StandardStandard

Partial least squares for discriminating variance components in global navigation satellite systems accuracy obtained under scots pine canopies. / Valbuena, Ruben; Mauro, Francisco; Rodríguez-Solano, Roberto et al.
Yn: Forest Science, Cyfrol 58, Rhif 2, 2012, t. 139-153.

Allbwn ymchwil: Cyfraniad at gyfnodolynErthygladolygiad gan gymheiriaid

HarvardHarvard

APA

CBE

MLA

VancouverVancouver

Valbuena R, Mauro F, Rodríguez-Solano R, Manzanera JA. Partial least squares for discriminating variance components in global navigation satellite systems accuracy obtained under scots pine canopies. Forest Science. 2012;58(2):139-153. doi: 10.5849/forsci.10-025

Author

Valbuena, Ruben ; Mauro, Francisco ; Rodríguez-Solano, Roberto et al. / Partial least squares for discriminating variance components in global navigation satellite systems accuracy obtained under scots pine canopies. Yn: Forest Science. 2012 ; Cyfrol 58, Rhif 2. tt. 139-153.

RIS

TY - JOUR

T1 - Partial least squares for discriminating variance components in global navigation satellite systems accuracy obtained under scots pine canopies

AU - Valbuena, Ruben

AU - Mauro, Francisco

AU - Rodríguez-Solano, Roberto

AU - Manzanera, Jose Antonio

PY - 2012

Y1 - 2012

N2 - This article applies inductive reasoning for explaining how the presence of a forest cover affects the performance of diverse global navigation satellite system (GNSS) receivers. We computed GNSS accuracy and precision obtained in a Scots pine forest situated in the Guadarrama mountain range, Spain. The quality of the GNSS occupations obtained was related to forest parameters and indices describing stand density at the receiver's position. We also characterized the terrain and the canopy gap surrounding the receiver to search for more sources of variability. We computed multiple regression models by means of both ordinary and partial least squares. Results with both techniques showed that most variables were clearly determining the quality of GNSS positioning, although we had to discard the practicality of using terrain slope, stem density, or characterizing only the trees that surround the receiver. Moreover, partial least-squares analysis was successfully used to discriminate two different components that were causing opposite effects on the vertical accuracy. We regarded the second component as being caused by the separate effect of tree needles, because the higher the tree canopy was, the lower the error. Therefore, tree height may describe opposed effects: while wood stock increases, the crown bulk distances itself from the GNSS receiver. We therefore suggest that it is mandatory to model the interactive effects of both the number of trees and their size and height. Hence, we propose relative spacing index, wood volume, and leaf area index as the variables with the best potential for predicting GNSS accuracy.

AB - This article applies inductive reasoning for explaining how the presence of a forest cover affects the performance of diverse global navigation satellite system (GNSS) receivers. We computed GNSS accuracy and precision obtained in a Scots pine forest situated in the Guadarrama mountain range, Spain. The quality of the GNSS occupations obtained was related to forest parameters and indices describing stand density at the receiver's position. We also characterized the terrain and the canopy gap surrounding the receiver to search for more sources of variability. We computed multiple regression models by means of both ordinary and partial least squares. Results with both techniques showed that most variables were clearly determining the quality of GNSS positioning, although we had to discard the practicality of using terrain slope, stem density, or characterizing only the trees that surround the receiver. Moreover, partial least-squares analysis was successfully used to discriminate two different components that were causing opposite effects on the vertical accuracy. We regarded the second component as being caused by the separate effect of tree needles, because the higher the tree canopy was, the lower the error. Therefore, tree height may describe opposed effects: while wood stock increases, the crown bulk distances itself from the GNSS receiver. We therefore suggest that it is mandatory to model the interactive effects of both the number of trees and their size and height. Hence, we propose relative spacing index, wood volume, and leaf area index as the variables with the best potential for predicting GNSS accuracy.

U2 - 10.5849/forsci.10-025

DO - 10.5849/forsci.10-025

M3 - Erthygl

VL - 58

SP - 139

EP - 153

JO - Forest Science

JF - Forest Science

SN - 0015-749X

IS - 2

ER -