No abstract

Large-area assessment of aboveground tree biomass (AGB) to inform regional or national forest monitoring programs can be efficiently carried out by combining remotely sensed data and field sample measurements through a generic statistical model, in contrast to site-specific models. We integrated forest inventory plot data with spatial predictors from Landsat time-series imagery and LiDAR strip samples at four sites across the eastern USA-Minnesota (MN), Maine (ME), Pennsylvania-New Jersey (PANJ) and South Carolina (SC)-in statistical modeling frameworks to analyze the performance of generic (all sites combined) versus site-specific models. The major objective was to evaluate the prediction accuracy of generic and site-specific models when applied to particular sites. Pixel-level polynomial model fitting was applied to the time-series of near-anniversary date Landsat variables to obtain projected metrics in the target year 2014 for which LiDAR strip samples were available. Two forms of models based on ordinary least-squares multiple linear regressions (MLR) and the random forest (RF) machine learning approach were developed for each site and for the pooled (i.e., generic) reference data frame. The models were evaluated using national forest inventory (NFI) data for the USA. We observed stronger fit statistics with the MLR than with RF for both the site-specific and the generic models. The proportions of variances explained (adjusted R 2 ) with the site-specific models were 0.86, 0.78, 0.82 and 0.92 for ME, MN, PANJ and SC, respectively while the generic model had adjusted R 2 = 0.85. A test of statistical equivalence of observed and predicted AGB for the NFI locations did not reveal equivalence with any of the models, possibly due to the different resolutions of the observed and predicted data. In contrast, predictions by the generic and site-specific models were equivalent. We conclude that a generic model provides accuracies comparable to the site-specific models for large-area AGB assessment across our study sites in the eastern USA.

The ability to harmonize data sources with varying temporal, spatial, and ecosystem measurements (e.g. forest structure to soil organic carbon) for creation of terrestrial carbon baselines is paramount to refining the monitoring of terrestrial carbon stocks and stock changes. In this study, we developed and examined the short-(5 years) and long-term (30 years) performance of matrix models for incorporating light detection and ranging (LiDAR) strip samples and time-series Landsat surface reflectance high-level data products, with field inventory measurements to predict aboveground biomass (AGB) dynamics for study sites across the eastern USA-Minnesota (MN), Maine (ME), Pennsylvania-New Jersey (PANJ) and South Carolina (SC). The rows and columns of the matrix were stand density (i.e. number of trees per unit area) sorted by inventory plot and by species group and diameter class. Through model comparisons in the short-term, we found that average stand basal area (B) predicted by three matrix models all fell within the 95% confidence interval of observed values. The three matrix models were based on (i) only field inventory variables (inventory), (ii) LiDAR and Landsat-derived metrics combined with field inventory variables (LiDAR+Landsat+inventory), and (iii) only Landsat-derived metrics combined with field inventory variables (Landsat+inventory), respectively. In the long term, predicted AGB using LiDAR+Landsat+inventory and Land-sat+inventory variables had similar AGB patterns (differences within 7.2 Mg ha −1 ) to those predicted by matrix models with only inventory variables from 2015-2045. When considering uncertainty derived from fuzzy sets all three matrix models had similar AGBs (differences within 7.6 Mg ha −1 ) by the year 2045. Therefore, the use of matrix models enabled various combinations of LiDAR, Landsat, and field data, especially Landsat data, to estimate large-scale AGB dynamics (i.e. central component of carbon stock monitoring) without loss of accuracy from only using variables from forest inventories. These findings suggest that the use of Landsat data alone incorporating elevation (E), plot slope (S) and aspect (A), and site productivity (C) could produce suitable estimation of AGB dynamics (ranging from 67.1-105.5 Mg ha −1 in 2045) to actual AGB dynamics using matrix models. Such a framework may afford refined monitoring and estimation of terrestrial carbon stocks and stock changes from spatially explicit to spatially explicit and spatially continuous estimates and also provide temporal flexibility and continuity with the Landsat time series.