5. Georeferencing Forest Inventory into the FRA Geographic Information System

Not all of the inventories had a map with which it was possible to geographically locate the area inventoried. The reasons for this were different. The map was annexed in a separate document but this document was not found. The map should have been annexed with the report but the envelope-page was empty. The map was just a scratch map without coordinates. There was map of the global area inventoried but not of the different units or strata. Or only the coordinates of the fours corners were available.

In order to georeference the volume, depending on the information available, different procedures were applied.

In the case of a missing map, the description of the area was very useful and from this it was occasionally possible to understand and locate the area inventoried in a generic country map. Infact very often inventories’ limits were considered natural limits such as rivers, lakes, or administrative boundaries. Due to this, it was very important to invest time and energy to find country maps with the same information reported in the inventory description. In order to achieve this, the library (A.G.L.) was very useful. Normally the map used was at 1:500000/1000000 scale.

When it was possible to locate the area inventoried in a map with coordinates, the procedure was the following:

-The borders of the area inventoried (the smallest unit for which the volume was reported) were drawn on an acetate with a 0.2 mm pen. In order to save time during the next steps, if the units were more than one block or additional inventories covered in a neighboring zone, they were included on the same lucid paper.

-At least four tic marks (normally six or nine) were drawn on the acetate and their id and coordinates reported in a tixxxxxx.txt file (xxxxxx represents the id code).

-The polygons drawn on the acetate were scanned at 200 dpi, and saved as a compressed .tif file (imxxxxxx.tif).

-The files were imported into Arc\Info GIS and vectorized. For corresponding tic marks, tic points were added.

-The cover was then edited to clean all the superfluous arcs. All the needed corrections were done.

-To georeference the map, a cover of just tics was created and projected. Very few inventory maps included information on the projection system characteristics. When it was not possible to figure out the projection system, the projection characteristics of the topographical maps used in the respective countries was adopted. The cover -with the polygons was "Transformed" on the tics cover earlier projected.
-A new Item in the coxxxxxx.pat file was added to register the Id_gis code.

-The polygon labels were added and coded with the Id_gis code.

When it was not possible to locate the area inventoried on a map, but the coordinates of the four corners were provided, the procedure consisted in building a cover of polygons directly from a .txt file where the coordinates of the fours corners were reported. The ID of the polygon reported in the .txt file corresponds to the Id_gis. In this way it was possible to avoid the labelling phase.

When two or more inventories covered the same area or part of it, to keep the polygons separate, and consequently the volume, the Region function of Arc\Info was used.

One of the problems with the Forest Inventory Data of India, coming from the "Goverment of India, Forest Survey of India, 1993. The State of Forest Report 1993" was the fact that we didn’t know with sufficient precision the position of the inventoried area and its shape. So the question was: At what feature and from what geographical point of view, could we link the Forest Inventory Data? Different possibilities were evaluated; e.g., to associate the inventory data to a feature point, a circle with the size proportional to the volume data, or a rectangular feature. In the end, considering the fact that forest inventories regard generally only forested areas we decided to proceed as follows:

The vegetation map¹ and the administrative unit map² were combined in an overlay and linked to a database. From this new map we knew where the forest was, what class of forest a certain polygon was, its size and the administrative name of the area. All the polygons of the vegetation map belonging to a single volume data were selected to develop the forest inventory map plot. As a result we had our data geographically georeferenced with a probable shape. Data inventory was georeferenced and the volume attributed to the equivalent vegetation map units (polygons delimiting the classes). As result, a forest inventory plot map was generated. A new item was added at the polygon cover and polygons have been labeled with the appropriate Id_gis code.

The vegetation map used is going to be replaced by the National Forest Vegetation Map at scale 1: 3.500.000 based on visual interpretation of Landsat imagery-1991-1993.

Data was available in a .dbf file for all the 2735 plot-clusters in a 20x20 km grid (see Sampling Design description). No useful map was provided. Due to the necessity to georeference them, an Arc/Info routine was developed to locate each of them.

You can see the data for The National Forest Inventory of Indonesia in the Annex 1.2 .

¹The map was prepared by the International Institute for Vegetation Mapping (ICIV) Toulouse, France, for FAO. It was drawn on FAO base maps of 1:5,000,000 scale in a Miller Oblated Stereographic Projection. The manuscript was prepared from information current in approximately 1980. The mapping projects and reports were finalized in 1987. The map was digitized between 1988-90.The map was digitized by the Canada Lands Data System (CLDS) Division of Environment, Canada, for FAO.

²For more information see the work “Estimating Actual Forest Volume using Forest Inventory and Climatological Data”.