Hyperspectral Imaging is a new and innovative method of mapping the spread ofimportant invasive plant species. It is a non-intrusive method that will provide a basic information baseline for the islands. The unique feature of this system is its adaptability to the needs of the ecosystem manager. It isinvaluable as a warning system and has features that can be expanded upon at alater date if necessary.
Hyperspectral Imaging is a relatively new technology that has important environmental applications. It would be an advantageous and applicable system in the Galapagos due to a number of important factors. First the unintrusive nature of the measurements taken will be key in an ecosystem where disturbance of often fragmented and fragile habitats can result in giving unintended advantages to invasive species that need to be eliminated. Secondly, measurements from a remote sensor can provide in depth information on distribution of species, something that is often labor intensive and based on mainly qualitative information when done on an island-wide scale. Finally, qualitative information on distribution of invasive species can be used by park managers in a number of useful ways. Park managers can use this information to determine the overall health of the Galapagos ecosystem or to facilitate removal of these invasive plants. The data can be used as a baseline for ecosystem variations and as an indicator in the studies of human impacts on the Galapagos. Hyperspectral analysis recreates the complex processes that occur in the human eye, sensing the unique spectral signatures of materials on earth. Spectral signatures result from the reflection of electromagnetic radiation from the sun. Hyperspectral monitoring uses the spectral reflectance (or ratio of reflected energy to incident energy as a function of wavelength) to analyze the makeup of different materials. The downward trends in the spectral curves show the wavelength ranges where the material absorbs the incident energy. Using these bands it is possible to distinguish between different materials. Chlorophyll in plants absorbs visible light but absorbs red and blue light more effectively than green light. Thus reflectance varies sharply across the boundary between red and infrared wavelengths due to the complex cellular processes going on within the leaves. Species type, canopy state and plant stress can all affect these readings.
Hyperspectral analysis was chosen over other methods of analysis because of its easily quantifiable nature. Also analysis by simple photography would not be efficient or effective in this situation. Photography would require a species easily identifiable by sight or color. Also pictures taken would need to be analyzed by experts in order to determine the density and nature of invasive species found in each area. This would require undue amounts of time. Often when using simple photography with species that are similar in nature valuable data cannot be obtained due to the amount of plants that occur within the sample area. Since the species that warrant monitoring on the Galapagos often form dense stands and occur in areas where the native vegetation has been disturbed, photography presents researchers with countless obstacles to definitive results. Analysis of two-dimensional photographs does not give clear indications of the amounts of invasive plants in one area. Manual data sampling was another option considered in this analysis. Manual data sampling, other than disturbing natural habitats, also has several disadvantages. Manual sampling can be done using a number of methods. The first method is to take entire species counts, which in addition to being costly and time consuming, is limited by human capabilities. Entire population sampling relies on the expertise of the individuals doing the surveys. If the participants in the survey are not trained properly large amounts of experimental error can occur. Training will add on to the amount of resources and time needed to set up an island wide sensor network. These surveys will cover areas on different islands and often take place in areas that are not easily accessed by researchers. This type of survey requires a large amount of people to perform and although it yields interesting information is not necessary for the purposes of a large nature reserve. The monitoring system in the Galapagos is intended to serve as an early warning system and to give park management a better idea of the health of the ecosystem as a whole. Survey sampling, another technique, was also ruled out because of its intrusion into the natural landscape as well as its dependence upon trained experts to perform frequent field studies in a variety of locations. Survey sampling is to remain a valuable supplement to hyperspectral analysis. This will allow park managers to receive valuable in the field data but will allow these surveys to be performed on a more infrequent basis making it feasible to do in depth analysis of invasive plants on each of the islands without the intrusive costs of frequent sampling. Survey sampling could be performed in the winter months when many plants go into more dormant states and would reinforce data received from remote sensors. Multispectral analysis was also considered but because of the larger band gap of this data the results are often less accurate. Accuracy is extremely important in a dynamic ecosystem like the Galapagos. Another advantage of hyperspectral analysis is its adaptability to other systems. The monitoring system of the Galapagos was designed to be expandable and hyperspectral monitoring has a wide variety of uses. It can be used to monitor turbidity in the waters of the Galapagos, pollution from point and non-point sources and even to monitor biodiversity in the many ecosystems of the Galapagos. It is perhaps the ability to monitor biodiversity that is the most exciting application of this system. As the spectral library of the Galapagos is built more plants can be analyzed. This will allow scientists to view the diversity of a patch of land in the Galapagos something nearly impossible to do in the past. As this technology improves the error occurring in the measurements of the spectral signatures of materials will decrease. One of the research initiatives funded will be the spectral identification of all the known plants on the Galapagos and the creation of a computer program that can identify invasive stands as they appear and send out warnings to park rangers who will deal with these problems. Also this research initiative will create computer programs that recognize new spectral signatures, which will be investigated and perhaps will find new species or species that are thought to be extinct. Over time this data will provide the most interesting information because scientists will be able to see evolutionary interactions as the spectral signatures of evolving plants begin to change. Another exciting use of this system that will be researched will be the protection of animal species that depend on the habitats monitored by these sensors.
The hyperspectral equipment chosen for the Galapagos is new and advanced equipment due to the highly variable nature of the Galapagos. For hyperspectral data the bandpass has to be fairly narrow (between one and five nanometers). Spectral analysis of plants must be done in the field and must be done in the visible to near infrared range to be applicable to our testing. The spectroradiometers that are chosen to perform this important function should include wavelength ranges from about 400 to 1000 nm. The amount of interference should be minimized as much as possible but because this is research done in the field some interference is expected. The equipment should be versatile and adaptable as well as durable. The spectrometer also must be integratable with satellite and GIS data needed to analyze the results of our studies. The computer program chosen to analyze the data was ENVI. This system was chosen mainly for its adaptability to many different types of systems. ENVI is compatible with GIS data as well as with many types of sensors and directly imports the binary data it needs from these devices. ENVI allows the user to input both scanned air data and satellite data into its system, calculates vegetation biomass data, calculates plant stress data, and overlays both GIS data or GPS data allowing the user to automatically extract channels, pits and other drainage features. Another important advantage to information analysis with ENVI is the fact that it has been used worldwide and is the only commercial software that allows users to read the HDF data format. It also allows users to manipulate large amounts of data, which surely would be the case in the Galapagos.
Currently satellites exist that give hyperspectral analysis (for instance EO-1 launched into space in December 2001 by NASA). More satellites of this nature are likely to be launched into space in the next few years. Data from these satellites would be pivotal to our research program and funding could be provided by Orgalla. It is also possible that for research rights in the Galapagos NASA and other space organizations would be willing to contribute to our research. If funding was plentiful enough a satellite could be built specifically for our purposes.
