A E Hewitt, E J Willoughby, R H Wilde and R M Andrew

Landcare Research, Dunedin and Palmerston North

The National Soils Database (NSD) is a database of well-characterised soil profiles with data describing sites, soil morphology, chemistry, mineralogy and physics. It represents a substantial part of our collective knowledge of New Zealand's soil resources. Together with associated digitised soil maps and the NZ Land Resource Inventory (NZLRI), the NSD has been designated a nationally significant database by MoRST. It is currently undergoing development and expansion. Our paper reviews its status, development progress, and discusses some of its applications.

The NSD contains digital data from 1449 soil profiles collected between 1964 and 1993; data from the period 1938 to 1964 are held in a card filing system. The digital database is rich with data for 550 soil attributes. Although sites are located throughout NZ, geographic representation is uneven and based on the location of soil surveys. Representation of land use is dominated by pasture (65%), with exotic and native forest sites least well represented. Of the 250 subgroups described in the New Zealand Soil Classification only 190 are represented. The database is available in PARADOX, and an SQL version is currently being tested.

The FRST-funded Environmental Data Programme is contracted to undertake an intensive development of the NSD. Substantial data sets not currently in the NSD will be included and a programme of field sampling and laboratory analysis will fill critical data gaps. Data gaps have been identified by a combination of two methods. In the first method, spatial models have been generated for critical soil attributes at national scales, and a standard error surface derived. For areas where the standard error is large, new sampling sites will be chosen to decrease the error. In the second method, samples will be chosen to fill combinations of soil groups, vegetation classes and land use. Our aim is to have adequate representation of major driving variables that can be used for spatial modelling of critical soil attributes at a range of scales sufficient to meet reasonable levels of uncertainty. A minimum data set has been defined for cost-effective sampling and analysis, and will enable efficient modelling of attribute values. A new quality assurance system is also being designed for the NSD.

The NSD will eventually be integrated with other Landcare Research databases. The ability to link and search across soil databases held by CRIs, universities and other agencies needs to be explored.

The NSD is used in two ways. The first is in non-spatial data analysis of soil attributes, for example, in the determination of normal ranges of soil health indicators. The second is in mapping soil classes or soil attributes. The trend is to model single soil attribute surfaces rather than soil classes. Although the data cannot generate maps with the resolution needed for precision agricultural applications at within-field scales, it can be used to stratify areas of comparable fine-scale variability, and to locate land suitable for specific developments.

Return to Paper List