Transferability of Lidar-derived Basal Area and Stem Density Models within a Northern Idaho Ecoregion

Abstract A patchwork of disjunct lidar collections is rapidly developing across the USA, often acquired with different acquisition goals and parameters and without field data for forest inventory. Airborne lidar and coincident field data have been used to estimate forest attributes across individual lidar extents, where forest measurements are collected using project-specific inventory designs. This research explores predicting forest attributes at locations not represented in the training data by combining lidar and field measurements from ecologically similar forests. Using field measurements from six lidar units, random forests regression models were created by systematically withholding forest inventory data from one lidar unit and using the forest inventory data from the other five units to predict basal area and stem density at the withheld unit. Results indicate that BA models produce more accurate predictions than stem density models when transferred to a lidar unit that did not contain field data. Relative root mean square errors calculated from the withheld field plots ranged between 32.3%–50.1% for BA and 40.7%–67.3% for stem density models. It is concluded that forest managers may use predictive models constructed from ecologically similar forests to obtain a preliminary estimate of resources, until local field measurement can be obtained. RÉSUMÉ Un ensemble disparate de collections de données lidar se développe rapidement aux États-Unis, souvent acquises avec des objectifs et des paramètres différents et sans données de terrains relatives aux inventaires forestiers. Des données de terrain concomitantes sont utilisées pour estimer des attributs forestiers au sein d’étendues de lidar aéroporté individuelles, pour lesquelles les mesures d’intérêt correspondent à des modèles d’inventaires déterminés. En combinant des mesures de terrain et de lidar provenant de forêts écologiquement similaires, cette étude examine la prédiction d’attributs forestiers à des endroits où des données d’entrainement sont absentes. En utilisant des mesures de terrain, des modèles de régression produits par des forêts d’arbres décisionnels ont été créés pour six unités de lidar. Un procédé systématique a été suivi par lequel les données de chaque unité sont retenues et la surface terrière et densité des tiges de cette unité sont ensuite estimés avec les données des cinq unités restantes. Les résultats indiquent que les estimations de surface terrière sont plus robustes que pour la densité de tiges lorsque les modèles sont transférés à une unité de lidar pour laquelle les données de terrain sont absents. Les erreurs quadratiques moyennes relatives calculées pour les échantillons retenus sont entre 32.3 % et 50.1 % pour les modèles de surface terrière et entre 40.7 % et 67.3 % pour la densité des tiges. Nous concluons que, jusqu'à ce que des mesures de terrain locales puissent être acquises, il est possible d’obtenir des estimations préliminaires de ressources forestières en utilisant des modèles prédictifs construits à partir d'échantillons de forêts écologiquement similaires.

[1]  Damian H. Evans Airborne laser scanning as a method for exploring long-term socio-ecological dynamics in Cambodia , 2016 .

[2]  G. Moisen,et al.  Random forests and stochastic gradient boosting for predicting tree canopy cover: Comparing tuning processes and model performance , 2016 .

[3]  Michael J. Falkowski,et al.  Temporal transferability of LiDAR-based imputation of forest inventory attributes , 2015 .

[4]  hya sree.M,et al.  Lidar Remote Sensing , 2015 .

[5]  Aaron Weiskittel,et al.  Evaluation of alternative methods for using LiDAR to predict aboveground biomass in mixed species and structurally complex forests in northeastern North America , 2015, Math. Comput. For. Nat. Resour. Sci..

[6]  Florian Hartig,et al.  Importance of sample size, data type and prediction method for remote sensing-based estimations of aboveground forest biomass , 2014 .

[7]  Joanne C. White,et al.  Validating estimates of merchantable volume from airborne laser scanning (ALS) data using weight scale data , 2014 .

[8]  Harold S. J. Zald,et al.  Influence of lidar, Landsat imagery, disturbance history, plot location accuracy, and plot size on accuracy of imputation maps of forest composition and structure , 2014 .

[9]  Joanne C. White,et al.  A best practices guide for generating forest inventory attributes from airborne laser scanning data using an area-based approach , 2013 .

[10]  D. Pitt,et al.  Parametric vs. nonparametric LiDAR models for operational forest inventory in boreal Ontario , 2013 .

[11]  Barbara Koch,et al.  Evaluation of most similar neighbour and random forest methods for imputing forest inventory variables using data from target and auxiliary stands , 2012 .

[12]  Jungho Im,et al.  Forest biomass estimation from airborne LiDAR data using machine learning approaches , 2012 .

[13]  J. Eitel,et al.  Quantifying aboveground forest carbon pools and fluxes from repeat LiDAR surveys , 2012 .

[14]  Warren B. Cohen,et al.  Modeling Percent Tree Canopy Cover: A Pilot Study , 2012 .

[15]  Thomas Hilker,et al.  Stability of Sample-Based Scanning-LiDAR-Derived Vegetation Metrics for Forest Monitoring , 2011, IEEE Transactions on Geoscience and Remote Sensing.

[16]  A. Hudak,et al.  Mapping snags and understory shrubs for a LiDAR-based assessment of wildlife habitat suitability , 2009 .

[17]  P. Gessler,et al.  Characterizing forest succession with lidar data: An evaluation for the Inland Northwest, USA , 2009 .

[18]  E. Næsset Effects of different sensors, flying altitudes, and pulse repetition frequencies on forest canopy metrics and biophysical stand properties derived from small-footprint airborne laser data , 2009 .

[19]  Paul E. Gessler,et al.  The influence of conifer forest canopy cover on the accuracy of two individual tree measurement algorithms using lidar data , 2008 .

[20]  D. R. Cutler,et al.  Utah State University From the SelectedWorks of , 2017 .

[21]  E. Næsset Airborne laser scanning as a method in operational forest inventory: Status of accuracy assessments accomplished in Scandinavia , 2007 .

[22]  A. R. Stage,et al.  Interactions of Elevation, Aspect, and Slope in Models of Forest Species Composition and Productivity , 2007, Forest Science.

[23]  Andrew Thomas Hudak,et al.  A Multiscale Curvature Algorithm for Classifying Discrete Return LiDAR in Forested Environments , 2007, IEEE Transactions on Geoscience and Remote Sensing.

[24]  C. Hopkinson The influence of flying altitude, beam divergence, and pulse repetition frequency on laser pulse return intensity and canopy frequency distribution , 2007 .

[25]  Nicholas C. Coops,et al.  Assessment of forest structure with airborne LiDAR and the effects of platform altitude , 2006 .

[26]  A. Prasad,et al.  Newer Classification and Regression Tree Techniques: Bagging and Random Forests for Ecological Prediction , 2006, Ecosystems.

[27]  Rick L. Lawrence,et al.  Mapping invasive plants using hyperspectral imagery and Breiman Cutler classifications (RandomForest) , 2006 .

[28]  P. Gessler,et al.  Regression modeling and mapping of coniferous forest basal area and tree density from discrete-return lidar and multispectral satellite data , 2006 .

[29]  J. Marshall,et al.  A regression-based equivalence test for model validation: shifting the burden of proof. , 2005, Tree physiology.

[30]  Kazuhiro Aruga,et al.  Application of an airborne laser scanner to forest road design with accurate earthwork volumes , 2005, Journal of Forest Research.

[31]  Andrew P. Robinson,et al.  Model validation using equivalence tests , 2004 .

[32]  E. Næsset Practical large-scale forest stand inventory using a small-footprint airborne scanning laser , 2004 .

[33]  Nicole A. Lazar,et al.  Testing Statistical Hypotheses of Equivalence , 2003, Technometrics.

[34]  W. Cohen,et al.  Lidar remote sensing of above‐ground biomass in three biomes , 2002 .

[35]  E. Næsset Predicting forest stand characteristics with airborne scanning laser using a practical two-stage procedure and field data , 2002 .

[36]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[37]  Shinichi Morishita,et al.  On Classification and Regression , 1998, Discovery Science.

[38]  David W. Roberts,et al.  Forest habitat types of northern Idaho: a second approximation , 1991 .

[39]  R. Bailey Delineation of ecosystem regions , 1983 .

[40]  R Core Team,et al.  R: A language and environment for statistical computing. , 2014 .

[41]  E. Næsset,et al.  Forestry Applications of Airborne Laser Scanning , 2014, Managing Forest Ecosystems.

[42]  J. Evans,et al.  Quantifying Bufo boreas connectivity in Yellowstone National Park with landscape genetics. , 2010, Ecology.

[43]  H. Temesgen,et al.  Imputing Mean Annual Change to Estimate Current Forest Attributes , 2009 .

[44]  P. McDaniel,et al.  A Geographically Weighted Regression Analysis of Douglas-Fir Site Index in North Central Idaho , 2008 .

[45]  Andy Liaw,et al.  Classification and Regression by randomForest , 2007 .