Ecological Spectral Information System (ESIS): Integration of Spectral Data with Measurements of Vegetation Functional Traits 
(Collaborative Research, NASA)

A. Overview

Spectroscopy spanning visible through thermal infrared wavelengths is critical for improving scientific understanding of vegetation ecology. Spectroscopy enables measurement of vegetation and soil biophysical and biochemical properties, assessing plant functional types and biodiversity, and quantifying biosphere-atmosphere gas and energy exchange (Roberts et al. 2004, Kokaly et al. 2009, Ustin et al. 2009, Ben Dor et al. 2009, Ustin & Gamon 2010, Ollinger 2011, Serbin et al. 2012). Spectral data measured in the lab or the field can be compared to image data collected from aircraft and satellites using a variety of techniques including empirical methods (Martin & Aber 1997) and radiative transfer modeling (Combal et al. 2003, Asner & Martin 2008, Toomey et al. 2009). The ability to easily obtain reference spectral data is vital to many investigations. A well-curated spectral repository will allow researchers to quickly identify and use relevant spectral reference data and share information and data products resulting from NASA missions and instruments. The Ecological Spectral Information System (ESIS) that we propose implements a repository consisting of a multi-dimensional relational database, holding extensive collections of spectra, metadata and associated measurements. ESIS will function like a spectral library providing unlimited public access, and a structure that encourages user contributions, and assures that the holdings will be scientifically rigorous and traceable.

The spectral repository will incorporate a spectral library to enable rapid and effective access to scientifically robust data that will enable researchers to investigate and find answers to questions vital to the NASA Earth Sciences community. These questions include: 1) Which chemical and physical properties (e.g., chlorophyll, leaf mass per area, nutrients) can be retrieved consistently across species, ecosystems and measurement platforms? 2) What are the relative contributions to spectral variability of chemistry/physiology, structure, and phenology? 3) How do we meaningfully define “optical types” (Ustin & Gamon 2010) given the continuous variation of optical properties within the world’s major plant taxa? 4) Are plant species/functional types spectrally distinct, and at what spatial, temporal and spectral scales? 5) Which portions of the spectrum are most stable across plant species and which are most variable, and 6) How do spectra vary diurnally, with weather change, and seasonally, with spatial scale and measurement error. The objective of our proposal is to provide a foundation to a) enhance usability and accessibility of NASA data products to researchers, natural resource and agricultural managers, and related decision makers at all levels and b) extend the usability and fidelity of airborne and space-borne spectral imagery.

Well-designed spectral databases are essential tools of scientific inquiry in ecological sciences for several reasons. They provide:

B. Project Synopsis
Our goal is to develop a working prototype of an effective spectral database system, with active community engagement. Through these key actions, ESIS will yield a tangible outcome: a functioning prototype system for warehousing a rich, validated set of data, metadata standards, and analytical tools to facilitate synthesis studies and help identify critical gaps in our understanding of the spectral variation within ecosystems. We expect the ESIS approach to provide a systematic basis for evaluating the causes of spectral variability, including error. ESIS will provide a quantitative framework for remote sensing of vegetation that is an essential foundation for future studies involving imaging spectroscopy.

ESIS will be developed as a resource for the ecosystem spectroscopy community. It will be the foundation for new initiatives in collaboration and cooperation within the Earth Sciences community and lower access barriers by those outside the community. Starting from existing data sets contributed by the PIs (and NEON), ESIS will house data in a standardized, reliable, and traceable format using open source tools.  ESIS has long-term parallel and overlapping objectives, and an intended long lifespan as a repository for spectral and associated data:
A. Integrate community-provided spectral data into a scientifically robust and accessible ESIS repository and identify critical gaps in current spectroscopic knowledge. Any one set of lab, field, or imaging spectrometer data cover a limited range of functional types, geographic locations, and phenological conditions. By allowing ready review and comparison of spectra, ESIS will provide the capacity to assess the coverage of existing spectral data, determine where coverage is inadequate, and allow integration of new spectra to fill data gaps. ESIS will create data ingest standards and methods to assure the scientific validity, traceability, and description of data.
B. Establish spectral data collection standards and best practices. Methodologies for collecting lab and field spectral data will be documented and evaluated. Approaches that facilitate improved inter-comparison will be published, following efforts by Rivard et al. 2008. We will actively engage (via virtual conferencing) ongoing work in the SpecNet community (Gamon et al. 2006;, including ACEAS/TERN (in Australia,, EuroSpec (, and NEON. In this way, we will engage community participants and facilitate an emerging community consensus to identify and implement such standards.
C. Establish and use metadata standards and best practices for spectra in the VNIR-SWIR and TIR regions consistent with ISO 19115 and others by the remote sensing, ecology, modeling and climate change communities. Again, we will engage the international community in this component of the project and take advantage of lessons learned from existing international synthetic activities. NEON is committed to ensuring that their metadata and best practices standards are consistent with those arising from community consensus, and thus, will take an active role in this task. The standards we propose will be flexible enough to incorporate definitions for spectral measurements from non-vegetative materials (soils, minerals, chemicals, reflectance standards, etc.).
D. Develop access to the spectral holdings through web-based and open access application programming interfaces (APIs). Through the use of APIs, different groups can effectively exchange data, develop and share tools, and engage in synthesis-oriented discovery. We will start with the HyspIRI Ecosystem Spectral Library (, building a queryable database (back-end) hosted from JPL computers and front-end access delivered via web interfaces. Additional programming and development of the database, its interface and tools will occur at the University PI institutions, drawing from consensus developed in regular web conferences. The system will be hosted on JPL computers from inception, with remote programmers accessing the system through VPN access granted via Affiliate Agreements with JPL. The full database system will transition to JPL management in Year 3 and the entire database will be mirrored on the NEON servers and accessible through the NEON Data Portal. Through SpecNet, we will engage the primary teams behind the SPECCHIO database in Switzerland and Australia to ensure data structures that promote integration of spectral datasets (see letters from Schaepman, Held, and Chisholm).
E. Develop hierarchical, linked, interconnected descriptions among data holdings with associated vegetation properties. Relational databases will facilitate the interpretation of spectra and scaling of biological, chemical and physical measurements. We will focus on a database design that facilitates discovery through integration of spectral data with ancillary measurements of leaf/canopy chemistry, structure and metabolism.
F. Create, collect, catalogue, and make available easily-accessible open-source tools for users to query, visualize, extract, and analyze spectral data and ancillary measurements in a readily interpretable format. Building upon the development of APIs, this activity includes tools for consistent spectral processing (such as the R package “FieldSpec”), spectral convolution, band simulation for current and future sensors, spectral averaging, filtering, time series visualizations, leaf-to-canopy modeling, and summary statistics. These tools will combine web-interface and stand-alone routines, open source scripts or R packages that can query the database remotely without downloading the database itself. These activities are central to developing community engagement and encouraging contributions and use of the data. The toolset library will be available from a centralized repository.
G. Evaluate approaches for data inter-comparison. Under the proposed framework, we will conduct a series of analyses involving legacy datasets (e.g., from the PIs) and new data emerging from the HyspIRI, NASA TE, SpecNet, FLUXNET (Baldocchi et al. 2001) and NEON communities to address important synthesis-oriented research questions. Our goal is to harness existing tools within the community, make them open-source and readily accessible, thereby facilitating wider usage and improvements that lead to development of new tools.
H. Provide a framework for partitioning variability of spectra into various sources (e.g., natural causes, user error, instrument error, etc.).  By providing a means to characterize sources of spectral error and variability, ESIS will offer a much-needed framework to answer a number of critical quantitative questions that are currently hard to address.
I.  Establish source citation policies that encourage the contribution of data sets and use of data holdings within the broader community. Those contributing data will be assured that use of their data in scientific papers, reports, and proposals will be accompanied by an appropriate citation of the providers, their institutions, and, if appropriate, their sponsors.

In accomplishing these objectives, ESIS will provide a stronger foundation for tackling a range of ecological and remote sensing research questions.