2017/05/09 by Chad Babcock, Babcock, Chad, Andrew O. Finley +21
Environmental Science · #Applications (stat.AP) #FOS: Computer and information sciences #Forest ecology and management #Remote Sensing and LiDAR Applications #Soil Geostatistics and Mapping
paper · pdf · doi:10.48550/arxiv.1705.03534
openalex publication_date 2017/05/09 · openalex created_date 2022/10/03 · openalex updated_date 2026/07/28
The goal of this research was to develop and examine the performance of a\ngeostatistical coregionalization modeling approach for combining field\ninventory measurements, strip samples of airborne lidar and Landsat-based\nremote sensing data products to predict aboveground biomass (AGB) in interior\nAlaska's Tanana Valley. The proposed modeling strategy facilitates pixel-level\nmapping of AGB density predictions across the entire spatial domain.\nAdditionally, the coregionalization framework allows for statistically sound\nestimation of total AGB for arbitrary areal units within the study area---a key\nadvance to support diverse management objectives in interior Alaska. This\nresearch focuses on appropriate characterization of prediction uncertainty in\nthe form of posterior predictive coverage intervals and standard deviations.\nUsing the framework detailed here, it is possible to quantify estimation\nuncertainty for any spatial extent, ranging from pixel-level predictions of AGB\ndensity to estimates of AGB stocks for the full domain. The lidar-informed\ncoregionalization models consistently outperformed their counterpart lidar-free\nmodels in terms of point-level predictive performance and total AGB precision.\nAdditionally, the inclusion of Landsat-derived forest cover as a covariate\nfurther improved estimation precision in regions with lower lidar sampling\nintensity. Our findings also demonstrate that model-based approaches that do\nnot explicitly account for residual spatial dependence can grossly\nunderestimate uncertainty, resulting in falsely precise estimates of AGB. On\nthe other hand, in a geostatistical setting, residual spatial structure can be\nmodeled within a Bayesian hierarchical framework to obtain statistically\ndefensible assessments of uncertainty for AGB estimates.\n