Multi-Source National Forest Inventory Methods and Applications

About this book

Building on more than a decade of innovative research into multi-source forest inventory (MS-NFI) this book presents full details of the development, outputs and applications of the improved k-NN method. The method, which was pioneered in Finland in 1990, is rapidly becoming a world standard in forest inventory, having been adopted as standard in Finland and Sweden, and recently introduced in Austria and across the US.

The book describes in detail the full MS-NFI process, and the input data used a" including field data, satellite images, and digital map data, as well as coarse-scale variation of forest variables. It also presents comprehensive information on the types of outputs which can be derived, including maps and statistics, describing, for example, stock volumes and development, dominant tree species, age-class distribution, and large and small-scale variation.

The book will provide an invaluable resource for those involved in forest inventory, including government departments and bodies involved in forest policy, management and monitoring, forest managers, and researchers and graduate students interested in forest inventory, modelling and analysis. It will find an additional market among those interested in Earth observation, ecology and broader areas of environmental and natural resource management. Erkki Tomppo was the winner of the 1997 Marcus Wallenberg Prize for his work on the k-NN method.

Contents

List of abbreviations.- Preface.-1. Introduction.-2. Materials. 2.1 Field data. 2.1.1 Sampling designs. 2.1.2 Measurements and assessments on field sample plots and plot stands. 2.2 Satellite images. 2.2.1 The applied satellite images. 2.2.2 Landsat 5 TM. 2.2.3 Landsat 7 ETM. 2.2.4 IRS-1C and IRS-1D. 2.3 Digital map data. 2.3.1 The use of the map data. 2.3.2 The main sources of map data. 2.3.3 Peatland. 2.3.4 Arable land. 2.3.5 Urban areas, houses and other built-up areas. 2.3.6 Roads. 2.3.7 Water. 2.3.8 Accuracy of the combined land use map data. 2.3.9 Digital boundaries of the computation units. 2.4 Digital elevation model. 2.5 Large area forest resource data.-3. Methods. 3.1 Image rectification and pre-processing of data. 3.1.1 Satellite image rectification. 3.1.2 Radiometric correction by means of digital elevation model. 3.1.3 Preparation of input data sets. 3.2 Estimation. 3.2.1 Field data based estimation and reliability analysis. Estimation and error estimation based on the field plot data. 3.2.2 The basic k-NN estimation method. 3.2.3 The improved k-NN (ik-NN) method, use of coarse scale forest variable estimates and genetic algorithm in the distance metric. 3.2.3.1 Simplified sketch of the genetic algorithm. 3.2.3.2 The application of the algorithm. 3.2.4 Selecting estimation parameters and their values for k-NN. 3.2.5 Area and volume estimates -- stratification, correction for map errors. 3.2.5.1 Calibrated MS-NFI estimators. 3.2.5.2 Stratified MS-NFI. 3.2.5.3 Calibration of the MS-NFI municipality estimates to the official land areas. 3.2.6 Assessing the errors -- current and potential methods. 3.2.6.1 The current methods in assessing the reliability of the results. 3.2.6.2 Model-based error estimation.-4. Results. 4.1 Forest resources by municipalities. 4.2 Comparison of the results by regions and to MS-NFI8 results. 4.2.1 Variables in the comparison. 4.2.2 Mean volume of growing stock. 4.2.3 Dominant tree species. 4.2.4 Age class distribution on forest land. 4.2.5 Distribution of development classes. 4.2.6 Available energy wood. 4.3 Accuracy of small-area estimates from MS-NFI8 and MS-NFI9. 4.3.1 Empirical errors of MS-NFI9 small-area estimates based on independent inventory data. 4.3.2 Assessing the systematic errors of the MS-NFI8 and MS-NFI9 municipality estimates. 4.4 Digital thematic output maps.-5. Discussion.-Acknowledgements.- References.-Appendix 1. Forest resource tables 1--8.-Appendix 2. Examples of forest resource maps.-Index.-

Customer Reviews