Dees, Matthias 
  
COMBINING REMOTE SENSING DATA SOURCES AND TERRESTRIAL SAMPLE-BASED 
INVENTORY DATA FOR THE USE IN FOREST MANAGEMENT INVENTORIES 
Matthias DEES', Jan DUVENHORST", Claus Peter GROSS", Barbara KOCH" 
" University of Freiburg, Germany 
Department for Remote Sensing and Landscape Information Systems 
ferninfoQ uni-freiburg.de 
" PROCUL Consulting Freiburg, Germany 
  
1 
Working Group VII/4 
KEY WORDS: Forestry, Sample Inventory, Stratification, Data Fusion, Photogrammetrie, Landsat TM. 
ABSTRACT 
This paper presents two elements of a study on forest inventory and mapping in the context of forest management with 
a test site in the state Nordrhein-Westfalen, Germany. The first section concerns the analysis of the sample based forest 
inventory in a systematic grid design. The use of aerial photos represents an inexpensive, exact means of mapping the 
borders of the stand. This is a requirement for the analysis of the sample based forest inventory as a stratified sample. 
This analysis option makes it possible to reduce the sampling error for the central assessment attributes compared to an 
analysis using the simple random sampling approach. Along with the possibility of increasing accuracy, it is also 
possible to reduce the size of the sample by 25% without any loss of accuracy for the central assessment attributes. The 
second section concerns the k-nearest-neighbour method, in which sample data and medium resolution satellite data 
(Landsat TM and IRS1C LISS) are used. This method can provide a representation of the spatial distribution of central 
attributes. So far the mapping of main tree species has been the subject of study. This method does not provide a 
sufficient information basis for the standwise forest management inventory under the forest conditions that apply to the 
area studied. It can, however, provide a good overview of the spatial distribution of the main tree types. 
1 INTRODUCTION 
The study on the system of forest inventory and mapping was commissioned by Civil Forest Service of the state Nord- 
rhein-Westfalen, Germany. The system of forest inventory and mapping is part of the system of forest management. In 
the civil forest enterprises, every ten years the forest management plan is revised or updated. Elements in gathering 
information are 1) remote sensing, 2) a terrestrial sample based inventory (whose introduction is planned), 3) surveys 
for the entire stands in a part of the stands, and 4) information from previous surveys. The general aim of the study is, 
on one hand, to examine the possibilities of introducing remote sensing into the system of forest inventory and mapping 
and, on the other, to examine the possibilities that arise from the introduction of the stand inventory on the basis of 
samples. The introduction of the stand inventory on the basis of samples makes statistically supported information 
possible at the level of the forest enterprise and for larger sections of the forest enterprise. The gathering of information 
on the level of the stand thus is no longer necessary to provide information about the entire forest enterprise that could 
be gained from the summation of the information of all single stands. Nevertheless, the single stands should continue to 
be described individually as they represent the smallest spatial element for forest management. The previous method of 
mapping and describing the stand, which is time-consuming and costly, should be replaced to the extent possible by less 
expensive survey methods. In this context, this study comprises the following elements: 
1. Evaluation of the possibilities of mapping the stands using various data sources and techniques by means of visual 
interpretation (identification of homogenous units, registering of borders, updating of borders). 
2. Evaluation of the possibilities of describing the mapped stands with various data sources and techniques by means of 
visual interpretation. 
3. Evaluation of the possibilities of describing the stands with the k-nearest-neighbour method by linking the sample 
information with the information from the satellite data from the sensors Landsat TM5 and IRSIC LISS. 
4. Identification of further possibilities of high quality stand descriptions with a minimum of terrestrial surveys, thus 
reducing costs. 
5. Investigation of the option of using stratification in the analysis of the sample inventory. 
The present paper presents two areas of investigation from the overall context of the study: on the one hand, the method 
of stratification in the analysis of sample-based inventory; on the other, the possibilities of the k-nearest-neighbour 
method in providing information for important attributes concerning the forest stands and within the stands. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 355