III. Educational Activities

In the review of our IDS effort, it was pointed out that we needed to expand our educational contribution beyond the walls of the University. We recognized this obligation and we have sought aggressively to meet these responsibilities.

Our expanded sense of educational responsibility does not stem only from the NASA review process, but also from our recognition that there is a failing in the American educational system.

For the last three decades, commissions, committees, and task forces have called for a new approach to science and mathematics education in American schools, citing continuing challenges to US economic strength and general well-being from increasing international competition. While many students do receive an excellent education, it is clear that many do not, as evidenced by poor performance relative to peers in other countries, inadequate preparation for post-secondary education, insufficient numbers pursuing post-secondary education, continued under-representation of women and minorities in science careers, and low levels of public science literacy. A recent comparative study of pre-college science and mathematics education in the United States, conducted for the National Science Foundation (NSF), identifies a lack of focus in school curricula and characterizes student knowledge as "a mile wide and an inch deep." The reason for this "...is because the (educational) system is deeply and fundamentally flawed," concludes W.L. Schmidt of Michigan State University and author for the NSF report. Reforms in our present methods of science and mathematics education are clearly needed.

The American Association for the Advancement of Science (AAAS) and the National Research Council (NRC) of the National Academy of Sciences have sponsored two science reform initiatives (Project 2061 and National Science Education Standards). Although these initiatives were generated independently, both focus on science education reform as a means of achieving the common goal of scientific literacy for all Americans. Further, the AAAS has established a series of benchmarks to be reached as a means of improving scientific literacy. A common goal of both Project 2061 and the National Science Education Standards is the recommendation that pre-college students develop a clear understanding of the nature of science as one step in the process of achieving scientific literacy. Inquiry is identified in both proposals as an essential component of science, and both identify a "hands-on" approach to the learning of science as an effective means of introducing students to the essential components of science.

One approach to introducing students to hands-on science (as opposed to textbook-based study) is to involve students with on-going scientific investigations as part of a student/scientist partnership. Several examples of such partnerships have been recognized, such as: GLOBE, GREEN, Forest Watch, SPaRCE, and SPAN. A Student/Scientist Partnership Conference, held in October, 1996 in Washington, DC, highlighted the pluses and minuses associated with such active participation by students in scientific research efforts and called attention to the need for data quality assurance as a key issue in the development of successful partnerships. It is obvious that partnerships between students and scientists must provide benefits for science as well as for students. Two of the most successful examples of Student/Scientist Partnerships cited at the Conference were Forest Watch and GLOBE. Forest Watch which was developed, in part, through our IDS project and GLOBE benefited enormously from significant input of our IDS team and other UNH faculty. The lessons learned in both of these effective educational programs are now being applied to the development of similar activities based on research efforts in the UNH/EOS project.

During 1996, several additional outreach educational activities were developed as part of our IDS effort.¹⁵ These activities fall into three general categories: a) developing remote sensing, image processing, and modeling tools for use in the classroom; b) continuing to connect and coordinate GLOBE and Forest Watch-types of student-generated datasets with on-going EOS and related global-scale research projects; and c) assisting in the development of outreach education components of graduate student research projects.

• Classroom Tools For Introducing Students To Remote Sensing, Image Processing, And Modeling.

MultiSpec Tutorial Development. The current version of the MultiSpec tutorial, developed for Forest Watch and GLOBE, contains new material designed to train students in various methods of creating land cover maps of their own areas. The methods include detailed descriptions of how to conduct manual interpretations, as well as unsupervised classifications and supervised classifications using the free MultiSpec software, and are designed around the use of a modified UNESCO Classification system that is international in scope. This tutorial is written for the high school freshman/sophomore level, but could be used in training personnel at all levels of education.

PnET Model Development for High School Classroom Use. A recent development leading to a more user-friendly version of PnET is the production of a Visual Basic version. In addition to being easier to use, this new format allows for easy conversion to a Macintosh operating system. PnET-Day and PnET-II are likely candidates for use in the classroom, although the PnET-CN version, which required past land-use history may also be appropriate for use in a classroom. Students will be able to use land cover maps based on the 512 X 512 pixel sub-scenes of Landsat TM data of their own area (produced using the MultiSpec unsupervised classification capability) as input data for parameterizing PnET, along with max/min temperatures, precipitation, soil moisture, and litterfall collected by the students.

We would like to point out that PnET is a significant research tool that contributes directly to expanding our understanding of terrestrial systems in the context of landuse and environmental change. Making this available in a high school situation is an important contribution.

Narrow-band Imaging Video Camera. As part of the continuing study of forest damage in the Czech Republic, it has become clear that broad band, Landsat-style datasets are not adequate for detection of the initial stages of forest damage. Recent work at UNH (Lambert, et al., 1995) has demonstrated that such broad-band data are only able to detect three categories of forest damage (low or healthy, moderate, and heavy damage).

This demonstrates a significant shortcoming if the objective is to be able to respond in time to correct the problem: a detection of early stages of damage is essential if corrective measures are to be taken. Hyper-spectral remote sensing methods offer this early (i.e. pre-visual) detection capability.

Field spectrometry offers early detection capabilities, but is too limited in terms of field-of-view (only a few branches can be scanned at a time - assessing a forest stand is not practical). Airborne or spaceborne spectrometry offer hope as a future option, but at present is not practical as a broad-scale capability. Both costs and computing power needs are present barriers to using such a hyper-spectral approach. As a low-cost alternative, UNH and the Stennis Space Center are coordinating the development of a video-based narrow-band imaging camera (NBIC).

The narrow-band imaging camera (NBIC) allows one to use narrow-band interference filters (+/- 5nm) with a standard security video camera (costing approximately $250.00) to monitor spectral fine-feature parameters in the 400-800nm spectral region. Subtle changes in the red edge position (related to minor changes in chlorophyll concentrations) can be monitored using the NBIC approach by comparing video images acquired for vegetation using 670, 700, and 720nm filters. Pre-visual detection of initial stages of damage by air pollution, insect infestation, and herbicide treatment have been documented. In the case of the Forest Watch white pine study, the NBIC has been used to demonstrate to students the value of monitoring the infrared to determine levels of damage in their branch samples.

If this work is successful, it will contribute to defining the requirements for hyper-spectral data in the AM-2 era. Thus in this area of work, the focus is on contributing to pre-college education challenges and in so doing we contribute to a wide array of activities including expanding our knowledge of terrestrial systems at regional scales and defining remote sensing requirements for the year 2003 (when many of these high school student will be graduating from college and hopefully some going to graduate school with a focus upon global change).

• Making Connections Between Student-Generated Datasets And On-Going EOS And Related Research Projects.

Two significant connections were established in 1996 between sources of student datasets such as land cover type maps produced in GLOBE and Forest Watch and our on-going large-scale IDS research efforts. The first is with the EOS/MODIS Land Cover modeling project at Boston University (Alan Strahler, Curtis Woodcock and Doug Muchony). The second is with the IGBP START (System for Analysis, Research, and Training).

MODIS Land Cover Modeling Project. At present the project is using AVHRR and transformed Landsat TM data (scaled to 1.1 km pixels) to simulate MODIS datasets. No reliable source of finer-resolution (i.e. TM 30m scale) land cover maps are currently available for much of the US or the rest of the world. Student-generated Land Cover Maps, produced from 512 X 512 pixel TM sub-scenes using either a manual visual interpretation approach or a computer-based unsupervised classification image (via MultiSpec) represent a unique source of such datasets.

During the past year, extensive time and effort has been put on revising the MultiSpec tutorial training manual (discussed above) developed for Forest Watch teachers and students. This revision has included development of detailed tutorial materials for the use of MultiSpec with both Macintosh and PCs, and for the production of both supervised and unsupervised classification images. In addition, a stand-alone tutorial for the production of a manual visual interpretation of Landsat TM hardcopy images (both true color and false color IR 512 X 512 sub-scenes centered on Forest Watch schools) was developed. These products were also made available to the GLOBE national program for inclusion in the latest version of the GLOBE Teacher's Guide.

Using these tutorials, and biometry measurements developed for Forest Watch, students are able to produced very detailed Land Cover maps. The Land Cover classification used is the Modified UNESCO Classification (MUC), developed because of its international application (as compared with the Anderson Classification System developed for use in the US).

As part of the UNH participation in the GLOBE Program, an error matrix approach to assessing the accuracy of the Land Cover maps has been developed. Thus, not only are students able to produce high-spatial resolution Land Cover maps for their local area (15km X 15km study areas centered on their school), but they are also able to determine the degree of accuracy of these products. An equal number of calibration sites and validation sites are to be used in production and accuracy assessment activities.

Land Cover maps from GLOBE schools in the 45 participating GLOBE countries will be made available to the MODIS Land Cover Modeling team for validation of the MODIS-simulation land cover products. It is estimated that GLOBE schools represent approximately 3,500 separate 15Km X 15km study sites (located on all continents except Antarctica) for which high-resolution land cover maps can be produced. Such a global network of land cover sites, each produced using the same types of data (TM images), using the same methods (manual or unsupervised classifications), assessed for accuracy using the same methods (error matrix), and using the same classification scheme (MUC), is unique and unavailable from any other source. These student datasets constitute an invaluable source of accurate, high-resolution, land cover information of importance to the global research community.

IGBP/START Project. Successful implementation of the IGBP scientific agenda requires that research - measurements, analysis, synthesis, modeling, and prediction - be carried out in all regions of the world, using similar approaches and methods. To meet this need, the concept of a global system of regional networks of institutions pursuing common IGBP objectives has been proposed. In this concept, each region comprises several cooperating nations trained in the appropriate methods. The approach described above (land cover classifications, each generated using the same type of data, using the same methods, assessed for accuracy using an error matrix approach, and using the same classification scheme) is an ideal goal. Participants in START will be trained using educational materials (curricula, software, video support material, etc.) developed as part of the EOS/IDS program.