Introduction Understanding the links between animal distribution and habitat plays a pivotal role... more Introduction Understanding the links between animal distribution and habitat plays a pivotal role in designing management for threatened species (Lecis and Noris, 2003). While an understanding of habitat preferences is important to improve ecological knowledge about a species, ...
This paper investigates people’s perceptions toward King Cobras in the tropical rainforests of th... more This paper investigates people’s perceptions toward King Cobras in the tropical rainforests of the Western Ghats ecoregion of southern India. We built logistic regression models to test if people’s perceptions (to kill/not to kill the snake) were influenced by factors such as the snake’s size and defensiveness, whether the snake was found near human habitation, the time of encounter and season. The model correctly classified over 80% of instances when people expressed an inclination to kill the snake. Results support our expectation that the snake’s defensiveness escalates the probability that the snake will be killed, but are contrary in that smaller snakes are more likely to be killed than larger ones, especially when encountered away from human habitation. Findings suggest a need for slight refocusing of King Cobra conservation outreach efforts towards smaller snakes, especially in regions where sizeable human habitations exist near fragmented King Cobra habitat.
Introduction Understanding the links between animal distribution and habitat plays a pivotal role... more Introduction Understanding the links between animal distribution and habitat plays a pivotal role in designing management for threatened species (Lecis and Noris, 2003). While an understanding of habitat preferences is important to improve ecological knowledge about a species, ...
Ecological applications : a publication of the Ecological Society of America, 2015
A major goal of remote sensing is the development of generalizable algorithms to repeatedly and a... more A major goal of remote sensing is the development of generalizable algorithms to repeatedly and accurately map ecosystem properties across space and time. Imaging spectroscopy has great potential to map vegetation traits that cannot be retrieved from broadband spectral data, but rarely have such methods been tested across broad regions. Here we illustrate a general approach for estimating key foliar chemical and morphological traits through space and time using NASA's Airborne Visible/Infrared Imaging Spectrometer (AVIRIS-Classic). We apply partial least squares regression (PLSR) to data from 237 field plots within 51 images acquired between 2008 and 2011. Using a series of 500 randomized 50/50 subsets of the original data, we generated spatially explicit maps of seven traits (leaf mass per area (M(area)), percentage nitrogen, carbon, fiber, lignin, and cellulose, and isotopic nitrogen concentration, δ15N) as well as pixel-wise uncertainties in their estimates based on error pro...
ABSTRACT Recent and forthcoming developments in imaging spectroscopy will soon provide unpreceden... more ABSTRACT Recent and forthcoming developments in imaging spectroscopy will soon provide unprecedented opportunities for assessing ecosystem function. Research has shown that spectroscopic methods have the potential to characterize key plant functional traits such as foliar nitrogen content (Majeke et al. 2008; Martin et al. 2008; Townsend et al. 2003), specific leaf area (SLA), foliar lignin and cellulose (Kokaly et al. 2009; Majeke et al. 2008) and δN15 concentrations (Elmore and Craine 2011; Kleinebecker et al. 2009) across landscapes. Moreover, it has been shown that variables that can be derived from imaging spectroscopy - foliar nitrogen and lignin/cellulose concentrations - characterize leaf lifespan (Santiago 2007; Shipley et al. 2006; Wright et al. 2005), influence soil C:N ratios (Ollinger et al. 2002) and in combination with disturbance events (Eshleman 2000; McNeil et al. 2007), may be strong controllers of nutrient cycling in forested landscapes (Aber et al. 1991; de Bello et al. 2010; Fortunel et al. 2009). New methods and improved signal-to-noise of imaging spectroscopy instruments (or hyperspectral imaging) have enabled the assessment of variations in forest functional traits across space and time (Asner and Vitousek 2005; Chambers et al. 2007; Martin and Aber 1997). However, standardized protocols for pre-processing airborne spectroscopic imagery do not exist in contrast to spectroscopic methods for chemometric analysis. Given the large volume of data forthcoming from the NEON AOP and future spaceborne missions such as ENMAP and HyspIRI, it is essential that assessments be performed to identify the consequences of pre-processing on consistent retrieval of forest traits across images and through time. Information retrieval from airborne spectroscopic imagery is especially difficult due to the multitude of artifacts in hyperspectral imagery. Whereas atmospheric effects can be partly corrected using established radiative transfer models, additional artifacts such as from along and across-track brightness gradients due to solar and sensor geometry, differential illumination effects due to complex terrain, and occlusions due to clouds and cloud shadows can persist. In combination, these effects may manifest themselves in the quality of spectra retrieved from radiative transfer models and eventually effect retrievals of key structural and biochemical traits of forest canopies. Here, we expand upon the methods and techniques (Figure 1) developed at the University of Wisconsin for dealing with many of these issues using a large database of spectroscopic imagery (~150 images) collected over three years (2008-2011) across different ecosystems in the US. Sites are located in the Rocky Mountains, the Appalachians and the Upper Midwest, and our objective was to develop empirical algorithms for the retrieval of forest phytochemical and functional traits across sites and years. As noted by Martin et al. (2008), there exists considerable potential to develop cross-site calibrations of imaging spectroscopy data for canopy chemistry, which our research confirms. However, variability in retrieved spectra – even within a single day – can provide challenges to application of generalized algorithms. We compare predictions of canopy chemical traits retrieved using artifact and atmospherically corrected (ACORN, TAFKAA, ATCOR) and uncorrected imagery to suggest a synthesis of the relative accuracies and biases associated with each technique.
Ash trees across North America are threatened by the invasive emerald ash borer (EAB), a bupresti... more Ash trees across North America are threatened by the invasive emerald ash borer (EAB), a buprestid beetle that continues to kill tens of millions of ash trees (Fraxinus spp) in 15 states and provinces with no signs of abatement. Invasion by EAB presents a challenge to resource managers ...
Background/Question/Methods Natural resource agencies are tasked with managing wildlife for a div... more Background/Question/Methods Natural resource agencies are tasked with managing wildlife for a diverse group of constituents with interests ranging from hunting to conservation. These agencies rely on population models that are limited by coarse spatiotemporal resolutions that can foster misunderstandings when patterns described at one scale are incompatible with decision-making at another. The challenge lies in developing an open and accessible method for collecting data on wildlife populations at a variety of scales to improve the information used in decision-support models, and to provide the public confidence that the information being used by decision-makers more accurately reflects conditions on the ground. As part of a unique partnership between NASA, the Wisconsin Department of Natural Resources, Zooniverse, and the University of Wisconsin, we are deploying over 2,000 camera traps throughout Wisconsin. These camera traps will be established and maintained by citizen scientist...
High nitrogen (N) concentrations in streams contribute to eutrophication and acidification of rec... more High nitrogen (N) concentrations in streams contribute to eutrophication and acidification of receiving waters, the loss of biological diversity, and present a public health concern (Vitousek et al., 1997). In temperate landscapes with mixed land cover, a particular concern is high N contributions in the form of nitrate (NO3-) derived from agricultural runoff and other sources. The ability to predict nitrate-N (NO3-N) levels of streams easily and accurately is needed to understand the consequences of increased nitrogen loading and to improve management. Recently, land-use/land-cover (LULC) classifications have been used in a variety of studies to predict stream water quality on broad scales (e.g., Stanley and Maxted 2008). These predictive models rely on LULC classifications derived from satellite imagery, combined with other environmental data to extend predictions over varying spatial and temporal scales. The majority of research using single image LULC techniques indicated a posi...
Recent and forthcoming developments in imaging spectroscopy will soon provide unprecedented oppor... more Recent and forthcoming developments in imaging spectroscopy will soon provide unprecedented opportunities for assessing ecosystem function. Research has shown that spectroscopic methods have the potential to characterize key plant functional traits such as foliar nitrogen content, specific leaf area (SLA), foliar lignin and cellulose and N15 concentrations. Moreover, it has been shown that variables that can be derived from imaging spectroscopy - foliar nitrogen and lignin/cellulose concentrations - characterize leaf lifespan, influence soil C:N ratios and in combination with disturbance events, may be strong controllers of nutrient cycling in forested landscapes. New methods and improved signal-to-noise of imaging spectroscopy instruments have enabled the assessment of variations in forest functional traits across space and time. However, standardized protocols for pre-processing spectroscopic imagery do not exist, in contrast to spectroscopic methods for chemometric analysis. Give...
Recent and forthcoming developments in imaging spectroscopy will soon provide unprecedented oppor... more Recent and forthcoming developments in imaging spectroscopy will soon provide unprecedented opportunities for assessing ecosystem function. Research has shown that spectroscopic methods have the potential to characterize key plant functional traits such as foliar nitrogen content (Majeke et al. 2008; Martin et al. 2008; Townsend et al. 2003), specific leaf area (SLA), foliar lignin and cellulose (Kokaly et al. 2009; Majeke et al. 2008) and δN15 concentrations (Elmore and Craine 2011; Kleinebecker et al. 2009) across landscapes. Moreover, it has been shown that variables that can be derived from imaging spectroscopy - foliar nitrogen and lignin/cellulose concentrations - characterize leaf lifespan (Santiago 2007; Shipley et al. 2006; Wright et al. 2005), influence soil C:N ratios (Ollinger et al. 2002) and in combination with disturbance events (Eshleman 2000; McNeil et al. 2007), may be strong controllers of nutrient cycling in forested landscapes (Aber et al. 1991; de Bello et al. ...
Urban forests provide important ecosystem services related to climate, nutrients, runoff and aest... more Urban forests provide important ecosystem services related to climate, nutrients, runoff and aesthetics. Assessment of variations in urban forest growth is critical to urban management and planning, as well as to identify responses to climate and other environmental changes. We estimated annual relative basal area increment by tree rings from 37 plots in Madison, Wisconsin and neighboring municipalities. We related relative basal area growth to variables of vegetation traits derived from remote sensing, including structure (aboveground biomass, diameter, height, basal area, crown width and crown length) from discrete-return airborne lidar, and biochemical variables (foliar nitrogen, carbon, lignin, cellulose, fiber and LMA), spectral indices (NDVI, NDWI, PRI, NDII etc.) and species composition from AVIRIS hyperspectral imagery. Variations in tree growth was mainly correlated with tree species composition (R2 = 0.29, RMSE = 0.004) with coniferous stands having a faster growth rate th...
Introduction Understanding the links between animal distribution and habitat plays a pivotal role... more Introduction Understanding the links between animal distribution and habitat plays a pivotal role in designing management for threatened species (Lecis and Noris, 2003). While an understanding of habitat preferences is important to improve ecological knowledge about a species, ...
This paper investigates people’s perceptions toward King Cobras in the tropical rainforests of th... more This paper investigates people’s perceptions toward King Cobras in the tropical rainforests of the Western Ghats ecoregion of southern India. We built logistic regression models to test if people’s perceptions (to kill/not to kill the snake) were influenced by factors such as the snake’s size and defensiveness, whether the snake was found near human habitation, the time of encounter and season. The model correctly classified over 80% of instances when people expressed an inclination to kill the snake. Results support our expectation that the snake’s defensiveness escalates the probability that the snake will be killed, but are contrary in that smaller snakes are more likely to be killed than larger ones, especially when encountered away from human habitation. Findings suggest a need for slight refocusing of King Cobra conservation outreach efforts towards smaller snakes, especially in regions where sizeable human habitations exist near fragmented King Cobra habitat.
Introduction Understanding the links between animal distribution and habitat plays a pivotal role... more Introduction Understanding the links between animal distribution and habitat plays a pivotal role in designing management for threatened species (Lecis and Noris, 2003). While an understanding of habitat preferences is important to improve ecological knowledge about a species, ...
Ecological applications : a publication of the Ecological Society of America, 2015
A major goal of remote sensing is the development of generalizable algorithms to repeatedly and a... more A major goal of remote sensing is the development of generalizable algorithms to repeatedly and accurately map ecosystem properties across space and time. Imaging spectroscopy has great potential to map vegetation traits that cannot be retrieved from broadband spectral data, but rarely have such methods been tested across broad regions. Here we illustrate a general approach for estimating key foliar chemical and morphological traits through space and time using NASA's Airborne Visible/Infrared Imaging Spectrometer (AVIRIS-Classic). We apply partial least squares regression (PLSR) to data from 237 field plots within 51 images acquired between 2008 and 2011. Using a series of 500 randomized 50/50 subsets of the original data, we generated spatially explicit maps of seven traits (leaf mass per area (M(area)), percentage nitrogen, carbon, fiber, lignin, and cellulose, and isotopic nitrogen concentration, δ15N) as well as pixel-wise uncertainties in their estimates based on error pro...
ABSTRACT Recent and forthcoming developments in imaging spectroscopy will soon provide unpreceden... more ABSTRACT Recent and forthcoming developments in imaging spectroscopy will soon provide unprecedented opportunities for assessing ecosystem function. Research has shown that spectroscopic methods have the potential to characterize key plant functional traits such as foliar nitrogen content (Majeke et al. 2008; Martin et al. 2008; Townsend et al. 2003), specific leaf area (SLA), foliar lignin and cellulose (Kokaly et al. 2009; Majeke et al. 2008) and δN15 concentrations (Elmore and Craine 2011; Kleinebecker et al. 2009) across landscapes. Moreover, it has been shown that variables that can be derived from imaging spectroscopy - foliar nitrogen and lignin/cellulose concentrations - characterize leaf lifespan (Santiago 2007; Shipley et al. 2006; Wright et al. 2005), influence soil C:N ratios (Ollinger et al. 2002) and in combination with disturbance events (Eshleman 2000; McNeil et al. 2007), may be strong controllers of nutrient cycling in forested landscapes (Aber et al. 1991; de Bello et al. 2010; Fortunel et al. 2009). New methods and improved signal-to-noise of imaging spectroscopy instruments (or hyperspectral imaging) have enabled the assessment of variations in forest functional traits across space and time (Asner and Vitousek 2005; Chambers et al. 2007; Martin and Aber 1997). However, standardized protocols for pre-processing airborne spectroscopic imagery do not exist in contrast to spectroscopic methods for chemometric analysis. Given the large volume of data forthcoming from the NEON AOP and future spaceborne missions such as ENMAP and HyspIRI, it is essential that assessments be performed to identify the consequences of pre-processing on consistent retrieval of forest traits across images and through time. Information retrieval from airborne spectroscopic imagery is especially difficult due to the multitude of artifacts in hyperspectral imagery. Whereas atmospheric effects can be partly corrected using established radiative transfer models, additional artifacts such as from along and across-track brightness gradients due to solar and sensor geometry, differential illumination effects due to complex terrain, and occlusions due to clouds and cloud shadows can persist. In combination, these effects may manifest themselves in the quality of spectra retrieved from radiative transfer models and eventually effect retrievals of key structural and biochemical traits of forest canopies. Here, we expand upon the methods and techniques (Figure 1) developed at the University of Wisconsin for dealing with many of these issues using a large database of spectroscopic imagery (~150 images) collected over three years (2008-2011) across different ecosystems in the US. Sites are located in the Rocky Mountains, the Appalachians and the Upper Midwest, and our objective was to develop empirical algorithms for the retrieval of forest phytochemical and functional traits across sites and years. As noted by Martin et al. (2008), there exists considerable potential to develop cross-site calibrations of imaging spectroscopy data for canopy chemistry, which our research confirms. However, variability in retrieved spectra – even within a single day – can provide challenges to application of generalized algorithms. We compare predictions of canopy chemical traits retrieved using artifact and atmospherically corrected (ACORN, TAFKAA, ATCOR) and uncorrected imagery to suggest a synthesis of the relative accuracies and biases associated with each technique.
Ash trees across North America are threatened by the invasive emerald ash borer (EAB), a bupresti... more Ash trees across North America are threatened by the invasive emerald ash borer (EAB), a buprestid beetle that continues to kill tens of millions of ash trees (Fraxinus spp) in 15 states and provinces with no signs of abatement. Invasion by EAB presents a challenge to resource managers ...
Background/Question/Methods Natural resource agencies are tasked with managing wildlife for a div... more Background/Question/Methods Natural resource agencies are tasked with managing wildlife for a diverse group of constituents with interests ranging from hunting to conservation. These agencies rely on population models that are limited by coarse spatiotemporal resolutions that can foster misunderstandings when patterns described at one scale are incompatible with decision-making at another. The challenge lies in developing an open and accessible method for collecting data on wildlife populations at a variety of scales to improve the information used in decision-support models, and to provide the public confidence that the information being used by decision-makers more accurately reflects conditions on the ground. As part of a unique partnership between NASA, the Wisconsin Department of Natural Resources, Zooniverse, and the University of Wisconsin, we are deploying over 2,000 camera traps throughout Wisconsin. These camera traps will be established and maintained by citizen scientist...
High nitrogen (N) concentrations in streams contribute to eutrophication and acidification of rec... more High nitrogen (N) concentrations in streams contribute to eutrophication and acidification of receiving waters, the loss of biological diversity, and present a public health concern (Vitousek et al., 1997). In temperate landscapes with mixed land cover, a particular concern is high N contributions in the form of nitrate (NO3-) derived from agricultural runoff and other sources. The ability to predict nitrate-N (NO3-N) levels of streams easily and accurately is needed to understand the consequences of increased nitrogen loading and to improve management. Recently, land-use/land-cover (LULC) classifications have been used in a variety of studies to predict stream water quality on broad scales (e.g., Stanley and Maxted 2008). These predictive models rely on LULC classifications derived from satellite imagery, combined with other environmental data to extend predictions over varying spatial and temporal scales. The majority of research using single image LULC techniques indicated a posi...
Recent and forthcoming developments in imaging spectroscopy will soon provide unprecedented oppor... more Recent and forthcoming developments in imaging spectroscopy will soon provide unprecedented opportunities for assessing ecosystem function. Research has shown that spectroscopic methods have the potential to characterize key plant functional traits such as foliar nitrogen content, specific leaf area (SLA), foliar lignin and cellulose and N15 concentrations. Moreover, it has been shown that variables that can be derived from imaging spectroscopy - foliar nitrogen and lignin/cellulose concentrations - characterize leaf lifespan, influence soil C:N ratios and in combination with disturbance events, may be strong controllers of nutrient cycling in forested landscapes. New methods and improved signal-to-noise of imaging spectroscopy instruments have enabled the assessment of variations in forest functional traits across space and time. However, standardized protocols for pre-processing spectroscopic imagery do not exist, in contrast to spectroscopic methods for chemometric analysis. Give...
Recent and forthcoming developments in imaging spectroscopy will soon provide unprecedented oppor... more Recent and forthcoming developments in imaging spectroscopy will soon provide unprecedented opportunities for assessing ecosystem function. Research has shown that spectroscopic methods have the potential to characterize key plant functional traits such as foliar nitrogen content (Majeke et al. 2008; Martin et al. 2008; Townsend et al. 2003), specific leaf area (SLA), foliar lignin and cellulose (Kokaly et al. 2009; Majeke et al. 2008) and δN15 concentrations (Elmore and Craine 2011; Kleinebecker et al. 2009) across landscapes. Moreover, it has been shown that variables that can be derived from imaging spectroscopy - foliar nitrogen and lignin/cellulose concentrations - characterize leaf lifespan (Santiago 2007; Shipley et al. 2006; Wright et al. 2005), influence soil C:N ratios (Ollinger et al. 2002) and in combination with disturbance events (Eshleman 2000; McNeil et al. 2007), may be strong controllers of nutrient cycling in forested landscapes (Aber et al. 1991; de Bello et al. ...
Urban forests provide important ecosystem services related to climate, nutrients, runoff and aest... more Urban forests provide important ecosystem services related to climate, nutrients, runoff and aesthetics. Assessment of variations in urban forest growth is critical to urban management and planning, as well as to identify responses to climate and other environmental changes. We estimated annual relative basal area increment by tree rings from 37 plots in Madison, Wisconsin and neighboring municipalities. We related relative basal area growth to variables of vegetation traits derived from remote sensing, including structure (aboveground biomass, diameter, height, basal area, crown width and crown length) from discrete-return airborne lidar, and biochemical variables (foliar nitrogen, carbon, lignin, cellulose, fiber and LMA), spectral indices (NDVI, NDWI, PRI, NDII etc.) and species composition from AVIRIS hyperspectral imagery. Variations in tree growth was mainly correlated with tree species composition (R2 = 0.29, RMSE = 0.004) with coniferous stands having a faster growth rate th...
Uploads
Papers by Aditya Singh