The Laboratory of Ecological Spectroscopy (LECOSPEC) was founded in January 2024 by Dr. Peter R. Nelson to provide commercially available spectroscopy imaging to Maine and beyond. LECOSPEC offers imaging/scanning, analysis and training using reflectanct spectroscopy for natural resource applications in agriculture, forestry and aquatic environments. Contact Principal and Founder Peter at [email protected] for more information.
Imaging Equipment: Through a partnership with the University of Maine, LECOSPEC offers Visible and Near-infrared (VNIR) airborne imaging using a Headwall Micro-A series imaging spectrometer on a heavy-lift UAS. This instrument has 326 spectral channels covering 400-1000nm within 10cm or smaller pixels. For ground-based applications such as leaves, rocks, water, and soils, LECOSPC offers scanning solutions covering Visible, Near and Shortwave infrared (VSWIR, 350-2500nm, 2151 spectral channels) using a Spectral Evolution PSR+
Analysis: LECOSPEC can provide end-to-end imaging to analysis or provide guidance on analysis and options in between. LECOSPEC can provide analysis or support with R, Python, ENVI and QGIS. Lecospec offers open=source R code workflows (lecospectR) for users to develop their own analysis, which we also use in training. Python deep learning (lecospy) workflows are in development for licensing and tuning to specific applications.
Imaging Equipment: Through a partnership with the University of Maine, LECOSPEC offers Visible and Near-infrared (VNIR) airborne imaging using a Headwall Micro-A series imaging spectrometer on a heavy-lift UAS. This instrument has 326 spectral channels covering 400-1000nm within 10cm or smaller pixels. For ground-based applications such as leaves, rocks, water, and soils, LECOSPC offers scanning solutions covering Visible, Near and Shortwave infrared (VSWIR, 350-2500nm, 2151 spectral channels) using a Spectral Evolution PSR+
Analysis: LECOSPEC can provide end-to-end imaging to analysis or provide guidance on analysis and options in between. LECOSPEC can provide analysis or support with R, Python, ENVI and QGIS. Lecospec offers open=source R code workflows (lecospectR) for users to develop their own analysis, which we also use in training. Python deep learning (lecospy) workflows are in development for licensing and tuning to specific applications.
Our Work
LECOSPEC is actively engaged in multiple spectral imaging projects centered on the enhancement of current remote sensing techniques through the inclusion of UAV-based spectroscopy, enabling the enhanced study of patterns of community assembly and vegetation health measurement across spatial scales. Through the use of UAVs, airborne imagery, hyperspectral remote sensing, and ground-truthing by spectroradiometer, the lab works to generate high resolution maps to answer the questions: What is it? Where is it? How much is there?
CURRENT Projects
Wymans blueberry nutrient and irrigation trials
Wyman's Blueberries is the world's largest wild blueberry producer and operates thousands of acres of managed bluberry fields in the US and Canada. LECOSPEC has flownover their experimental trials multiple times each growing season for the last several years. The goal is to help understand what blueberry traits can be estimated with visible and near infrared (VNIR) imagery and how that might inform management, such risk management and yield estimation.
NASA ABoVE: Biome shift implications for resource management.

The NASA Arctic-Boreal Vulnerability Experiment (ABoVE) is a multi-year field campaign to gain a better understanding of the sensitive ecosystems affected by the rapid effects of climate change.
The Spectroscopy lab collected over 1000 reflectance scans (using Spectral Evolution PSR+, 350-2500 nm) from over 100 vascular and non-vascular plants across boreal and Arctic Alaska between 2017-2019. Data products will be provided as rasters in GeoTIFF format coded as 8 bit unsigned integers (data values 0–100 and separate value for NODATA). A QC GeoTIFF will also be provided for each data raster that includes codes for unprocessed areas (e.g. water, snow, void) and confidence metrics for cover values.
The current work for the project centers on processing the gathered hyperspectral reflectance data in R, and orthorectifying them to generate accurate, high resolution, plant functional type maps based on target vegetation’s unique reflectance signatures.
View the lab's Alaska spectral library data repository here.
The Spectroscopy lab collected over 1000 reflectance scans (using Spectral Evolution PSR+, 350-2500 nm) from over 100 vascular and non-vascular plants across boreal and Arctic Alaska between 2017-2019. Data products will be provided as rasters in GeoTIFF format coded as 8 bit unsigned integers (data values 0–100 and separate value for NODATA). A QC GeoTIFF will also be provided for each data raster that includes codes for unprocessed areas (e.g. water, snow, void) and confidence metrics for cover values.
The current work for the project centers on processing the gathered hyperspectral reflectance data in R, and orthorectifying them to generate accurate, high resolution, plant functional type maps based on target vegetation’s unique reflectance signatures.
View the lab's Alaska spectral library data repository here.
PAST Projects
Field Museum Polar Studies: Fine-scale, species-level hyperspectral mapping of Arctic plants and lichens.

The study of Alaskan tundra focuses on hyperspectral landscape data collected at very high spatial resolution (< 5 cm) from drones and ground level surveys for the purpose of identifying and mapping individual vascular plants and lichens. At each plot, a drone-mounted hyperspectral camera (Headwall Micro A-series VNIR imaging spectrometer) was used to capture 4 cm-pixel resolution data (400-1000 nm). Hyperspectral readings of all species were taken with a field-portable spectroradiometer (Spectral Evolution PSR+, 350-2500 nm).
For the species targeted for collaborator Richard Ree 's genetic study, 100 individuals were sampled from each species across all sites, with spectroradiometric readings taken in the field and after drying for all individuals.
Mixed-Tuned Matched Filter (MTMF) will be used to connect spectral libraries to aerial imagery to attempt to map each species present in the plot. The results will increase our understanding of how ecological communities are assembled at different spatial scales, and reveal how hyperspectral reflectance may be influenced by both genotype and environment. Our study will aid the development of rapid, low-cost ecological monitoring protocols in an environment particularly susceptible to climate change.
View the lab's Alaska spectral library data repository here.
For the species targeted for collaborator Richard Ree 's genetic study, 100 individuals were sampled from each species across all sites, with spectroradiometric readings taken in the field and after drying for all individuals.
Mixed-Tuned Matched Filter (MTMF) will be used to connect spectral libraries to aerial imagery to attempt to map each species present in the plot. The results will increase our understanding of how ecological communities are assembled at different spatial scales, and reveal how hyperspectral reflectance may be influenced by both genotype and environment. Our study will aid the development of rapid, low-cost ecological monitoring protocols in an environment particularly susceptible to climate change.
View the lab's Alaska spectral library data repository here.
MSGC RID: Enhanced assessment of Maine forests through VNIR imaging spectroscopy.

The observational network for monitoring Maine tree species distribution, forest disturbance history, and biomass stocks, has employed multi-scale remote sensing from repeat aerial image collections, existing data and new acquisitions coincident with previous NASA G-LiHT sensor.
The Spectroscopy lab builds on this framework by refining the spatial scale of measurements taken to match those of individual trees, using point-intercept sampling of plant presence along transect as method of measuring forest understory composition and abundance. Each plant recorded at each point along the transect were assigned a categorical group (dominant canopy, sub-canopy, shrub layer, and forb layer). These categories are rough vertical strata groups that will allow grouping of the vertical strata so each plant can be attributed to a specific cloud of LIDAR points.
Our remote sensing campaigns employed the DJI Matrice600 UAV with the Headwall Micro A-series VNIR imaging spectrometer and Applanix APX-15 GPS/GNSS system operated between 50-100 m AGL. The Micro A-series high capacity spectral range enables for detection of physically small or spectrally narrow features (eg. branches with a pathogen).
The Spectroscopy lab builds on this framework by refining the spatial scale of measurements taken to match those of individual trees, using point-intercept sampling of plant presence along transect as method of measuring forest understory composition and abundance. Each plant recorded at each point along the transect were assigned a categorical group (dominant canopy, sub-canopy, shrub layer, and forb layer). These categories are rough vertical strata groups that will allow grouping of the vertical strata so each plant can be attributed to a specific cloud of LIDAR points.
Our remote sensing campaigns employed the DJI Matrice600 UAV with the Headwall Micro A-series VNIR imaging spectrometer and Applanix APX-15 GPS/GNSS system operated between 50-100 m AGL. The Micro A-series high capacity spectral range enables for detection of physically small or spectrally narrow features (eg. branches with a pathogen).
MEIF SCI: Hyperspectral Imaging for Mapping and Health Assessment of Maine Forest and Agriculture.

The ongoing project expands ground spectroradiometric sampling and hyperspectral sensing throughout Maine to a series of target sites connected to key economically important crops, forest resources and associated problems (eg. insects or disease) or positive signals (exceptional growth). For these targets, we continue to build image and spectral libraries for use in mapping target surfaces (eg. healthy or diseased crop species) for 500 samples throughout Maine.
The spectral library provides spectral end-members of each target surface so that the quantity or presence of the target chemistry or plant/soil type can be estimated for each pixel in the hyperspectral image.
Long-term, we plan to grow a flexible platform of UAVs or airplane-based hyperspectral image pipelines, including ground-based spectral data and chemical analyses of samples to produce a map of client-generated targets.
The spectral library provides spectral end-members of each target surface so that the quantity or presence of the target chemistry or plant/soil type can be estimated for each pixel in the hyperspectral image.
Long-term, we plan to grow a flexible platform of UAVs or airplane-based hyperspectral image pipelines, including ground-based spectral data and chemical analyses of samples to produce a map of client-generated targets.