Remote-sensing support for the Estonian National Forest Inventory, facilitating the construction of maps for forest height, standing-wood volume, and tree species composition

Mait Lang; Allan Sims; Kalev Pärna; Raul Kangro; Märt Möls; Marta Mõistus; Andres Kiviste; Mati Tee; Toivo Vajakas; Mattias Rennel

doi:10.2478/fsmu-2020-0016

.blurhash-client-img { display: none !important; }

Remote-sensing support for the Estonian National Forest Inventory, facilitating the construction of maps for forest height, standing-wood volume, and tree species composition

Forestry Studies

Volume 73 (2020): Issue 1 (December 2020)

By: Mait Lang, Allan Sims, Kalev Pärna, Raul Kangro, Märt Möls, Marta Mõistus, Andres Kiviste, Mati Tee, Toivo Vajakas and Mattias Rennel

Open Access

|Mar 2021

Figures & Tables

Cluster network of the Estonian NFI (2014–2018).
*Joonis 1. Statistilise metsainventuuri proovitükkide klastrite paiknemine (2014–2018).*

Number of measured permanent (P), temporary (T), and new permanent (NP) sample plots per year.
*Joonis 2. Aastas mõõdetav alaliste (P), ajutiste (T) ja uute alaliste (NP) proovitükkide arv.*

Estonian NFI cluster design and sample plot arrangement.
*Joonis 3. Eesti statistilise metsainventuuri proovitükkide paiknemine klastris.*

Location of target areas for fitting forest height and standing-wood volume models, corresponding to the Estonian Land Board ALS measurement schema. The areas are partially overlapping.
*Joonis 4. Maa-ameti aerofotomõõdistuse objektid, mille piires lähendatakse puistu kõrguse ja tüvemahu prognoosmudelid. Alad on mõningase ülekattega.*

Predicted forest height on the overlap area of ALS data: (a) blocks SW and SE, (b) blocks NW and NE. The blocks correspond to lidar data from summer (SU) and springtime (SP).
*Joonis 5. Lidarmõõtmise alade ülekatetel asuvatele puistutele prognoositud kõrgus: (a) plokid SW ja SE, (b) plokid NW ja NE. Plokid vastavad suvistele (SU) ja kevadistele (SP) lidarandmetele.*

Table 1

Parameters for the forest-height H (dm) prediction model (1). ALS data are from spring (SP) and summer (SU), with RSE the model residual standard error, R² the coefficient of determination, and DF the number of degrees of freedom. Insignificant values (p > 0.05) are in italics.

Tabel 1. Metsa kõrguse H (dm) prognoosmudeli (1) parameetrid. Eristatud on suvised (SU) ja kevadised (SP) laserskaneerimise lennud. RSE on mudeli jääkhälve, R² on determinatsioonikordaja ja DF on vabadusastmete arv. Mitteolulised väärtused (p > 0,05) on kursiivis.

ALS flight ALS lend	Block Ala	Model parameters^* / Mudeli parameetrid^*

		a	b	RSE (dm)	R² %	DF
SP 2017	SW	9.05	11.86	24	89.5	292
SU 2017	SE	6.09	11.58	21	94.7	281
SP 2018	NE	13.80	11.47	17	94.8	313
SU 2018	NW	3.94	12.07	21	91.8	312

* Height percentiles are in metres. / Punktipilve kõrgusprotsentiilid on meetrites.

Table 2

Parameters for the standing-wood volume M (m³ ha⁻¹) prediction model (2) for the upper layer. ALS data are from springtime (SP) and summer (SU). RSE is the model residual standard error, R² the coefficient of determination, and DF the number of degrees of freedom. Insignificant values (p > 0.05) are in italics.

Tabel 2. Esimese rinde tüvemahu M (m³ ha⁻¹) prognoosmudeli (2) parameetrid. Eristatud on suvised (SU) ja kevadised (SP) laserskaneerimise lennud. RSE on mudeli jääkhälve, R² on determinatsioonikordaja ja DF on vabadusastmete arv. Mitteolulised väärtused (p > 0.05) on kursiivis.

ALS flight ALS lend	Model parameters^* / Mudeli parameetrid^*

	a	b	c	d	RSE (m³ ha⁻¹)	R²	DF
SP 2017	1.438	1.264	0.200	0.416	78.6	84.2	287
SU 2017	0.083	1.452	0.121	0.861	97.0	85.2	276
SP 2018	1.277	1.233	0.308	0.454	65.7	91.7	308
SU 2018	0.298	1.492	0.003	0.598	76.3	85.1	304

* Height percentiles are in metres, and canopy cover values within the range of 0–100. / Punktipilve kõrgusprotsentiilid on meetrites ja katvuse väärtus on vahemikus 0–100.

Measured and predicted wood volume for NFI sample plots. Dominant species codes: HB = European aspen; KS = silver birch; KU = Norway spruce; LM = black alder; LV = grey alder; MA = Scots pine; XK = other.
*Joonis 6. Mõõdetud ja prognoositud tüvemaht SMI proovitükkidel. Enamuspuuliikide koodid on metsakorralduse juhendi järgi.*

Table 3

Standing-wood volume M (m³ ha⁻¹) prediction errors under model (2), by dominant tree species, with ALS data distinguished according to spring (SP) and summer (SU).

Tabel 3. Esimese rinde tüvemahu M (m³ ha⁻¹) prognoosmudeli (2) vead peapuuliikide järgi. Eristatud on suvised (SU) ja kevadised (SP) laserskaneerimise lennud.

ALS data ALS andmed	Target area Piirkond	Variable^* Tunnus^*	Dominant species^*/ Enamuspuuliik^*

			HB	KS	KU	LM	LV	MA	XK
SP 2017	SW	M̅	190	161	179	263	152	250	233
SP 2017	SW	RSE	77	67	70	116	96	83	81
SP 2017	SW	MRE	−37	11	18	−45	25	−1	−59
SP 2017	SW	N	22	93	43	13	7	106	7
SU 2017	SE	M̅	265	190	287	331	199	302	87
SU 2017	SE	RSE	102	92	106	90	76	98	73
SU 2017	SE	MRE	−3	37	−13	−51	22	−26	57
SU 2017	SE	N	18	83	57	8	18	92	4
SP 2018	NE	M̅	157	197	235	188	145	237	166
SP 2018	NE	RSE	95	60	73	78	53	56	30
SP 2018	NE	MRE	−37	−3	6	−26	−9	10	−26
SP 2018	NE	N	19	59	66	10	31	120	3
SU 2018	NW	M̅	239	162	248	314	158	211	235
SU 2018	NW	RSE	104	59	91	125	59	72	144
SU 2018	NW	MRE	54	31	−29	−17	12	−24	144
SU 2018	NW	N	9	99	54	16	36	93	1

* M̅ = mean of measured values (m³ ha⁻¹); N = number of observations; RSE = residual standard error (m³ ha⁻¹); MRE = mean residual error (m³ ha⁻¹).

** Species codes: HB = European aspen; KS = silver birch; KU = Norway spruce; LM = black alder; LV = grey alder; MA = Scots pine; XK = other. / ^*M̅: vaatluste aritmeetiline keskmine (m³ ha⁻¹), N: vaatluste arv, RSE: keskmine ruutviga (m³ ha⁻¹), MRE: keskmine viga (m³ ha⁻¹).^** Puuliikide koodid on metsakorralduse juhendi järgi.

Table 4

The average difference between predicted forest height Ĥ (m) and standing-wood volume $\hat{M} (m^{3} {ha}^{- 1})$ in the case of both evergreen (EGR) and deciduous (DEC) stands, upon comparing ALS springtime (SP) against summer (SU) data.

Tabel 4. Prognoositud kõrguse Ĥ (m) ja esimese rinde tüvemahu $\hat{M} (m^{3} {ha}^{- 1})$ keskmine erinevus okaspuupuistutes (EGR) ja lehtpuupuistutes (DEC) kasutades suvised (SU) ja kevadisi (SP) laser-skaneerimise andmeid.

Block SP Ala SP	Block SU Ala SU	$\bar{({\hat{H}}_{SP} - {\hat{H}}_{SU})}, m$		$\bar{({\hat{M}}_{SP} - {\hat{M}}_{SU})}, m^{3} {ha}^{- 1}$

		EGR	DEC	EGR	DEC
SW	SE	0.53	−0.90	34.5	−46.5
NE	NW	0.13	−0.71	40.4	−38.0

Predicted standing-wood volume on the overlap area of ALS data: (a) blocks SW and SE, (b) blocks NW and NE. The blocks correspond to lidar data from summer (SU) and springtime (SP).
*Joonis 7. Prognoositud puistu tüvemaht lidarmõõtmise alade ülekatetel: (a) plokid SW ja SE, (b) plokid NW ja NE.*

Predicted proportion (%) of evergreen conifers for 5,011 NFI sample plots with stands older than 24 years.
*Joonis 8. Prognoositud okaspuude osakaal 5011 SMI proovitükil, kus kasvab 24-aastane või vanem puistu.*

Table 5

Predicted dominant species and field observations on 6,239 NFI sample plots.

Tabel 5. Prognoositud enamuspuuliik võrrelduna 6239 SMI proovitüki andmetega.

NFI / SMI	Predicted dominant species^* / Prognoositud enamuspuuliik^*

	KS	KU	MA	LV	LM	HB	XK
KS	1,110	186	114	153	137	106	16
KU	201	694	162	47	12	37	6
MA	173	211	1,532	4	8	15	2
LV	93	25	3	334	14	23	7
LM	67	10	7	25	123	20	4
HB	81	34	5	65	23	108	13
XK	47	13	15	72	30	35	17

* Species codes: HB = European aspen; KS = silver birch; KU = Norway spruce; LM = black alder; LV = grey alder; MA = Scots pine; XK = other. / Puuliikide koodid on metsakorralduse juhendi järgi.

References

Authors

Metrics

Articles in this issue

DOI: https://doi.org/10.2478/fsmu-2020-0016 | Journal eISSN: 1736-8723 | Journal ISSN: 1406-9954

Journal RSS Feed

Language: English, Estonian

Page range: 77 - 97

Submitted on: Jul 19, 2020

Accepted on: Nov 23, 2020

Published on: Mar 11, 2021

Published by: Estonian University of Life Sciences

In partnership with: Paradigm Publishing Services

Keywords:

National Forest Inventory,

lidar,

multispectral satellite images,

sample plots

Related subjects:

Life sciences,

Plant science

© 2021 Mait Lang, Allan Sims, Kalev Pärna, Raul Kangro, Märt Möls, Marta Mõistus, Andres Kiviste, Mati Tee, Toivo Vajakas, Mattias Rennel, published by Estonian University of Life Sciences
This work is licensed under the Creative Commons Attribution 4.0 License.

Volume 73 (2020): Issue 1 (December 2020)

Remote-sensing support for the Estonian National Forest Inventory, facilitating the construction of maps for forest height, standing-wood volume, and tree species composition

Figures & Tables

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Table 1

Table 2

Figure 6

Table 3

Table 4

Figure 7

Figure 8

Table 5

Paradigm

My account