Search

Efforts to monitor the broad-scale impacts of drought on forests often come up short. Drought is a direct stressor of forests as well as a driver of secondary disturbance agents, making a full accounting of drought impacts challenging. General impacts can be inferred from moisture deficits quantified using precipitation and temperature measurements. However, derived meteorological indices may not meaningfully capture drought impacts because drought responses can differ substantially among species, sites and regions. Meteorology-based approaches also require the characterization of current moisture conditions relative to some specified time and place, but defining baseline conditions over large, ecologically diverse regions can be as difficult as quantifying the moisture deficit itself. In contrast, remote sensing approaches attempt to observe immediate, secondary, and longer-term changes in vegetation response, yet they too are no panacea. Remote sensing methods integrate responses across entire mixed-vegetation pixels and rarely distinguish the effects of drought on a single species, nor can they disentangle drought effects from those caused by various other disturbance agents. Establishment of suitable baselines from remote sensing may be even more challenging than with meteorological data. Here we review broad-scale drought monitoring methods, and suggest that an integrated data-mining approach may hold the most promise for enhancing our ability to resolve drought impacts on forests. A big-data approach that integrates meteorological and remotely sensed data streams, together with other datasets such as vegetation type, wildfire occurrence and pest activity, can clarify direct drought effects while filtering indirect drought effects and consequences. This strategy leverages the strengths of meteorology-based and remote sensing approaches with the aid of ancillary data, such that they complement each other and lead toward a better understanding of drought impacts. [ABSTRACT FROM AUTHOR]

Abstract: We investigated the use of principal components analysis (PCA) to visualize dominant patterns and identify anomalies in a multi-year land surface phenology data set (231 m × 231 m normalized difference vegetation index (NDVI) values derived from the Moderate Resolution Imaging Spectroradiometer (MODIS)) used for detecting threats to forest health in the conterminous United States (CONUS). Our goal is to find ways that PCA can be used with this massive data set to automate the process of detecting forest disturbance and attributing it to particular agents. We briefly describe the parallel computational approaches we used to make PCA feasible, and present some examples in which we have used it to visualize the seasonal vegetation phenology for the CONUS and to detect areas where anomalous NDVI traces suggest potential threats to forest health. [Copyright &y& Elsevier]

Abstract: Identification of geographic ecoregions has long been of interest to environmental scientists and ecologists for identifying regions of similar ecological and environmental conditions. Such classifications are important for predicting suitable species ranges, for stratification of ecological samples, and to help prioritize habitat preservation and remediation e_orts. Hargrove and Ho_man have developed geographical spatio-temporal clustering algorithms and codes and have successfully applied them to a variety of environmental science domains, including ecological regionalization; environmental monitoring network design; analysis of satellite-, airborne-, and ground-based remote sensing, and climate model-model and model-measurement intercomparison. With the advances in state-of-the-art satellite remote sensing and climate models, observations and model outputs are available at increasingly high spatial and temporal resolutions. Long time series of these high resolution datasets are extremely large in size and growing. Analysis and knowledge extraction from these large datasets are not just algorithmic and ecological problems, but also pose a complex computational problem. This paper focuses on the development of a massively parallel multivariate geographical spatio-temporal clustering code for analysis of very large datasets using tens of thousands processors on one of the fastest supercomputers in the world. [Copyright &y& Elsevier]

Abstract: From field-scale measurements to global climate simulations and remote sensing, the growing body of very large and long time series Earth science data are increasingly di_cult to analyze, visualize, and interpret. Data mining, information theoretic, and machine learning techniques—such as cluster analysis, singular value decomposition, block entropy, Fourier and wavelet analysis, phase-space reconstruction, and artificial neural networks—are being applied to problems of segmentation, feature extraction, change detection, model-data comparison, and model validation. The size and complexity of Earth science data exceed the limits of most analysis tools and the capacities of desktop computers. New scalable analysis and visualization tools, running on parallel cluster computers and supercomputers, are required to analyze data of this magnitude. This workshop will demonstrate how data mining techniques are applied in the Earth sciences and describe innovative computer science methods that support analysis and discovery in the Earth sciences. [Copyright &y& Elsevier]

Abstract: We investigate methods for geospatiotemporal data mining of multi-year land surface phenology data (250 m2 Normalized Di_erence Vegetation Index (NDVI) values derived from the Moderate Resolution Imaging Spectrometer (MODIS) in this study) for the conterminous United States (CONUS) as part of an early warning system for detecting threats to forest ecosystems. The approaches explored here are based on k-means cluster analysis of this massive data set, which provides a basis for defining the bounds of the expected or “normal” phenological patterns that indicate healthy vegetation at a given geographic location. We briefly describe the computational approaches we have used to make cluster analysis of such massive data sets feasible, describe approaches we have explored for distinguishing between normal and abnormal phenology, and present some examples in which we have applied these approaches to identify various forest disturbances in the CONUS. [Copyright &y& Elsevier]

We describe a polar coordinate transformation of vegetation index profiles which permits a broad-scale comparison of location-specific phenological variability influenced by climate, topography, land use, and other factors. We apply statistical data reduction techniques to identify fundamental dimensions of phenological variability and to classify phenological types with intuitive ecological interpretation. Remote sensing-based land surface phenology can reveal ecologically meaningful vegetational diversity and dynamics across broad landscapes. Land surface phenology is inherently complex at regional to continental scales, varying with latitude, elevation, and multiple biophysical factors. Quantifying phenological change across ecological gradients at these scales is a potentially powerful way to monitor ecological development, disturbance, and diversity. Polar coordinate transformation was applied to Moderate Resolution Imaging Spectroradiometer (MODIS) normalized difference vegetation index (NDVI) time series spanning 2000-2018 across North America. In a first step, 46 NDVI values per year were reduced to 11 intuitive annual metrics, such as the midpoint of the growing season and degree of seasonality, measured relative to location-specific annual phenological cycles. Second, factor analysis further reduced these metrics to fundamental phenology dimensions corresponding to annual timing, productivity, and seasonality. The factor analysis explained over 95% of the variability in the metrics and represented a more than ten-fold reduction in data volume from the original time series. In a final step, phenological classes ('phenoclasses') based on the statistical clustering of the factor data, were computed to describe the phenological state of each pixel during each year, which facilitated the tracking of year-to-year dynamics. Collectively the phenology metrics, factors, and phenoclasses provide a system for characterizing land surface phenology and for monitoring phenological change that is indicative of ecological gradients, development, disturbance, and other aspects of landscape-scale diversity and dynamics. [ABSTRACT FROM AUTHOR]