Publications

In this paper, we describe constructing an algorithm and providing an open-source package to analyze the overall trend and responses of both carbon use efficiency (CUE) and corn yield to climate factors at the continental scale. Our algorithm enables automatic retrieval of remote sensing data through the Google Earth Engine and USDA agricultural production data at the county level across the United States through Application Programming Interface (API). We (1) integrated satellite images of MODIS-based net primary productivity (NPP) and gross primary productivity (GPP), and ECMWF-based climatic variables, (2) calculated CUE and commonly used climate metrics, and then (3) investigated the spatial heterogeneity of the variables. We applied a random forest algorithm to identify the key climate drivers of CUE and crop yield, and estimated the responses of CUE and yield to climate variability using time series data in each county across the United States. Our results show that growing degree days (GDD) has the highest predictive power for both CUE and yield, while extreme degree days (EDD) is the least important explanatory variable. We also found that yield increases with greater GDD and precipitation in most areas of the United States, but with more mixed and fragmented interactions in the southern regions—and that CUE decreases with GDD in the northern regions but increases with GDD in the southern ones. As global warming increases, both carbon sequestration and yield will increase in the south, though CUE will decrease in the north.

Evaluating the interactions between above- and below-ground processes is important to understand and quantify how ecosystems respond differently to atmospheric forcings and/or perturbations and how this depends on their intrinsic characteristics and heterogeneity. Improving such understanding is particularly needed in snow-impacted mountainous systems where the complexity in water and carbon storage and release arises from strong heterogeneity in meteorological forcing and terrain, vegetation and soil characteristics. This study investigates spatial and temporal interactions between terrain, soil moisture, and plant seasonal dynamics at the intra- and inter-annual scale along a 160 m long mountainous, non-forested hillslope-to-floodplain system in the upper East River Watershed in the upper Colorado River Basin. To this end, repeated UAV-based multi-spectral aerial imaging, ground-based soil electrical resistivity imaging, and soil moisture sensors were used to quantify the interactions between above and below-ground compartments. Results reveal significant soil-plant co-dynamics. The spatial variation and dynamics of soil water content and electrical conductivity, driven by topographic and soil intrinsic characteristics, correspond to distinct plant types, with highest plant productivity in convergent areas. Plant productivity in heavy snow years benefited from more water infiltration as well as a shallow groundwater table depth. In comparison, low snowpack years with an early first bare-ground date, which are linked to an early increase in plant greenness, imply a short period of saturated conditions that leads to lower average and maximum greenness values during the growing season. Overall, these results emphasize the strong impact of snowpack dynamics, and terrain and subsurface characteristics on the heterogeneity in plant type and seasonal dynamics.

Landslides are a major natural hazard, threatening communities and infrastructure worldwide. The mitigation of these hazards relies on the understanding of their causes and triggering processes, which depends directly on soil properties, land use, and their changes over time. In this study, we propose a novel framework to estimate the probability of failure in highly developed urban areas. The framework combines remote sensing and geophysical data to estimate soil properties and land covers. Such estimate properties are then integrated into a hydrogeomechanical model to provide a robust estimate of the probability of failure. To assess the importance and sensitivity of the input parameters to the probability of failure assessment, a sensitivity analysis was performed on the seven main parameters (density, friction angle, cohesion, soil thickness, slope, water recharge and saturated hydraulic conductivity) of the hydro-geomechanical model. Slope angle, soil thickness and cohesion are shown to be the most important parameters. While the slope angle can be derived from high-resolution digital elevation models, soil thickness and cohesion cannot be assessed. To incorporate the variability of these two parameters into the model, seismic noise measurements were performed to estimate soil thickness. Supervised classification of remote sensing data was used to map vegetation type and related root cohesion, which can impact the cohesion significantly. The results show that slopes with relatively thick soil layers (above 2 m) have up to four times higher probability of failure. Slopes with tall vegetation cover, and hence comparably high root cohesion, reduce the probability of failure, particularly when the soil layer is relatively thin (< 3 m). The developed approach makes use of rapid to acquire geophysical and easily to obtain remote sensing data, and hence is transferable to other study sites. This approach may be of particular importance to areas of active vegetation management that may cause considerable changes in landslide hazard maps.

Landslide inventory mapping (LIM) is an important application in remote sensing for assisting in the relief of landslide geohazards. However, while conducting LIM tasks performing change detection analysis using bitemporal very high-resolution (VHR) remote sensing images, due to landslide usually occurred in a mountain area, the phenological difference and outcrop rock may bring pseudochanges to LIM results. In this article, a novel change detection approach based on adaptive region shape similarity (ARSS) is proposed for LIM with VHR remote sensing images to improve detection performance. First, an adaptive region around each pixel is extended to explore the contextual information. Then, direction lines within an adaptive region are defined to describe the shape of the adaptive region. Finally, the pixels located on each direction line are taken into account to build the corresponding histogram. The shape similarity between the pairwise histogram curves is measured by using the discrete Frchet distance (DFD). Once the bitemporal images are processed by using the abovementioned steps, a change magnitude image (CMI) is generated, while a threshold is then used to obtain a final binary change map. The proposed approach is applied to three pairs of landslide site images acquired with aerial plane and one land use change dataset acquired by Quick Bird Satellite. Compared with ten state-of-the-art methods, the proposed approach achieved LIMs and detection results with higher accuracies and better performance.

Ecosystems at coastal terrestrial–aquatic interfaces play a significant role in global biogeochemical cycles. In this study, we aimed to characterize coastal wetlands with particular focus on the co-variability between plant dynamics, topography, soil, and other environmental factors. We proposed a functional zonation approach based on machine learning clustering to identify the spatial regions, i.e., zones that capture these co-varied properties. This approach was applied to publicly available datasets along Lake Erie, in the Great Lakes Region. We investigated the heterogeneity of coastal ecosystem structures as a function of along-shore distance and transverse distance, based on the spatial data layers, including topography, wetland vegetation cover, and the time series of Landsat’s enhanced vegetation index (EVI) between 1990 and 2020. Results showed that the topographic metrics (elevation and slope), soil texture, and plant productivity influence the spatial distribution of wetland land-covers (emergent and phragmites). These results highlight a natural organization along the transverse axis, where the elevation and the EVI increase further away from the coastline. In addition, the clustering analysis allowed us to identify regions with distinct environmental characteristics, as well as the ones that are more sensitive to interannual lake-level variations.

Bedrock property quantification is critical for predicting the hydrological response of watersheds to climate disturbances. Estimating bedrock hydraulic properties over watershed scales is inherently difficult, particularly in fracture-dominated regions. Our analysis tests the covariability of above- and belowground features on a watershed scale, by linking borehole geophysical data, near-surface geophysics, and remote sensing data. We use machine learning to quantify the relationships between bedrock geophysical/hydrological properties and geomorphological/ vegetation indices and show that machine learning relationships can estimate most of their covariability. Although we can predict the electrical resistivity variation across the watershed, regions of lower variability in the input parameters are shown to provide better estimates, indicating a limitation of commonly applied geomorphological models. Our results emphasize that such an integrated approach can be used to derive detailed bedrock characteristics, allowing for identification of small-scale variations across an entire watershed that may be critical to assess the impact of disturbances on hydrological systems.

Understanding the interactions among agricultural processes, soil, and plants is necessary for optimizing crop yield and productivity. This study focuses on developing effective monitoring and analysis methodologies that estimate key soil and plant properties. These methodologies include data acquisition and processing approaches that use unmanned aerial vehicles (UAVs) and surface geophysical techniques. In particular, we applied these approaches to a soybean farm in Arkansas to characterize the soil–plant coupled spatial and temporal heterogeneity, as well as to identify key environmental factors that influence plant growth and yield. UAV‑based multitemporal acquisition of high‑resolution RGB (red–green–blue) imagery and direct measurements were used to monitor plant height and photosynthetic activity. We present an algorithm that efficiently exploits the highresolution UAV images to estimate plant spatial abundance and plant vigor throughout the growing season. Such plant characterization is extremely important for the identification of anomalous areas, providing easily interpretable information that can be used to guide near‑real‑time farming decisions. Additionally, high‑resolution multitemporal surface geophysical measurements of apparent soil electrical conductivity were used to estimate the spatial heterogeneity of soil texture. By integrating the multiscale multitype soil and plant datasets, we identified the spatiotemporal co‑variance between soil properties and plant development and yield. Our novel approach for early season monitoring of plant spatial abundance identified areas of low productivity controlled by soil clay content, while temporal analysis of geophysical data showed the impact of soil moisture and irrigation practice (controlled by topography) on plant dynamics. Our study demonstrates the effective coupling of UAV data products with geophysical data to extract critical information for farm management.

In complex terrain, non-parallel surfaces receive emitted radiation from adjacent surfaces. Qualitatively, where surface skin temperatures and lower tropospheric temperature and humidity are not uniform, the downwelling longwave radiation (DLR) will be determined not just by radiation from the atmosphere above a given location, but also by adjacent surface temperatures. We quantify this three-dimensional longwave radiative effect over the Upper Colorado River Basin in clear-sky conditions by calculating surface DLR with observed land-surface temperatures from ECOSTRESS. We find that this effect is due to terrain-subtended sky-view and represents ∼22% of the surface longwave flux, rising to ∼28% and ∼24% in the East and Southeast of the Basin, respectively, and can be >50% in extreme cases. The common omission of this effect in atmospheric radiation models leads to an underestimation of DLR in complex terrain, especially at higher elevations, which has significant implications for mountainous ecohydrology simulations.

In this study, we develop a watershed zonation approach for characterizing watershed organization and functions in a tractable manner by integrating multiple spatial data layers. We hypothesize that (1) a hillslope is an appropriate unit for capturing the watershed-scale heterogeneity of key bedrock-through-canopy properties and for quantifying the co-variability of these properties representing coupled ecohydrological and biogeochemical interactions, (2) remote sensing data layers and clustering methods can be used to identify watershed hillslope zones having the unique distributions of these properties relative to neighboring parcels, and (3) property suites associated with the identified zones can be used to understand zone-based functions, such as response to early snowmelt or drought and solute exports to the river. We demonstrate this concept using unsupervised clustering methods that synthesize airborne remote sensing data (lidar, hyperspectral, and electromagnetic surveys) along with satellite and streamflow data collected in the East River Watershed, Crested Butte, Colorado, USA. Results show that (1) we can define the scale of hillslopes at which the hillslope-averaged metrics can capture the majority of the overall variability in key properties (such as elevation, net potential annual radiation, and peak snow-water equivalent – SWE), (2) elevation and aspect are independent controls on plant and snow signatures, (3) near-surface bedrock electrical resistivity (top 20 m) and geological structures are significantly correlated with surface topography and plan species distribution, and (4) K-means, hierarchical clustering, and Gaussian mixture clustering methods generate similar zonation patterns across the watershed. Using independently collected data, we show that the identified zones provide information about zone-based watershed functions, including foresummer drought sensitivity and river nitrogen exports. The approach is expected to be applicable to other sites and generally useful for guiding the selection of hillslope-experiment locations and informing model parameterization.

Climate change is reshaping vulnerable ecosystems, leading to uncertain effects on ecosystem dynamics, including evapotranspiration (ET) and ecosystem respiration (Reco). However, accurate estimation of ET and Reco still remains challenging at sparsely monitored watersheds, where data and field instrumentation are limited. In this study, we developed a hybrid predictive modeling approach (HPM) that integrates eddy covariance measurements, physically based model simulation results, meteorological forcings, and remote-sensing datasets to estimate ET and Reco in high space–time resolution. HPM relies on a deep learning algorithm and long short-term memory (LSTM) and requires only air temperature, precipitation, radiation, normalized difference vegetation index (NDVI), and soil temperature (when available) as input variables. We tested and validated HPM estimation results in different climate regions and developed four use cases to demonstrate the applicability and variability of HPM at various FLUXNET sites and Rocky Mountain SNOTEL sites in Western North America. To test the limitations and performance of the HPM approach in mountainous watersheds, an expanded use case focused on the East River Watershed, Colorado, USA. The results indicate HPM is capable of identifying complicated interactions among meteorological forcings, ET, and Reco variables, as well as providing reliable estimation of ET and Reco across relevant spatiotemporal scales, even in challenging mountainous systems. The study documents that HPM increases our capability to estimate ET and Reco and enhances process understanding at sparsely monitored watersheds.

Soil thickness plays a central role in the interactions between vegetation, soils, and topography, where it controls the retention and release of water, carbon, nitrogen, and metals. However, mapping soil thickness, here defined as the mobile regolith layer, at high spatial resolution remains challenging. Here, we develop a hybrid model that combines a process-based model and empirical relationships to estimate the spatial heterogeneity of soil thickness with fine spatial resolution (0.5 m). We apply this model to two aspects of hillslopes (southwest and northeast-facing, respectively) in the East River watershed in Colorado. Two independent measurement methods – auger and cone penetrometer – are used to sample soil thickness at 78 locations to calibrate the local value of unconstrained parameters within the hybrid model. Sensitivity analysis using the hybrid model reveals that the diffusion coefficient used in hillslope diffusion modeling has the largest sensitivity among all input parameters. In addition, our results from both sampling and modeling show that, in general, the northeast-facing hillslope has a deeper soil layer than the southwest-facing hillslope. By comparing the soil thickness estimated between a machine-learning approach and this hybrid model, the hybrid model provides higher accuracy and requires less sampling data. Modeling results further reveal that the southwest-facing hillslope has a slightly faster surface soil erosion rate and soil production rate than the northeast-facing hillslope, which suggests that the relatively less dense vegetation cover and drier surface soils on the southwest-facing slopes influence soil properties. With seven parameters in total for calibration, this hybrid model can provide a realistic soil thickness map with a relatively small amount of sampling dataset comparing to machine-learning approach. Integrating process-based modeling and statistical analysis not only provides a thorough understanding of the fundamental mechanisms for soil thickness prediction but also integrates the strengths of both statistical approaches and process-based modeling approaches.

‍

Change vector analysis (CVA) is a simple yet attractive method to detect changes with remote sensing images. Since its first introduction in 1980, CVA has received increased attention from the remote sensing community, leading to the definition of several new methodologies based on the CVAs concept while extending its applicability. In this article, we provide an extensive review of CVA-based approaches in the context of land-cover change detection (LCCD). We first reviewed the development of the CVA-based LCCD method with remote sensing images, and some classical-related methods were discussed. Then, we analyze and compare the performance of five selected methods. The analysis was carried out on seven real datasets acquired by different sensors and platforms (e.g., Landsat, Quick Bird, and airborne) and spatial resolutions (from 0.5 to 30 m/pixel), with scenes from both urban and natural landscapes. The analysis shows several findings, which include that the performance of CVA-based approaches is, in general, resolution dependent, and the detection accuracies of a specific method vary with different input datasets, for example, when applying the classical CVA to the datasets with a resolution from 0.5 to 30 m/pixel, the accuracy of false alarm (FA) ranges from 2.26% to 23.22%. Furthermore, the diagnoses also remind that the detection accuracies for a specific method varied with the size of the area being considered for a given dataset, such as when applying deep CVA (DCVA) to the image with 200 × 200 to 1600 × 1600 size of the Landsat dataset, the FA of DCVA reduced from 11.8% to 1.0%. Moreover, comparing the detection accuracies of different methods implies that the content of an image scene still plays an important role when disregarding the unique preferences of different methods.

Land cover change detection (LCCD) with remote sensing images is an important application of Earth observation data because it provides insights into environmental health, global warming, and city management. In particular, very-high-resolution (VHR) remote sensing images can capture details of a ground object and offer an opportunity to detect land cover changes in detail. However, VHR images usually have high spatial resolution but contain limited spectral information. Therefore, LCCD with VHR optimal images performs poorly because of high intraclass variation and low interclass variance. In the past decades, various approaches have been promoted to smoothen noise, reduce pseudochanges, and preserve the details of detection maps with VHR images. In this article, we first present an overview of the main issues in terms of algorithms, applications, and referred characteristics to promote a comprehensive and general understanding of the development of LCCD with VHR remote sensing images. Several key methodologies are compared with three pairs of real VHR optimal images. Finally, we discuss future challenges and opportunities in applying VHR remote sensing images in LCCD.

Background: Biogeochemical exports from watersheds are modulated by the activity of microorganisms that function over micron scales. Here, we tested the hypothesis that meander-bound regions share a core microbiome and exhibit patterns of metabolic potential that broadly predict biogeochemical processes in floodplain soils along a river corridor.

Results: We intensively sampled the microbiomes of floodplain soils located in the upper, middle, and lower reaches of the East River, Colorado. Despite the very high microbial diversity and complexity of the soils, we reconstructed 248 quality draft genomes representative of subspecies. Approximately one third of these bacterial subspecies was detected across all three locations at similar abundance levels, and ~ 15% of species were detected in two consecutive years. Within the meander-bound floodplains, we did not detect systematic patterns of gene abundance based on sampling position relative to the river. However, across meanders, we identified a core floodplain microbiome that is enriched in capacities for aerobic respiration, aerobic CO oxidation, and thiosulfate oxidation with the formation of elemental sulfur. Given this, we conducted a transcriptomic analysis of the middle floodplain. In contrast to predictions made based on the prominence of gene inventories, the most highly transcribed genes were relatively rare amoCAB and nxrAB (for nitrification) genes, followed by genes involved in methanol and formate oxidation, and nitrogen and CO2 fixation. Within all three meanders, low soil organic carbon correlated with high activity of genes involved in methanol, formate, sulfide, hydrogen, and ammonia oxidation, nitrite oxidoreduction, and nitrate and nitrite reduction. Overall, the results emphasize the importance of sulfur, one-carbon and nitrogen compound metabolism in soils of the riparian corridor.

Conclusions: The disparity between the scale of a microbial cell and the scale of a watershed currently limits the development of genomically informed predictive models describing watershed biogeochemical function. Meanderbound floodplains appear to serve as scaling motifs that predict aggregate capacities for biogeochemical transformations, providing a foundation for incorporating riparian soil microbiomes in watershed models. Widely represented genetic capacities did not predict in situ activity at one time point, but rather they define a reservoir of biogeochemical potential available as conditions change.

‍

Working within a vineyard in the Pessac Léognan Appellation of Bordeaux, France, this study documents the potential of using simple statistical methods with spatially-resolved and increasingly available electromagnetic induction (EMI) geophysical and normalized difference vegetation index (NDVI) datasets to accurately estimate Bordeaux vineyard soil classes and to quantitatively explore the relationship between vineyard soil types and grapevine vigor. First, co-located electrical tomographic tomography (ERT) and EMI datasets were compared to gain confidence about how the EMI method averaged soil properties over the grapevine rooting depth. Then, EMI data were used with core soil texture and soil-pit based interpretations of Bordeaux soil types (Brunisol, Redoxisol, Colluviosol and Calcosol) to estimate the spatial distribution of geophysically-identified Bordeaux soil classes. A strong relationship (r = 0.75, p< 0.01) was revealed between the geophysicallyidentified Bordeaux soil classes and NDVI (both 2 m resolution), showing that the highest grapevine vigor was associated with the Bordeaux soil classes having the largest clay fraction. The results suggest that within-block variability of grapevine vigor was largely controlled by variability in soil classes, and that carefully collected EMI and NDVI datasets can be exceedingly helpful for providing quantitative estimates of vineyard soil and vigor variability, as well as their covariation. The method is expected to be transferable to other viticultural regions, providing an approach to use easy-to-acquire, high resolution datasets to guide viticultural practices, including routine management and replanting.

Climate warming in alpine regions is changing patterns of water storage, a primary control on alpine plant ecology, biogeochemistry, and water supplies to lower elevations. There is an outstanding need to determine how the interacting drivers of precipitation and the critical zone (CZ) dictate the spatial pattern and time evolution of soil water storage. In this study, we developed an analytical framework that combines intensive hydrologic measurements and extensive remotely-sensed observations with statistical modeling to identify areas with similar temporal trends in soil water storage within, and predict their relationships across, a 0.26 km2 alpine catchment in the Colorado Rocky Mountains, U.S.A. Repeat measurements of soil moisture were used to drive an unsupervised clustering algorithm, which identified six unique groups of locations ranging from predominantly dry to persistently very wet within the catchment. We then explored relationships between these hydrologic groups and multiple CZ-related indices, including snow depth, plant productivity, macro- (102->103 m) and microtopography (<100-102 m), and hydrological flow paths. Finally, we used a supervised machine learning random forest algorithm to map each of the six hydrologic groups across the catchment based on distributed CZ properties and evaluated their aggregate relationships at the catchment scale. Our analysis indicated that ∼40–50% of the catchment is hydrologically connected to the stream channel, lending insight into the portions of the catchment that likely dominate stream water and solute fluxes. This research expands our understanding of patch-to-catchment-scale physical controls on hydrologic and biogeochemical processes, as well as their relationships across space and time, which will inform predictive models aimed at determining future changes to alpine ecosystems.

In the headwater catchments of the Rocky Mountains, plant productivity and its dynamics are largely dependent upon water availability, which is influenced by changing snowmelt dynamics associated with climate change. Understanding and quantifying the interactions between snow, plants and soil moisture is challenging, since these interactions are highly heterogeneous in mountainous terrain, particularly as they are influenced by microtopography within a hillslope. Recent advances in satellite remote sensing have created an opportunity for monitoring snow and plant dynamics at high spatiotemporal resolutions that can capture microtopographic effects. In this study, we investigate the relationships among topography, snowmelt, soil moisture and plant dynamics in the East River watershed, Crested Butte, Colorado, based on a time series of 3-meter resolution PlanetScope normalized difference vegetation index (NDVI) images. To make use of a large volume of high-resolution time-lapse images (17 images total), we use unsupervised machine learning methods to reduce the dimensionality of the time lapse images by identifying spatial zones that have characteristic NDVI time series. We hypothesize that each zone represents a set of similar snowmelt and plant dynamics that differ from other identified zones and that these zones are associated with key topographic features, plant species and soil moisture. We compare different distance measures (Ward and complete linkage) to understand the effects of their influence on the zonation map. Results show that the identified zones are associated with particular microtopographic features; highly productive zones are associated with low slopes and high topographic wetness index, in contrast with zones of low productivity, which are associated with high slopes and low topographic wetness index. The zones also correspond to particular plant species distributions; higher forb coverage is associated with zones characterized by higher peak productivity combined with rapid senescence in low moisture conditions, while higher sagebrush coverage is associated with low productivity and similar senescence patterns between high and low moisture conditions. In addition, soil moisture probe and sensor data confirm that each zone has a unique soil moisture distribution. This cluster-based analysis can tractably analyze high-resolution time-lapse images to examine plant-soil-snow interactions, guide sampling and sensor placements and identify areas likely vulnerable to ecological change in the future.

1. In recent years, the availability of airborne imaging spectroscopy (hyperspectral) data has expanded dramatically. The high spatial and spectral resolution of these data uniquely enable spatially explicit ecological studies including species mapping, assessment of drought mortality and foliar trait distributions. However, we have barely begun to unlock the potential of these data to use direct mapping of vegetation characteristics to infer subsurface properties of the critical zone. To assess their utility for Earth systems research, imaging spectroscopy data acquisitions require integration with large, coincident ground-based datasets collected by experts in ecology and environmental and Earth science. Without coordinated, well-planned field campaigns, potential knowledge leveraged from advanced airborne data collections could be lost. Despite the growing importance of this field, documented methods to couple such a wide variety of disciplines remain sparse. 2. We coordinated the first National Ecological Observatory Network Airborne Observation Platform (AOP) survey performed outside of their core sites, which took place in the Upper East River watershed, Colorado. Extensive planning for sample tracking and organization allowed field and flight teams to update the ground-based sampling strategy daily. This enabled collection of an extensive set of physical samples to support a wide range of ecological, microbiological, biogeochemical and hydrological studies. 3. We present a framework for integrating airborne and field campaigns to obtain high-quality data for foliar trait prediction and document an archive of coincident physical samples collected to support a systems approach to ecological research in the critical zone. This detailed methodological account provides an example of how a multi-disciplinary and multi-institutional team can coordinate to maximize knowledge gained from an airborne survey, an approach that could be extended to other studies. 4. The coordination of imaging spectroscopy surveys with appropriately timed and extensive field surveys, along with high-quality processing of these data, presents a unique opportunity to reveal new insights into the structure and dynamics of the critical zone. To our knowledge, this level of co-aligned sampling has never been undertaken in tandem with AOP surveys and subsequent studies utilizing this archive will shed considerable light on the breadth of applications for which imaging spectroscopy data can be leveraged.

Long-term plot-scale studies have found water limitation to be a key factor driving ecosystem productivity in the Rocky Mountains. Specifically, the intensity of early summer (the ‘foresummer’ period from May to June) drought conditions appears to impose critical controls on peak ecosystem productivity. This study aims to (1) assess the importance of early snowmelt and foresummer drought in controlling peak plant productivity, based on the historical Landsat normalized-difference vegetation index (NDVI) and climate data; (2) map the spatial heterogeneity of foresummer drought sensitivity; and (3) identify the environmental controls (e.g. geomorphology, elevation, geology, plant types) on drought sensitivity. Our domain (15 × 15 km) includes four drainages within the East Water watershed near Gothic, Colorado, USA. We define foresummer drought sensitivity based on the regression slopes of the annual peak NDVI against the June Palmer Drought Severity Index between 1992 and 2010. Results show that foresummer drought sensitivity is spatially heterogeneous, and primarily dependent on the plant type and elevation. In support of the plot-based studies, we find that years with earlier snowmelt and drier foresummer conditions lead to lower peak NDVI; particularly in the low-elevation regions. Using random forest analysis, we identify additional key controls related to surface energy exchanges (i.e. potential net radiation), hydrological processes (i.e. microtopography and slope), and underlying geology. This remote-sensing-based approach for quantifying foresummer drought sensitivity can be used to identify the regions that are vulnerable or resilient to climate perturbations, as well as to inform future sampling, characterization, and modeling studies.

Morphological attribute profiles are multilevel decompositions of images obtained with a sequence of transformations performed by connected operators. They have been extensively employed in performing multiscale and region-based analysis in a large number of applications. One main, still unresolved, issue is the selection of filter parameters able to provide representative and non-redundant threshold decomposition of the image. This paper presents a framework for the automatic selection of filter thresholds based on Granulometric Characteristic Functions (GCFs). GCFs describe the way that non-linear morphological filters simplify a scene according to a given measure. Since attribute filters rely on a hierarchical representation of an image (e.g., the Tree of Shapes) for their implementation, GCFs can be efficiently computed by taking advantage of the tree representation. Eventually, the study of the GCFs allows the identification of a meaningful set of thresholds. Therefore, a trial and error approach is not necessary for the threshold selection, automating the process and in turn decreasing the computational time. It is shown that the redundant information is reduced within the resulting profiles (a problem of high occurrence, as regards manual selection). The proposed approach is tested on two real remote sensing data sets, and the classification results are compared with strategies present in the literature.

Random Forest (RF) is a widely used classifier to show a good performance of hyperspectral data classification. However, such performance could be improved by increasing the diversity that characterizes the ensemble architecture. In this paper, we propose a novel ensemble approach, namely rotation random forest via kernel principal component analysis (RoRF-KPCA). In particular, the original feature space is first randomly split into several subsets, and KPCA is performed on each subset to extract high order statistics. The obtained feature sets are merged and used as input to an RF classifier. Finally, the results achieved at each step are fused by a majority vote. Experimental analysis is conducted using real hyperspectral remote sensing images to evaluate the performance of the proposed method in comparison with RF, rotation forest, support vector machines, and RoRF-PCA. The obtained results demonstrate the effectiveness of the proposed method.

In this paper, we propose a new version of ro- tation forest (RoF) method for the pixel-wise classification of hyperspectral image. RoF, which is an ensemble of decision tree classifiers, uses random feature selection and data transformation techniques (i.e. principal component analysis) to improve both the accuracy of base classifiers and the diversity within the ensemble. Traditional RoF performs data transformation on the training samples of each subset. In order to further improve the perfor- mance of RoF, the data transformation is separately performed on each class, extracting sets of transformation matrices that are strictly dependent on the training samples of each single class. The approach, namely class separation-based RoF (RoF_CS), is experimentally investigated on a hyperspectral image collected by Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) sensor. Experimental results demonstrate that the proposed methodology achieves excellent performances in comparisons with random forest (RF) and RoF classifiers. Index

‍

The analysis of multi-temporal remote-sensing images is one of the main applications in Earth's observation and monitoring. In this paper, we present a Matlab toolbox for change detection analysis of optical multi-temporal remote-sensing data in which unsupervised approaches, iterative principal component analysis (ITPCA), and iteratively reweighted multivariate alteration detec- tion (IR-MAD) are implemented and optimized. The optimization is represented by the implementation of novel pre- and post-proces- sing strategies that aim to mitigate the side effects introduced by different acquisition conditions affecting change detection analy- sis. Special modules have been designed in order to decrease the required memory when large data sets are processed.

‍

The availability of hyperspectral images with improved spectral and spatial resolutions, provides the opportunity to obtain accurate land-cover classification. In this paper, a novel methodology that combines spectral and spatial information for supervised hyperspectral image classification is proposed. A feature reduction strategy based on Independent Component Analysis (ICA) is the main core of the spectral analysis, where the exploitation of prior information coupled to the evaluation of the reconstruction error assures the identification of the best class-informative sub-set of independent components. Reduced Attribute Profiles (rAPs), designed to address well known issues related to information redundancy that affect the common morphological APs, are then employed for the modelling and fusion of the contextual information. Four real hyperspectral data sets, characterized by different spectral and spatial resolution with a variety of scene typologies (urban, agriculture areas), have been used for assessing the accuracy and generalization capabilities of the proposed methodology. The obtained results demonstrate the classification effectiveness of the proposed approach in all different scene typologies, with respect to other state-of-the-art techniques.

‍

This paper presents a thorough study on the performances of different independent component analysis (ICA) algorithms for the extraction of class-discriminant information in remote sensing hyperspectral image classification. The study considers the three implementations of ICA that are most widely used in signal processing, namely Infomax, FastICA, and JADE. The analysis aims to address a number of important issues regarding the use of ICA in the RS domain. Three scenarios are considered and the performances of the ICA algorithms are evaluated and compared against each other, in order to reach the final goal of identifying the most suitable approach to the analysis of hyperspectral images in supervised classification. Different feature extraction and selection techniques are used for dimensionality reduction with ICA and are then compared to the commonly used strategy, which is based on preprocessing data with principal components analysis (PCA) prior to classification. Experimental results obtained on three real hyperspectral data sets from each of the considered algorithms are presented and analyzed in terms of both classification accuracies and computational time.

A new approach to change detection in very high resolution remote sensing images based on morphological attribute profiles (APs) is presented. A multiresolution contextual transformation performed by APs allows the extraction of geometrical features related to the structures within the scene at different scales. The temporal changes are detected by comparing the geometrical features extracted from the image of each date. The experiments performed on panchromatic QuickBird images related to an urban area show the effectiveness of the proposed technique in detecting changes on the basis of the spatial morphology by preserving geometrical detail.