News Contact Information

Dpt of Forest Sciences
P.O. Box 27
(Latokartanonkaari 7)
FI-00014 University of Helsinki

Student Services:
Registration of credits:
Communications and Community Relations:
PhD Studies:

Doctoral dissertation: ”Use of remotely sensed auxiliary data for improving sample-based forest inventories”

New possibilities for the acquisition of forest information are currently emerging. Different remote sensing techniques like airborne and space-borne lasers, digital air photography, and radar satellites can be used as complements to traditional field based inventories. In this thesis statistical methods for co-utilizing field and remotely sensed data for large-area forest surveys are developed and evaluated. It is shown that these methods may improve the cost-efficiency of forest inventories and it is suggested that they may be important tools in meeting society's global demands for accurate forest information.

The conversion from a fossil-based to a bio-based economy is a major challenge for society and requires substantial changes in our techniques for making efficient use of renewable resources. The most important source of renewable raw materials in Finland as well as in many other boreal and tropical countries is forests. New chemicals, fuels, and biomaterials derived from forest biomass are emerging and a large number of other ecosystem services are derived from forests as well. Thus, forests are of vital importance for the transition towards a bio-based economy. 

The management of forests require accurate information. For large-area forest inventories, such as national forest inventories, remotely sensed (RS) data can be combined with samples of field data in order to provide cost-efficient and up-to-date descriptions of forest resources. Well-known statis­tical approaches to utilize RS data are stratification and post-stratification. More advanced, and only recently introduced, methods include design-based (DB) model-assisted (MA) and model-based (MB) estimation approaches. The latter approaches rely on advanced modeling of the relationships between field and RS data.

DB inference has been applied in forest inventories since the 1920s in the Nordic countries. However, it is not the only inferential mode that can be applied in survey sampling.  Only recently the MB and DB MA approaches have been introduced in forest inventories. MB inference is an alternative to MA estimation in case RS auxiliary data are available. Described briefly, MB inference relies more heavily on the correctness of the model(s) applied in the estimators. While the dependence on the model is a drawback, this mode of inference also has advantages over MA approaches; e.g., in some cases smaller sample sizes might be applied for reaching a certain level of accuracy.
In perspective of the recently introduced MA and MB estimation approaches, the present  thesis focused on analysis and further development of these techniques for large-area forest inventories. Empirical data for the studies were acquired from a boreal forest area in the Kuortane region of western Finland. The data comprised a combination of auxiliary information derived from airborne LiDAR and Landsat data, and field sample plot data collected using a modification of the 10th Finnish National Forest Inventory. The studied forest attribute was growing stock volume.

In Paper I, RS data were applied at the design stage, using a newly developed design which spreads the sample efficiently in the space of auxiliary data. The analysis was carried out through Monte Carlo sampling simulation using a simulated population developed by way of a copula technique utilizing empirical data from Kuortane. The results of the study showed that the new design resulted in a higher precision when compared to a traditional design where the samples were spread only in the space of geographical data.

In Paper II, RS auxiliary data were applied in connection with MA estimation. The auxiliary data were used mainly in the estimation stage, but also in the design stage through probability-proportional-to-size sampling utilizing Landsat data. The results showed that LiDAR auxiliary data considerably improved the precision compared to estimation based only on field samples. Additionally, in spite of their low correlation with growing stock volume, adding Landsat data as auxiliary data further improved the precision of the estimators.

In Paper III, the focus was set on MB inference and the influence of the use of different models on the precision of estimators.  For this study, a second simulated population was developed utilizing the empirical data, including only non-zero growing stock volume observations. The results revealed that the choice of model form in MB inference had minor to moderate effects on the precision of the estimators. Furthermore, as expected, it was found that MB prediction and MA estimation performed almost equally well.

In Paper IV, the precision of MB prediction and MA estimation was compared in a case where field and RS data were geographically mismatched. The same simulated population as used in Paper III was employed in this study. The results showed that the precision in most cases decreased considerably, and more so when LiDAR auxiliary data were applied, compared to when Landsat auxiliary data were used. As for the choice of inferential framework, it was revealed that MB inference in this case had some advantages compared to DB inference through MA estimators.

The results of this thesis are important for the development of forest inventories to meet the requirements which stem from an increasing number of international commitments and agreements related to forests.

Svetlana Saarela, Engineer, majoring in Forestry, is one of the pioneers in the field of MB inference for forest surveys. During her studies she conducted the first in-depth analysis of MB inference for large-area forest inventories. Her work is of interest to those who apply mathematical statistics for surveys of natural resources using the MA and MB inferential frameworks. She is defending her doctoral dissertation titled “Use of remotely sensed auxiliary data for improving sample-based forest inventories” (“Användning av fjärranalysdata för att förbättra stickprovsbaserade skogsinventeringar”) at the Department of Forest Sciences, Faculty of Agriculture and Forestry, University of Helsinki, on September 25th, 2015, at 14:00 o’clock.

The public examination will take place at the following address: B-building, Luentosali 5, Latokartanonkaari 7.

Professor Timothy G. Gregoire, School of Forestry and Environmental Studies, Yale University, U.S.A., will serve as the opponent, and Professor Bo Dahlin as the custos.

The dissertation has been published in the series Dissertationes Forestales 201. The dissertation is also available in electronic form through the E-thesis service.

Picture: The hybrid inference for population mean prediction based on ordinary least squares regression with homo- and heteroskedastic residuals; the background photo was taken in the Red Canyon (Utah, U.S.A.) by Svetlana Saarela.