- Tytuł:
-
Określanie wybranych cech drzewostanów sosnowych z wykorzystaniem chmur punktów pozyskanych w procesie automatycznego dopasowania cyfrowych zdjęć lotniczych
Estimating selected characteristics of Scots pine stands based on point clouds derived from automatic matching of airborne images - Autorzy:
- Hawryło, P.
- Powiązania:
- https://bibliotekanauki.pl/articles/986689.pdf
- Data publikacji:
- 2017
- Wydawca:
- Polskie Towarzystwo Leśne
- Tematy:
-
lesnictwo
fotogrametria
zdjecia lotnicze
automatyczne dopasowanie zdjec lotniczych
chmura punktow
drzewostany sosnowe
wysokosc Loreya
pole powierzchni przekroju piersnicowego
zasobnosc drzewostanu
dokladnosc szacowania
modele predykcyjne
area−based approach
forest inventory
lorey's height
basal area
growing stock volume - Opis:
- Recent research show that image−derived point clouds (IPC) can be successfully utilized in the context of estimation of selected forest parameters. However there are still some topics for further investigation in the context of using archival data. Particularly, in case of forest stands in Poland the number of such studies is limited. This study investigates the usefulness of airborne IPC for estimation of Lorey's height (H), basal area (G) and growing stock volume (V) of Scots pine stands. Additionally, the accuracy of predictive models is analyzed for images acquired in the same year as the field reference data were collected (2015) and for archival images acquired in 2012. The study was performed in Janów Lubelski Forest District (south−eastern Poland) on approximately 7,800 ha. The measurements from 94 circular plots were used as the reference data. The Area Based Approach and multiple linear regression were used to create predictive models. For each analyzed dependent variable variety of IPC characteristics were evaluated to select the most accurate predictive models. The best model for each variable was selected based on Akaike Information Criteria and RMSE. The same predictor variables were used for both years of aerial images aquisition. For models comparison the one−sided bootstrap test was used. The RMSPE for H, G and V in case of images acquired in 2015 amounted to 5.3, 16.3 and 17.3%, respectively. Analogically, the relationship between predicted and observed values (R2) amounted to 0.96, 0.47 and 0.81. The finally selected models had three, one and two predictors for H, G and V, respectively. The difference between accuracy of models based on aerial images from 2012 and 2015 was not statistically significant for all analyzed variables. The obtained results show high potential of using airborne image−derived point clouds for forest inventory activities in Scots pine dominated stands.
- Źródło:
-
Sylwan; 2017, 161, 09; 707-714
0039-7660 - Pojawia się w:
- Sylwan
- Dostawca treści:
- Biblioteka Nauki
Artykuł