Catalogue description Forestry Commission: Research Division: National Inventory of Woodland and Trees: Datasets

Hardware: Husky Hunter 16/80 hand-held computers were used for data input in the field. After transfer to the Forestry Commission, data for both surveys has been held in variety of hardware platforms, initially a Prime mainframe; followed by a DEC VAX, followed by a VAX Workstation 4000, VLC VMS PCs, and Dell Pentium PCs.

Operating System: Microsoft DOS 3.3 was used for field input. The original Prime system ran the PRIMOS v19.x operating system; VMS 6.1 was originally used with the DEC VAX. Following the transfer of applications to networked PCs, several versions of Microsoft Windows were used.

Application Software: Data for the surveys was downloaded by surveyors as ASCII files and transferred to the Forestry Commission, where it was originally loaded into a Rapport database with a customised interface written in FORTRAN. Later, the software was changed to an Oracle database (version 7.0 for the Grampian pilot, version 8i for the full survey).

Digital map data: in Scotland, the digital maps from the Land Cover of Scotland 1988 survey were produced by the Macaulay Land Use Research Institute, using 'SPANS software with NUMONICS digitizing tablets'. SPANS is the name of a suite of spatial analysis software originally produced by TYDAC Research Inc and owned, since 1994, by PCI Geomatics. Numonics digitizing tablets are input devices. The data was exported to the Forestry Commission in ARC Export format. It was then loaded into LAMPS1, a digital mapping package (produced by Laser-Scan Limited, Cambridge, UK). LAMPS1 has also been used to hold digital map data generated for the NIWT in England and Wales. Since c.1997 Woodland Surveys Branch have analysed NIWT data and produced summary statistics using ESRI's ArcView GIS software.

User Interface: Input and manipulated via a series of data entry screens and menus on the Husky hand-helds.

Logical structure and schema: See the catalogues of individual datasets.

How data was originally captured and validated: Different data gathering and sampling techniques have been applied in each of the three main sections of the NIWT (the production of digital maps, the main woodland survey and the small woodland and trees survey). It should be noted that for the purposes of the NIWT, 'woodland' is defined as 'land under stands of trees with, or the potential to achieve, tree crown cover of more than 20%'. 'Areas of open space integral to the woodland' are included with woodland, as are 'intervening land classes such as roads, rivers or pipelines' if they are less than 50 metres in extent. The digital maps identifying woodland over 2 hectares have been produced from 1:25,000 aerial photographs and using information about more recent private plantings taken from records of the Forestry Commission's woodland grants scheme. In Scotland, the NIWT's digital maps were derived from those produced for the Land Cover of Scotland 1988 survey. This was a complete census of the land cover of Scotland, in which digital maps recording 126 land cover features were produced through the interpretation of specially flown aerial photographs taken between 1987 and 1989.

In the main woodland survey, the digital maps provide the raw material for selecting areas for ground surveying. Sampling is done using systematic random clusters of 1 hectare square plots (100 metres x 100 metres). Woodland identified on the maps as being over 2 hectares in extent is sorted into one of three size bands. A sample square grid, consisting of clusters of one to five randomly selected sample squares, is laid over the digital maps. The number of areas of woodland sampled increases with each size band, while the density of squares in the sample grid (i.e. the intensity of the sampling) decreases. The overall aim has been to achieve a 1% sample of each size category:

• Woodland of 2.0 - <100.0 hectares: every fifth woodland is selected with a cluster grid of sample plots at 5% intensity.
• Woodland of 100.0 - <500.0 hectares: two woods in five are selected with a cluster grid of sample plots at 2.5% intensity.
• Woodland of 500 hectares and over: all woods are selected with a cluster grid of sample plots at 1% intensit.

Each sample square selected for surveying is divided into sections of at least 0.1 hectare on the basis of species, age, height and mixture. Up to five sections may be assessed for each sample square. Sections are in turn divided into up to three elements, to record details of the three primary tree species in a section. Sections and elements are treated as being analogous to 'sub-compartments' and 'components' in the Forestry Commission's Sub-Compartment Database (a database recording forest land uses and growing stock, which was established in 1976 for use in the Forestry Commission's quinquennial forecast of production and valuation).

During the pilot survey in the Grampian Region, woodland owned by the Forestry Commission and managed by the Commission's commercial arm, Forest Enterprise, was sampled separately from private woodland. Data on Forestry Commission woodland was taken from the Forestry Commission's Sub-Compartment Database, and from structure assessments carried out in the field. After the Grampian pilot, it was decided to sample Forestry Commission woodland and private woodland as part of the same sampling exercise. Another change which took place after the Grampian pilot was the decision to sample within the 100 km square 'tiles' of the Ordnance Survey National Grid. During the Grampian pilot it had been assumed that sampling would take place by local authority area (in Scotland, by Region), but this proved to be impractical, as some woodland overlapped the areas of more than one local authority.

Different sampling methods have been followed in the small woodland and trees survey. Data gathering is based on a combination of assessing data from aerial surveys plus ground surveying of sampled plots. 1 kilometre squares based on the Ordnance Survey grid are each divided into 16 squares of 250 metres x 250 metres, two of which are surveyed. The aim is a 1% sample consisting of 4764 250 m x 250 m plots, for the whole of Great Britain.

Data validation checks were performed by the Husky software at the time of data input. Surveyors were required to periodically download data onto 3.5 inch floppy disks, which were sent to the Forestry Commission. Other validation checks (including cross-field checking, hierarchical checking and checks for missing data) were carried out by the Forestry Commission after data transfer.

Validation performed after transfer: Details of the content and transformation validation checks performed by NDAD on the datasets for the NIWT are recorded in the Dataset Catalogues.

The early phases of the National Inventory of Woodland and Trees (NIWT), including the planning of the survey in 1993 and the Grampian pilot in 1993-1994, were conducted by the Surveys Branch of Policy Studies Division within the Policy and Resources Group of the Forestry Commission. Responsibility for field surveys and for the NIWT is thought to have passed to Research, Development and Surveys Group when Policy Studies Division was broken up in 1995. The NIWT later became the responsibility of the Woodland Surveys Branch of Forest Research, which was created as an Executive Agency of the Forestry Commission in 1997. Hardware and programming support for the NIWT database has been provided by the Forestry Commission's Business Services Division and its predecessor, Business Systems Division.