundary shape and with only limited information on aquifer properties. AquiferSim is now ready for Environment Canterbury to use in their upcoming Land Use and Water Quality project.
Our work on the mapping of irrigated land based on satellite imagery has been further refined by adding constraints from DEM data. There has been a great deal of interest in maps of irrigated land, for a variety of applications. We are currently working with MAF to develop a strategy to map irrigated areas nationally.
Processing of soil bulk density and soil carbon for MfE
The assessment of soil carbon quantities has become an important part of Government goals to meet carbon emission targets. Systematic errors in the way that soil fine earth bulk density had been measured have recently been brought to light. Informatics researchers have developed a new regression model to estimate bulk density for records where laboratory or field-information is unavailable. This information can then be used to update the Ministry for the Environment’s (MfE) Historic Soils database. The updated version of the Historic Soils database will then be processed to produce a new statistical model to estimate soil carbon; this will be used by MfE to refine estimates for national soil carbon.
The bulk density regression model uses a relatively modern regression technique drawn from the data mining field, called stochastic gradient boosting. This method is particularly suitable for datasets (such as the National Soils Database) where the variables are sparsely available. This new method gives greater accuracy than previous methods for bulk density estimation, and operates on a mix of field and laboratory measurements.
Fact Sheet Manager: A content management system (CMS) for biological data
The Landcare Research Biodiversity Informatics group have recently completed phase one of a Fact Sheet Manager development project that allows information drawn dynamically from a variety of digital sources to be coordinated, combined, and presented to users as web-accessible content. The system allows content authors to control what is displayed and how it is combined with other resources, e.g. images. It is specifically tuned to biological data, especially where information about organisms from specific groups is being described, but its design is also general enough to be used in the future for other information series, e.g. soils and their physical properties. The presentation component will integrate with other more general CMSs allowing organisations to use the tools that are appropriate for the coordination and arrangement of scientific content of this nature.
The system has services-orientated architecture and has a number of components including: a source data template and accompanying XML transfer schema, a data loading service, orchestration routines for linking and integrating content from different data sources, a published data warehouse (which stores the integrated content), a data integration service that combines data and externally held resources, (e.g. metadata and images), and a web delivery mark-up tool allowing providers to fine-tune the style of presentation and control publication of sensitive data. The infrastructure also consumes a number of supporting services also developed by the Biodiversity Informatics group; for example, authorisation is handled through web service calls to our security module. Images and metadata are similarly provided from independent modules via web services. The tool has already been used in collaboration with Plant & Food Research to generate the Interesting Insects and Other Invertebrates fact sheet series.
Tuakiri – the NZ Access Federation
Landcare Research, together with subcontractors The University of Auckland and the University of Canterbury, is responsible for the technological side of setting up the New Zealand national identity and access management federation, funded by MoRST. The New Zealand Federation (Tuakiri) will ensure that staff, students and researchers can access and personalise the online resources they are entitled to, by giving appropriate information to the service provider. Examples of such resources are virtual environments to share data and exchange ideas, institutional repositories, grid resources for data storage or computation, and online journals.
We are creating a toolkit of best practices and standard installation templates to enable New Zealand institutions to join the federation easily. The project will establish a number of best endeavour Identity Providers among New Zealand institutions and provide technical assistance to deliver federated services.
The federation is based on Shibboleth and will use uApprove so users can review and give consent to any information released about them to service providers. The New Zealand Federation will be closely aligned with its counterpart in Australia, the Australian Access Federation (AAF).
New Zealand Organisms Register
The New Zealand Organisms Register (NZOR) is a national information infrastructure project, funded by TFBIS, to efficiently mobilise, integrate, and share authoritative taxonomic information critical to maintaining New Zealand’s conservation and biosecurity decision support systems and processes, and it is being developed by the Landcare Research Biodiversity Informatics group. Once complete NZOR will be an online (web and web services) digital catalogue of taxonomic data associated with around 100 000 organisms relevant to New Zealand. The register will be a distributed harvesting and caching infrastructure with web-services (SOAP and REST) architecture for dynamically and automatically merging data from authoritative data providers within New Zealand and globally. Web-services will allow end-users to dynamically integrate NZOR content into local databases as well as deliver updated content into the real-time cache. Web-based access to names and other content will also be available.