The effects of photo-interpreted variables in the estimation of stand-level merchantable volumes in the Province of Quebec

Context

The Quebec forest inventory is currently carried out using a stratified sampling design. Aerial photos of a target Forest Management Unit (FMU) are first taken. The areas of such FMU’s vary between 160 km2 and 27,000 km2. Photo-interpreters divide the territory into homogeneous polygons and, for each of those polygons, estimate the value of specific biological and physical variables. Once the photo-interpretation is completed, polygons with similar photo-interpreted characteristics are grouped into strata. There can be more than 1000 strata in a FMU. The stratification is used to direct the field inventory in which 400 m2 plots are established in stratum and in which forest properties such as basal area and merchantable volume per species are measured. The target sample size is of 15 per stratum, but only a fraction of the plots required can be established within the target FMU. Several plots are therefore recruited from other FMU’s or from previous inventories through heuristic rules. Plots are assigned to strata, and strata-level forest properties are imputed from plot averages. Field measurements are taken in the sampled plots such as the basal area and wood volume by species.

As can be seen from the synthetic description of the current forest inventory practice in Québec, the photo-interpreted variables are used only for constructing the strata. Can they be used to estimate the stand-level merchantable total and by species volume?

Objectives

Primary objective

The primary objective is to predict the total stand-level merchantable volume using the information from photo-interpreted, climatic, geographical and ecological classification variables.

Secondary objectives

It is also of interest to learn the following:

How the explicative variables can be used to predict the hardwood and softwood total volume?
How the explicative variables can be used to predict each of the species’ volume?

Research Question

To what extent do photo-interpreted variables explain the stand-level merchantable total volume and the wood volume by species?

Variables

Potential explanatory variables:

Group	Num	Name	Units	Description
	1	ID _ PET _ MES		Temporary sample plot identifier
	2	Tenure	0/1	0=Public land, 1=Private land
Photo-interpreted	3	Pert	0/1	0 = Undisturbed plot, 1 = Disturbed plot
	4	h _ moy	m	Stand height
	5	pct _ couv	%	Stand density expressed as the proportion of the plot area occupied by the projection of tree crown on the ground
	6	pct _ RES	%	Forest type expressed as the proportion of basal area occupied by softwoods
	7	Pct _ EB	%	Proportion of basal area occupied by White Spruce
	8	Pct _ EN	%	Proportion of basal area occupied by Black Spruce
	9	Pct _ EP	%	Proportion of basal area occupied by Red and Black Spruces
	10	Pct _ ML	%	Proportion of basal area occupied by Tamarack
	11	Pct _ PG	%	Proportion of basal area occupied by Jack Pine
	12	Pct _ RX	%	Proportion of basal area occupied by undetermined softwoods
	13	Pct _ SB	%	Proportion of basal area occupied by Balsam Fir
	14	Pct _ SE	%	Proportion of basal area occupied by Balsam Fir and White Spruce
	15	Pct _ TO	%	Proportion of basal area occupied by Eastern White Cedar
	16	Pct _ BJ	%	Proportion of basal area occupied by Yellow Birch
	17	Pct _ BP	%	Proportion of basal area occupied by White Birch
	18	Pct _ EO	%	Proportion of basal area occupied by Red Maple
	19	Pct _ ER	%	Proportion of basal area occupied by Maples
	20	Pct _ ES	%	Proportion of basal area occupied by Sugar Maple
	21	Pct _ FI	%	Proportion of basal area occupied by shade intolerant hardwoods
	22	Pct _ FH	%	Proportion of basal area occupied by hardwoods on humid stations
	23	Pct _ FN	%	Proportion of basal area occupied by non-commercial tree species
	24	Pct _ FO	%	Proportion of basal area occupied by Black Ash
	25	Pct _ FT	%	Proportion of basal area occupied by shade tolerant hardwoods
	26	Pct _ FX	%	Proportion of basal area occupied by undetermined hardwoods
	27	Pct _ PE	%	Proportion of basal area occupied by Poplars
	28	Age _ moy	Year	Mid value of stand age class
	29	Age _ structure	0/1	0=Even-aged forest stand, 1=Uneven-aged forest stand
	30	Dep _ epais	cm	Modal deposit thickness
	31	Drainage		Drainage index
Climatic	32	Deg _ jr	°C	Degree-day temperature
	33	Ptot	mm	Mean annual total precipitation
	34	Aridite	cm	Aridity index
Geographic	35	latitude	° decimal	Latitude
	36	longitude	° decimal	Longitude
	37	ALTITUDE	m	Altitude
	38	PENTE	%	Slope
	39	EXPOSITION	°	Aspect
	40	potentiel		Site potential productivity index
	41	topex		Wind exposure index
Ecological Land Classification Hierarchy	42	sous _ dom		Bioclimatic subdomain
	43	reg _ ecol		Land region
	44	sous _ reg		Land subregion
	45	unit _ pays		Regional landscape
	46	dist _ eco		Land distinct

Dependent variables:

Num	Name	Units	Description
1	ID _ PET _ MES		Temporary sample plot identifier
2	V _ BJ	m3/ha	Yellow Birch volume
3	V _ BP	m3/ha	White Birch volume
4	V _ EB	m3/ha	White Spruce volume
5	V _ SB	m3/ha	Balsam Fir volume
6	V _ TO	m3/ha	Eastern White-Cedar volume
7	V _ PE	m3/ha	Poplars volume
8	V _ EP	m3/ha	Black Spruce and Red Spruce volume
9	V _ FEU	m3/ha	Hardwood volume
10	V _ RES	m3/ha	Softwood volume
11	V _ TOT	m3/ha	Total volume

Data Access

A training data set of 5479 plots is available to construct the different models. This data set includes all the photo-interpreted variables along with some climatic and geophysical variables. Total volume in each plot is also available. Prediction of volumes has to be made for plots included in the test data set (UAF=012-54 with 1214 plots)

References

Bernier PY, Guindon L, Kurz WA, Stinson G. Reconstructing and modelling 71 years of forest growth in a Canadian boreal landscape: testing the CBM-CF3 carbon accounting model. Canadian Journal of Forest Research [In review].
Borders BE, Harrison WM,Clutter ML, Shiver BD, Souter RA. The value of timber inventory information for management planning. Canadian Journal of Forest Research 2008, 38: 2287 – 2294.
Coulombe S, Bernier PY, Raulier F. Uncertainty in detecting climate change impact on the projected yield of black spruce (Picea mariana). Forest Ecology and Management 2009 [In press].
Franco-Lopez H, Ek AR, Bauer ME. Estimation and mapping of forest stand density, volume, and cover type using the k-nearest neighbors method. Remote Sensing of Environment 2001, 77: 251 – 274.
Gjertsen AK, Tomppo E, Erkki, Tomter S. National forest inventory in Norway: using sample plots, digital maps, and satellite images. Proceedings of the International Geoscience and Remote Sensing Symposium (IGARSS) 2 1999, 729 – 731.
Lemay V, Temesgen H. Comparison of nearest neighbour methods for estimating basal area and stems per hectare using auxiliary variables. Forest Science 2005, 51: 109 – 119.
McRoberts RE, Nelson MD, Wendt DG. Stratified estimation of forest area using satellite imagery, inventory data and the k-nearest neighbors technique. Remote Sensing of Environment 2002, 82: 457 – 468.
McRoberts RE, Tomppo E, Finlay AO, Heikkine J. Estimating areal means and variances of forest attributes using the k-nearest neighbors technique and satellite imagery. Remote Sensing of Environment 2007, 111: 466 – 480.
McRoberts RE. A two-step nearest neighbors algorithm using satellite imagery for predicting forest structure within species composition class. Remote Sensing of Environment 2009, 113: 532 – 545.
Miranda M, Ung C-H, Guindon L, Condal A, Bernier P, Beaudoin A, Patry A. A simple Mahalanobis distance in the kNN method using aerial photograph for predicting stand table in a boreal mixed forest. In R Fournier and R McRoberts (editors) Extended forest inventory. Proceedings of the IUFRO- Division 4 Conference, Quebec City, 19 – 22 May 2009, 5 p.
Temesgen H, Lemay VM, Froese KL, Marshall PL. Imputing tree lists from areal attributes for complex stands of south-eastern British Columbia. Forest Ecology and Management 2003, 177: 277 – 285.
Tomppo E. Recent status and further development in the Finnish multi-source inventory. Lectures given at the 1997 Marcus Wallenburg Prize Symposium, Stockholm, Sweden, 14 October 1997, pp. 53 – 68.
Tomppo E, Goulding C, Katila M. Adapting Finnish multi-source forest inventory techniques to the New Zealnd preharvest inventory. Scandinavian Journal of Forest Research 1999, 14: 182 – 192.
Tomppo E, Korhonen KT, Heikkinen J, Yli-Kojola H. Multi-source inventory of the forests of the Hebei Forestry Bureau, Heilongjiang, China. Silva Fennica 2001, 35: 309 – 328.
Tuominen S and Pekkarinen A. Performance of different spectral and textural aerial photograph features in multi-source forest inventory. Remote Sensing of Environment 2004, 94: 256 – 268.