Graduation date: 1992

Various methods have been used to estimate the amount of above ground forest biomass across landscapes and to create biomass maps for specific stands or pixels across ownership or project areas. Without an accurate estimation method, land managers might end up with incorrect biomass estimate maps, which could lead them to make poorer decisions in their future management plans. Previous research has shown that nearest-neighbor imputation methods can accurately estimate forest volume across a landscape by relating variables of interest to ground data, satellite imagery, and light detection and ranging (LiDAR) data. Alternatively, parametric models, such as linear and non-linear regression and geographic weighted regression (GWR), have been used to estimate net primary production and tree diameter. The goal of this study was to compare various imputation methods to predict forest biomass, at a project planning scale (<20,000 acres) on the Malheur National Forest, located in eastern Oregon, USA. In this study I compared the predictive performance of, 1) linear regression, GWR, gradient nearest neighbor (GNN), most similar neighbor (MSN), random forest imputation, and k-nearest neighbor (k-nn) to estimate biomass (tons/acre) and basal area (sq. feet per acre) across 19,000 acres on the Malheur National Forest and 2) MSN and k-nn when imputing forest biomass at spatial scales ranging from 5,000 to 50,000 acres. To test the imputation methods a combination of ground inventory plots, LiDAR data, satellite imagery, and climate data were analyzed, and their root mean square error (RMSE) and bias were calculated. Results indicate that for biomass prediction, the k-nn (k=5) had the lowest RMSE and least amount of bias. The second most accurate method consisted of the k-nn (k=3), followed by the GWR model, and the random forest imputation. The GNN method was the least accurate. For basal area prediction, the GWR model had the lowest RMSE and least amount of bias. The second most accurate method was k-nn (k=5), followed by k-nn (k=3), and the random forest method. The GNN method, again, was the least accurate. The accuracy of MSN, the current imputation method used by the Malheur Nation Forest, and k-nn (k=5), the most accurate imputation method from the second chapter, were then compared over 6 spatial scales: 5,000, 10,000, 20,000, 30,000, 40,000, and 50,000 acres. The root mean square difference (RMSD) and bias were calculated for each of the spatial scale samples to determine which was more accurate. MSN was found to be more accurate at the 5,000, 10,000, 20,000, 30,000, and 40,000 acre scales. K-nn (k=5) was determined to be more accurate at the 50,000 acre scale. / Graduation date: 2013

Quantifying and modeling processes involved in the global carbon cycle is important to evaluate the temporal and spatial variability of these processes and understand the effect of this variability on future response to changing climate and land use patterns. Biomass accumulation and Net Primary Productivity (NPP) are large components of ecosystem carbon exchange with the atmosphere and thus are the focus of many modeling efforts. When scaling estimates of NPP temporally from days to years and spatially from square meters to landscapes and regions the spatial coherence of these processes through time must be taken into account. Spatial coherence is the degree to which pairs of sites across space are synchronous (i.e., correlated) through time with respect to a given process or variable. In this thesis I determined the spatial coherence of a major component of NPP, tree bole productivity (NPP[subscript B]), and examine how it influences scaling and our ability to predict NPP and forecast change of this flux. In Chapter 2 I developed and tested a method modeling radial tree increment growth from sub-sampled trees and estimating annual site-level biomass accumulation that allows quantification of the uncertainty in these estimates. Results demonstrated that a simple model using the mean and standard deviation of growth increments underestimated bole biomass increment in all three age classes examined by 1% at the largest sample sizes and up to 15% at the smallest sample sizes. The long term average NPP[subscript B] and inter-annual variability were also underestimated by as much as 10% and 22%, respectively. Stratification of trees by size in sampling and modeling methods increased accuracy and precision of estimates markedly. The precision of both models was sufficient to detect patterns of inter-annual variability. To estimate bole biomass accumulation with acceptable levels of accuracy and precision our results suggest sampling at least 64 trees per site, although one site required a sample size of more than 100 trees. In Chapter 3 I compared year to year variability of NPP for tree boles (NPP[subscript B]) for two adjacent small watersheds (second-growth and old-growth) in the western Cascades of Oregon using the methods developed in Chapter 2. Spatial coherence of NPP[subscript B] within and between watersheds was assessed using multivariate analysis techniques. NPP[subscript B] was found to be less coherent between watersheds than within watersheds, indicating decreased spatial coherence with differences in age class and increased spatial scale. However, a larger degree of spatial coherence existed within the old-growth watershed compared to the second-growth watershed, which may be a result of the smaller degree of variation in environmental characteristics in the former watershed. Within a watershed, potential annual direct incident radiation and heat load were more strongly associated with the variation of NPP[subscript B] than climate. Climatic factors correlated with the temporal variation of annual NPP[subscript B] varied between the two watersheds. Results suggest that inter-annual variability and spatial coherence of forest productivity is a result of both internal (e.g., environment and stand dynamics) and external (climate) factors. An unexpected conclusion was that spatial coherence was not consistent and changed through time. Therefore, the coherence of sites over time is not a simple relationship. Instead the patterns of spatial coherence exhibit complex behaviors that have implications for scaling estimates of productivity. This result also indicates that a correlation coefficient alone may not capture the complexity of change through time across space. In Chapter 4 I estimated year to year variation of NPP[subscript B] for eleven sites of varying age, elevation, moisture, and species composition in the Western Cascades of Oregon. Spatial coherence of tree growth within sites and NPP[subscript B] between sites was assessed using Pearson's product-moment correlation coefficient (r). Results suggest that spatial coherence is highly variable between sites (r=-O.18 to 0.92). The second-growth sites exhibited the greatest temporal variability of annual NPP[subscript B] due to the large accumulation of biomass during stand initiation, but old-growth sites exhibited the greatest variation of coherence of NPP[subscript B] between sites. In some years all sites behaved similarly, but for other years some sites were synchronous while others were not. As growth of individual trees and NPP[subscript B] at the site scale increased, inter-annual variability of those variables increased. Climate in part affected annual NPP[subscript B], but intrinsic factors and spatial proximity also affected the coherence between sites in this landscape. / Graduation date: 2006