Earth Observation (EO) data can provide added value to nations' assessments of vegetation aboveground biomass density (AGBD) with minimal additional costs. Yet, neither open access to global-scale EO datasets of vegetation heights or biomass, nor the availability of computational power, has proven sufficient for their wide uptake in climate policy-related assessments. Using Mexico as an example, one of the primary obstacles to enhancing their National Forest Inventory (NFI) with such global EO datasets is the lack of statistically defensible methodologies that do so, while addressing the nation's existing reporting needs and gaps. In collaboration with the Comisi & oacute;n Nacional Forestal (CONAFOR), this study develops a geostatistical model that integrates vegetation height and AGBD estimates from NASA's Global Ecosystem Dynamics Investigation (GEDI) and ESA's Climate Change Initiative (CCI) with Mexico's NFI to attain sub-national and geographically-explicit biomass predictions. The posited model includes spatially varying parameters, allowing flexibility to capture non-stationary relations between the EO-based covariates and NFI-estimated AGBD. Inference is conducted with Bayesian methods, allowing the computation of summary statistics, such as the standard deviations for single-location and area-wide predictions of AGBD. This enables the transparent disclosure and traceability of sources of uncertainty throughout the prediction approach. Results indicate strong model performance; the EO-based covariates explain 79% of the variance in NFI-estimated AGBD in a randomly withheld sample of 10% of observations and a heuristic root mean squared error (RMSE) of 21.55 Mg/ha. Approximately 96% of the observations falling within the 95% credible intervals of our predictions, with some systematic under-prediction observed at AGBD ranges of >100 Mg/ha. To ease the operational uptake of the model for policy purposes, source code based in the 'R' language with the optional use of urban and (non)forest masks for AGBD predictions is released. It includes demonstrations for predicting AGBD in Mexico's Natural Protected Areas, terrestrial ecological strata, and community forest management or payment for environmental services projects, which are commonly used delineations in its climate policy reports. For other nations considering the presented approach for policy purposes, the study discusses challenges concerning the use of EO-based covariates and the limitations of the model. It concludes with a broader call toward ensuring consistency in EO data streams, and prioritizing the co-development of EO-NFI integration approaches with nations in the future, thereby directly addressing their long-term climate policy needs.

Great Lakes Basin landscapes are undergoing rapid land cover and land use (LCLU) change. The goal for this study was to identify changes in land cover occurring in the Great Lakes Basin over three time periods to provide insights into historical land cover changes occurring on a bi-national watershed scale. To quantify potential impacts of anthropogenic changes on important yet vulnerable Great Lakes Wetland ecosystems, the historical changes in land cover over time are assessed via remote sensing. The goal is to better understand legacy effects on current conditions, including wetland gain and loss and the impacts of upland ecosystems on wetland health and water quality. Three key time periods with respect to Great Lakes water level changes and coastal wetland plant invasions were mapped using Landsat-derived land cover maps: 1985, 1995, and 2010. To address change between the three time periods of interest, we incorporate both radiometric and categorical change analysis and open-source tools available for assessing time series data including LandTrendr and TimeSync. Results include maps of annual land cover transition from 1985 to 1995 and 1995 to 2010 basin-wide and by ecoregion and an assessment of the magnitude and direction of change by land cover type. Basin-wide validated change results show approximately 776,854 ha of land changed from c.1980–1995 and approximately 998,400 ha of land changed from c.1995–2010. Both time periods displayed large net decreases in both deciduous forest and agricultural land and net increases in suburban cover. Change by ecoregion is reviewed in this study with many of the change types in central plains showing change in and out of agriculture and suburban land covers, the mixed wood plain ecoregion consisted of a mixture of agricultural, suburban, and forestry changes, and all top five change types in the mixed wood shield consisted of various stages of the forestry cycle for both time periods. In comparison with previous LCLU change studies, overall change products showed similar trends. The discussion reviews why, while most changes had accuracies better than 84%, accuracies found for change from urban to other classes and from other classes to agriculture were lower due to unique aspects of change in these classes which are not relevant for most change analyses applications. The study found a consistent loss in the deciduous forest area for much of the time studied, which is shown to influence the aquatic nitrogen implicated in the expansion of the invasive plant Phragmites australis in the Great Lakes Basin. This underscores the importance of LCLU maps, which allow for the quantification of historical land change in the watersheds of the Great Lakes where invasive species are expanding.

Navigating uncertainty is a critical challenge in all fields of science, especially when translating knowledge into real-world policies or management decisions. However, the wide variance in concepts and definitions of uncertainty across scientific fields hinders effective communication. As a microcosm of diverse fields within Earth Science, NASA’s Carbon Monitoring System (CMS) provides a useful crucible in which to identify cross-cutting concepts of uncertainty. The CMS convened the Uncertainty Working Group (UWG), a group of specialists across disciplines, to evaluate and synthesize efforts to characterize uncertainty in CMS projects. This paper represents efforts by the UWG to build a heuristic framework designed to evaluate data products and communicate uncertainty to both scientific and non-scientific end users. We consider four pillars of uncertainty: origins, severity, stochasticity versus incomplete knowledge, and spatial and temporal autocorrelation. Using a common vocabulary and a generalized workflow, the framework introduces a graphical heuristic accompanied by a narrative, exemplified through contrasting case studies. Envisioned as a versatile tool, this framework provides clarity in reporting uncertainty, guiding users and tempering expectations. Beyond CMS, it stands as a simple yet powerful means to communicate uncertainty across diverse scientific communities.

Aboveground biomass (AGB) estimates derived from Landsat’s spectral bands are limited by spectral saturation when AGB densities exceed 150–300 Mg ha−1. Statistical features that characterize image texture have been proposed as a means to alleviate spectral saturation. However, apart from Gray Level Co-occurrence Matrix (GLCM) statistics, many spatial feature engineering techniques (e.g., morphological operations or edge detectors) have not been evaluated in the context of forest AGB estimation. Moreover, many prior investigations have been constrained by limited geographic domains and sample sizes. We utilize 176 lidar-derived AGB maps covering ∼9.3 million ha of forests in the Pacific Northwest of the United States to construct an expansive AGB modeling dataset that spans numerous biophysical gradients and contains AGB densities exceeding 1000 Mg ha−1. We conduct a large-scale inter-comparison of multiple spatial feature engineering techniques, including GLCMs, edge detectors, morphological operations, spatial buffers, neighborhood vectorization, and neighborhood similarity features. Our numerical experiments indicate that statistical features derived from GLCMs and spatial buffers yield the greatest improvement in AGB model performance out of the spatial feature engineering strategies considered. Including spatial features in Random Forest AGB models reduces the root mean squared error (RMSE) by 9.97 Mg ha−1. We contextualize this improvement model performance by comparing to AGB models developed with multi-temporal features derived from the LandTrendr and Continuous Change Detection and Classification algorithms. The inclusion of temporal features reduces the model RMSE by 18.41 Mg ha−1. When spatial and temporal features are both included in the model’s feature set, the RMSE decreases by 21.71 Mg ha−1. We conclude that spatial feature engineering strategies can yield nominal gains in model performance. However, this improvement came at the cost of increased model prediction bias.

Like much of western North America, The Klamath-Siskiyou Ecoregion of southwest Oregon and northern California is experiencing a departure from its historical fire regime, with greater proportions of high severity fire and less low to moderate severity fire compared to historical norms. Reburn events (i.e. when wildfires burn through areas that have recently burned) can exacerbate this departure when high reburn severity contributes to a conversion from mixed-conifer forests to an alternative stable state dominated by pyrogenic shrubs. Spatially explicit understanding of the predictors of reburn severity is necessary to identify areas where the risk of high reburn severity should be mitigated and areas where low to moderate severity fire can be encouraged to promote resilience to future wildfires. Previous research concerning drivers of reburn severity has considered the severity of the initial fire, topography, and fuel and weather conditions immediately preceding the reburning fire, but has not evaluated the potential effects of vegetation dynamics after a fire on reburn severity. We investigated whether remotely sensed rates of vegetation recovery following an initial fire, measured by change in Normalized Burn Ratio (NBR) over 2, 5, and 10 years, are important predictors of reburn severity, once other known drivers of fire severity have been accounted for. Though a Random Forests analysis revealed that other variables (notably pre-reburn vegetation cover, average fire season PDSI, and initial fire severity) were more important predictors of reburn severity, vegetation recovery rates after an initial fire were found to be moderately important predictors of reburn severity, with comparable importance to fire weather variables and higher importance than some topographical variables. Short-term (2-year and 5-year) vegetation responses to wildfire representing either delayed mortality or elevated vegetation recovery were both associated with higher reburn severity, as were higher 10-year rates of annual vegetation accumulation. The results indicate that along with previously identified drivers of reburn severity, remotely sensed metrics of vegetation recovery relatively soon after a fire may provide important insight that can support post-fire management decisions and wildfire response planning for future reburn events.

Glaciers are important sentinels of a changing climate, crucial components of the global cryosphere and integral to their local landscapes. However, many of the commonly used methods for mapping glacier change are labor-intensive and limit the temporal and spatial scope of existing research. This study addresses some of the limitations of prior approaches by developing a novel deep-learning-based method called GlacierCoverNet. GlacierCoverNet is a deep neural network that relies on an extensive, purpose-built training dataset. Using this model, we created a record of over three decades long at a fine temporal cadence (every two years) for the state of Alaska. We conducted a robust error analysis of this dataset and then used the dataset to characterize changes in debris-free glaciers and supraglacial debris over the last ~35 years. We found that our deep learning model could produce maps comparable to existing approaches in the capture of areal extent, but without manual editing required. The model captured the area covered with glaciers that was ~97% of the Randolph Glacier Inventory 6.0 with ~6% and ~9% omission and commission rates in the southern portion of Alaska, respectively. The overall model area capture was lower and omission and commission rates were significantly higher in the northern Brooks Range. Overall, the glacier-covered area retreated by 8425 km2 (−13%) between 1985 and 2020, and supraglacial debris expanded by 2799 km2 (64%) during the same period across the state of Alaska.

The Landsat record represents an amazing resource for discovering land-cover changes and monitoring the Earth’s surface. However, making the most use of the available data, especially for automated applications ingesting thousands of images without human intervention, requires a robust screening of cloud and cloud-shadow, which contaminate clear views of the land surface. We constructed a deep convolutional neural network (CNN) model to semantically segment Landsat 8 images into regions labeled clear-sky, clouds, cloud-shadow, water, and snow/ice. For training, we constructed a global, hand-labeled dataset of Landsat 8 imagery; this labor-intensive process resulted in the uniquely high-quality dataset needed for the creation of a high-quality model. The CNN model achieves results on par with the ability of human interpreters, with a total accuracy of 97.1%, omitting only 3.5% of cloud pixels and 4.8% of cloud shadow pixels, which is seven to eight times fewer missed pixels than the masks distributed with the imagery. By harnessing the power of advanced tensor processing units, the classification of full images is I/O bound, making this approach a feasible method to generate masks for the entire Landsat 8 archive.

Formal planning and development of what became the first Landsat satellite commenced over 50 years ago in 1967. Now, having collected earth observation data for well over four decades since the 1972 launch of Landsat1, the Landsat program is increasingly complex and vibrant. Critical programmatic elements are ensuring the continuity of high quality measurements for scientific and operational investigations, including ground systems, acquisition planning, data archiving and management, and provision of analysis ready data products. Free and open access to archival and new imagery has resulted in a myriad of innovative applications and novel scientific insights. The planning of future compatible satellites in the Landsat series, which maintain continuity while incorporating technological advancements, has resulted in an increased operational use of Landsat data. Governments and international agencies, among others, can now build an expectation of Landsat data into a given operational data stream. International programs and conventions (e.g., deforestation monitoring, climate change mitigation) are empowered by access to systematically collected and calibrated data with expected future continuity further contributing to the existing multi-decadal record. The increased breadth and depth of Landsat science and applications have accelerated following the launch of Landsat-8, with significant improvements in data quality. Herein, we describe the programmatic developments and institutional context for the Landsat program and the unique ability of Landsat to meet the needs of national and international programs. We then present the key trends in Landsat science that underpin many of the recent scientific and application developments and followup with more detailed thematically organized summaries. The historical context offered by archival imagery combined with new imagery allows for the development of time series algorithms that can produce information on trends and dynamics. Landsat-8 has figured prominently in these recent developments, as has the improved understanding and calibration of historical data. Following the communication of the state of Landsat science, an outlook for future launches and envisioned programmatic developments are presented. Increased linkages between satellite programs are also made possible through an expectation of future mission continuity, such as developing a virtual constellation with Sentinel-2. Successful science and applications developments create a positive feedback loop-justifying and encouraging current and future programmatic support for Landsat.

Few studies have examined post-fire vegetation recovery in temperate forest ecosystems with Landsat time series analysis. We analyzed time series of Normalized Burn Ratio (NBR) derived from LandTrendr spectral-temporal segmentation fitting to examine post-fire NBR recovery for several wildfires that occurred in three different coniferous forest types in western North America during the years 2000 to 2007. We summarized NBR recovery trends, and investigated the influence of burn severity, post-fire climate, and topography on post-fire vegetation recovery via random forest (RF) analysis.

Modeling the spatial and temporal dynamics of soil temperature is deterministically complex due to the wide variability of several influential environmental variables, including soil column composition, soil moisture, air temperature, and solar energy. Landscape incident solar radiation is a significant environmental driver that affects both air temperature and ground-level soil energy loading; therefore, inclusion of solar energy is important for generating accurate representations of soil temperature. We used the U.S. Environmental Protection Agency’s Oregon Crest-to-Coast (O’CCMoN) Environmental Monitoring Transect dataset to develop and test the inclusion of ground-level solar energy driver data within an existing soil temperature model currently utilized within an ecohydrology model called Visualizing Ecosystem Land Management Assessments (VELMA). The O’CCMoN site data elucidate how localized ground-level solar energy between open and forested landscapes greatly influence the resulting soil temperature. We demonstrate how the inclusion of local ground-level solar energy significantly improves the ability to deterministically model soil temperature at two depths. These results suggest that landscape and watershed-scale models should incorporate spatially distributed solar energy to improve spatial and temporal simulations of soil temperature.

One way of analyzing satellite images for land use and land cover change (LULCC) is time series analysis (TSA). Most of the many TSA based LULCC algorithms proposed in the remote sensing community perform well on datasets for which they were designed, but their performance on randomly chosen datasets from across the globe has not been studied. A polyalgorithm combines several basic algorithms, each meant to solve the same problem, producing a strategy that unites the strengths and circumvents the weaknesses of constituent algorithms. The foundation of the proposed TSA based ‘polyalgorithm’ for LULCC is three algorithms (BFAST, EWMACD, and LandTrendR), precisely described mathematically, and chosen to be fundamentally distinct from each other in design and in the phenomena they capture. Analysis of results representing success, failure, and parameter sensitivity for each algorithm is presented. For a given pixel, Hausdorff distance is used to compare the distance between the change times (breakpoints) obtained from two different algorithms. Timesync validation data, a dataset that is based on human interpretation of Landsat time series in concert with historical aerial photography, is used for validation. The polyalgorithm yields more accurate results than EWMACD and LandTrendR alone, but counterintuitively not better than BFAST alone. This nascent work will be directly useful in land use and land cover change studies, of interest to terrestrial science research, especially regarding anthropogenic impacts on the environment.

Background: Satellite-based aboveground forest biomass maps commonly form the basis of forest biomass and carbon stock mapping and monitoring, but biomass maps likely vary in performance by region and as a function of spatial scale of aggregation. Assessing such variability is not possible with spatially-sparse vegetation plot networks. In the current study, our objective was to determine whether high-resolution lidar-based and moderate-resolution Landsat-base aboveground live forest biomass maps converged on similar predictions at stand- to landscape-levels (10 s to 100 s ha) and whether such differences depended on biophysical setting. Specifically, we examined deviations between lidar- and Landsat-based biomass mapping methods across scales and ecoregions using a measure of error (normalized root mean square deviation), a measure of the unsystematic deviations, or noise (Pearson correlation coefficient), and two measures related to systematic deviations, or biases (intercept and slope of a regression between the two sets of predictions). Results: Compared to forest inventory data (0.81-ha aggregate-level), lidar and Landsat-based mean biomass predictions exhibited similar performance, though lidar predictions exhibited less normalized root mean square deviation than Landsat when compared with the reference plot data. Across aggregate-levels, the intercepts and slopes of regression equations describing the relationships between lidar- and Landsat-based biomass predictions stabilized (i.e., little additional change with increasing area of aggregates) at aggregate-levels between 10 and 100 ha, suggesting a consistent relationship between the two maps at landscape-scales. Differences between lidar- and Landsat-based biomass maps varied as a function of forest canopy heterogeneity and composition, with systematic deviations (regression intercepts) increasing with mean canopy cover and hardwood proportion within forests and correlations decreasing with hardwood proportion. Conclusions: Deviations between lidar- and Landsat-based maps indicated that satellite-based approaches may represent general gradients in forest biomass. Ecoregion impacted deviations between lidar and Landsat biomass maps, highlighting the importance of biophysical setting in determining biomass map performance across aggregate scales. Therefore, regardless of the source of remote sensing (e.g., Landsat vs. lidar), factors affecting the measurement and prediction of forest biomass, such as species composition, need to be taken into account whether one is estimating biomass at the plot, stand, or landscape scale.

Landscape solar energy is a significant environmental driver, yet it remains complicated to model well. Several solar radiation models simplify the complexity of light by estimating it at discrete point locations or by averaging values over larger areas. These modeling approaches may be useful in certain cases, but they are unable to provide spatially distributed and temporally dynamic representations of solar energy across entire landscapes. We created a landscape-scale ground-level shade and solar energy model called Penumbra to address this deficiency. Penumbra simulates spatially distributed ground-level shade and incident solar energy at user-defined timescales by modeling local and distant topographic shading and vegetative shading. Spatially resolved inputs of a digital elevation model, a normalized digital surface model, and landscape object transmittance are used to estimate spatial variations in solar energy at user-defined temporal timesteps. The research goals for Penumbra included: 1) simulations of spatiotemporal variations of shade and solar energy caused by both objects and topographic features, 2) minimal user burden and parameterization, 3) flexible user defined temporal parameters, and 4) flexible external model coupling. We test Penumbra's predictive skill by comparing the model's predictions with monitored open and forested sites, and achieve calibrated mean errors ranging from -17.3 to 148.1 μmoles/m2/s. Penumbra is a dynamic model that can produce spatial and temporal representations of shade percentage and ground-level solar energy. Outputs from Penumbra can be used with other ecological models to better understand the health and resilience of aquatic, near stream terrestrial, and upland ecosystems.

California has a pressing need to measure and manage forest carbon. Fusion of satellite-based data with plot-level information provides a promising means to measure forest biomass at relevant spatial and temporal scales. Key questions remain regarding accuracy and feasibility. Over a century of fire suppression complicates managing forest carbon in California's dry forests. Live tree biomass is at risk of loss due to wildfire. Here, we evaluated the performance of an emerging technology using Landsat imagery, forest inventory data, and gradient nearest neighbor imputation (referred to as LT-GNN) to measure annual aboveground live tree biomass (AGB) across multiple spatial scales. We also developed a means to quantify the trade-off between biomass storage and stability for fire-prone forests. We relied on two independent estimates of AGB to evaluate LT-GNN results: local assessments calculated from field-data and airborne light detection, and county estimates calculated from Forest Inventory and Assessment plot results. We also used repeated measurements conducted in Forest Inventory and Analysis plots to quantify the ability of LT-GNN to detect trends in AGB. Finally, we extended a field experiment at Blodgett Forest Research Station in Georgetown, California to gain insights into biomass dynamics of fire-prone forests. LT-GNN is a promising method to monitor live tree biomass. Its success at interpolating county-level tree biomass suggests an application-ready means to track annual biomass at a policy-relevant scale. However, improvements are needed to track change under stable conditions. At finer scales applications must be pursued with more caution. In particular, LTGNN did not accurately predict AGB in an old-growth redwood forest. At Blodgett, we quantified the trade-off between biomass storage and stability. Fuel treatments did lower the overall biomass stored, but more biomass survived fire compared to the untreated forest. However, trade-off between biomass storage and stability critically depends on the probability of fire occurring in these stands.