Friday, August 5

ESSE 21 Skills Workshops

1:30 – 3:00 pm

Using STELLA for Earth System Models – Part I

John Snow, University of Oklahoma and Arthur Few, Rice University

Using Web-based GIS and other online resources to promote Web-based Inquiry

Alec Bodzin, Lehigh University

HyperInteractive Education

Owen E. Thompson, University of Maryland

Using the Evaluation Toolkit

David Reider, Education Design; Sabra Lee, Lesley University

Hands on with GPS and GIS (12)

Rudi Gens, Bill Witte, Anupma Prakash, University of Alaska

3:30 – 5:00 pm

Using STELLA for Earth System Models – Part II

John Snow, University of Oklahoma and Arthur Few, Rice University

Introduction to Raster GIS – IDRISI

Miriam Cope, Cal Poly Pomona, Bob Ford, Loma Linda University

Build and Fly CricketSat (10)

Neal Brown, Anupma Prakash, University of Alaska

Translating Science

Gina Maranto, University of Miami

Spatial Analysis – Locating the Right Technology

Jean ThiÈbaux, USRA

Using STELLA for Earth System Models (Part I and Part II)

John Snow, University of Oklahoma and Arthur Few, Rice University

The STELLA modeling language offers a powerful, highly flexible, graphically-oriented software tool for both students and instructors seeking to learn more about the Earth System through numerical modeling. Models of various components of the Earth System are easily constructed and then used to explore, in highly simplified form, "how the world works". In this two-part workshop presentation, we demonstrate a small suite of STELLA-based Earth System models and recount both personal and student experiences with the STELLA modeling language in the classroom. The first half of the presentation illustrates the layout and operation of a number of different models, with emphasis on the organization of their organization as interactive "learning environments". The second portion of the presentation addresses the many "lessons learned" during the development and from student use of the models. A CD with all notes and copies of the models will be provided which will allow the participants to continue to learn about STELLA after the meeting.

Helpful background material to read before the workshop:

http://www.hps-inc.com/Community/STArticles/SystemsThinking.aspx

http://www.hps-inc.com/resources/Articles/ST%204%20Key%20Questions.pdf

http://www.hps-inc.com/resources/Articles/STELLA%20IST%20-%20Chapter%201.pdf

http://www.hps-inc.com/resources/Articles/SDSTletsjustgetonwithit.pdf

Using Web-based GIS and other online resources to promote Web-based Inquiry

Alec Bodzin, Lehigh University

This session demonstrates the use of Web-based interactive GIS map coverages and Science-Technology-Society role-playing debate simulations to promote scientific inquiry for interdisciplinary watershed studies.

Inquiry refers to activities through which students develop knowledge and understanding of scientific ideas and how scientists study the natural world. In this workshop, participants will gain a theoretical and practical understanding about how to take advantage of Web-based GIS and other online instructional materials and approaches to promote inquiry learning with students.

This session will introduce a manual and instrument designed to identify and classify Web-based Inquiry for Learning Science activities (WBIs) along a continuum from learner-directed to materials-directed for each of the five essential features of inquiry as described in Inquiry and the National Science Education Standards. 

Participants will have hands-on experience working with a series of LEO (Lehigh Earth Observatory) GIS coverages we have developed to address questions about our watershed. In addition, an overview of our Web-based Science-Technology-Society (STS) role-playing debate simulations that are indexed in DLESE will be presented.

Workshop Outline:

1. Overview of Web-based inquiry for learning science

2. GIS for Watershed Investigations

http://www.leo.lehigh.edu/envirosci/watershed/gis/investigations.html

GIS and Interactive Mapping Web Links

http://www.leo.lehigh.edu/envirosci/watershed/gis/gislinks.html

3. STS role-playing debate simulations

Sprawl in the Lehigh River Watershed

http://www.leo.lehigh.edu/envirosci/enviroissue/sprawl/

Abandoned Mine Drainage in Pennsylvania 

http://www.leo.lehigh.edu/envirosci/enviroissue/amd/index.html

Stockertown Sinkhole Dilemma

http://www.leo.lehigh.edu/envirosci/enviroissue/sinkholes/

HyperInteractive Education

Owen E. Thompson, University of Maryland

The use of advanced 3D and 4D computer visualization technology is clearly established as a key tool for research scientists trying to understand complex things. During the 20th Century, adapting this technology to educational classrooms required the same tens of thousands of dollars per seat as it cost for research scientists. Things have changed ...

This workshop session explores the possibility of developing interactive 3D/4D educational web experiences in earth system science. This subject will be explored using interactive 3D/4D visualization in a common web web browser on common desktops and laptops Hands-on examples will include:

1. Interactive Virtual Experimental Laboratories

* Students Discovering Geostationary Satellite Orbits for Their First Time

* Investigating the "Coriolis Effect" in a virtual reality lab

2. Hyperinteraction with Global, Regional and Local Data and Models

* Utilizing Existing Web-based Images and Movies

* Utilizing Deep, Multi-Variate, Multi-Dimensional Databases across the web

3. Using Advanced Visualization to Compare Two ES Scientific Scenarios

* Weather on Rotating and Non-Rotating Earths

* Two Mathematical Predictions of One Tomorrow

Evaluation Toolkit

David Reider, Education Design ; Sabra Lee, Lesley University

Evaluation is often at the core of understanding how our teaching and learning works and how effective it is. As an activity, it helps us improve our courses and teaching. However, we often do not conduct thoughtful measurement of our own teaching and learning. The Evaluation Mini-Workshop will explore methods to evaluate ESSE-21 courses and activities. We will discuss outcome-based evaluations, the goals of evaluation, and describe commonly used methods. Activities include constructing surveys and rubrics, measurement strategies that are very useful and commonly found in college science courses.

Hands on with GPS and GIS

Rudi Gens, Bill Witte, Anupma Prakash – University of Alaska Fairbanks

A few of the skills that the GPS-GIS workshop will address include map reading, orientation, mobile field data capture techniques, and data integration. The skill set is critical in acquiring new data and using it for integrated analysis for promoting the concepts of Earth System Science. In this workshop the participants will

* learn the basics of global positioning systems (GPS) and geographic information systems (GIS)

* use hand held GPS units and take measurements

* plot GPS data on aerial photos / topo maps while in the field (on palmtops or upper end GPS units)

* download GPS data as Arc GIS shape files and integrate them in GIS in the computer lab

* compare the hardware/software available, limitations, prices, etc. to determine what is best for their needs

(Maximum number of participants: 12)

Introduction to Raster GIS – IDRISI and Image Classification

Miriam Cope, Cal Poly Pomona, Bob Ford, Loma Linda University

The goal of our mini-workshop is to introduce non-users of image processing software to IDRISI Kilimanjaro from Clark Labs, with hands on training in image classification. We have found it to be the most affordable and yet most robust in analytic capacity and it is particularly good for those training students or partners who plan to work internationally. Since our ESSE courses focus strongly on ≥spatial analysis≤—particularly remote sensing--we have adopted it along with STELLA, SPSS, SAS, and ArcGIS as critical tools in our teaching toolkit.

In our session we will first do a ≥show-and-tell≤ on IDRISI then lead participants in one or two quick intro ≥hands-on≤ exercises, and there will be time for Q & A. We will create training sites and develop signatures for the hard classifiers found in Idrisi: parallelpiped, minimum distance, maximum likelihood and fisher (linear discriminant analysis). Users will filter out isolated or unknown pixels and generate an accuracy assessment for each of the classified images. Participants will also receive ≥timed-out CD≤ copies of the software for evaluation back home. We will be glad to share what we have learned about ≥what works well≤ as well as ≥what doesnπt≤. Please join us! The workshop will be limited by the lab seats available - there will be a signup sheet at registration.

Below is a description by Clark Labs of IDRISI Kilimanjaro = http://www.clarklabs.org/IdrisiSoftware.asp?cat=2>http://www.clarklabs.org/IdrisiSoftware.asp?cat=2

Build and Fly CricketSat

Neal Brown, Anupma Prakash - University of Alaska Fairbanks

A few of the skills the Cricket Sat mini-workshop will addresses include experimental design, electronic assemblage, schematics reading, data analysis, and meteorologicl interpretation. The experiment is a great way of demonstrating the concepts of Earth System Science as it integrates the disciplines of electronics, physics, meteorology, earth science, and ecology. In this workshop the participants will

* construct a miniature weather balloon consisting of a temperature sensor, RF transmitter and a receiver

* mount their sensor on to a ballon and launch it from the parking lot on UAF lower campus

* retrieve the temperature data from their receivers (The receiver will either be attached to a computer for later analysis or a speaker whose output chirping frequency varies proportionately with temperature).

* discuss fidelity and prediction of temperature data

(Maximum number of participants: 10)

Translating Science

Gina Maranto, University of Miami

This workshop will focus on methods and goals for communicating science to students, scientists in general, and the media. It will look at the ways in which scientists at different career stages--from students to post-docs and beyond--can improve their abilities to write and speak about their work to a range of audiences. Ir will provide a sequence of assignments meant to introduce undergraduates to writing about science for a general audience, as well as outlining a curriculum for improving graduate students' communication skills. A hands-on computer session will demonstrate how to use Science Citation Index to help students gain a more comprehensive understanding of "hot" topics, researchers, and institutions. In addition, we will discuss means for increasing the chances that scientific work is reported accurately in the media.

Spatial Data Analysis ä So you need to analyze data from a large-scale (probably heterogeneous) observing array?: a consulting session approach to locating the right technology ä

Jean ThiÈbaux, USRA

The foundation of earth system science is comprised of the basic sciences: physics, chemistry, biology, and "societal evolution and its footprint on the Earth". It is a substantial foundation; and it becomes even more complex when the impact of each element on the others is added in. Nonetheless, understanding the Earth System in all of these ways is the necessary challenge to those most concerned with the nurture and husbandry of our planet.

That is the substantive or "natural components" aspect of "Earth System Science". The complimentary aspect is the quantitative, or measurement science aspect. In addition to numerical modeling, the latter includes quantitative diagnostics and multivariate estimation of the natural components and of their interactions, namely, "spatial analysis". Without the tools of spatial analysis, we can describe how the elements of the system work and interact, but we cannot use data inputs from observations of the natural components to make quantitative inferences. Specifically we cannot make data-based estimates of states of the Earth System other than at the points of the observations, nor make statements about relationships of its components with assignable measures of certainty. The latter are the purview of spatial analysis. And they are central to the challenge of understanding the workings and nurturing the health of our planet.

This lecture/discussion session will provide a guide through the labyrinth of spatial analysis techniques. Participants are invited to bring their own spatial analysis projects and problems to the session. Please send a one-page summary to jeanthiebaux@verizon.net , including the following:

ÿ The research objective of the project.

ÿ What questions do you wish the project to answer?

ÿ Description of your data, actual or anticipated, including instrument

and error characteristics. (Please do not include the data themselves.).

ÿ What methodology do you think you might use to achieve your objective?

If you have a clear idea of how you will analyze the data, include this in your description along with specific questions you would like addressed during the workshop.

The proposed format will use participantsπ works-in-progress as the bases for examining the utilities of different methods of analyzing data from spatial observing arrays, as these methods serve differing scientific objectives and data types. It will also provide guidance in locating software and reference literature.

Methodologies for the analysis of data from spatially-distributed observing arrays: candidates for discussion in the workshop ≥How do we choose, use and understand the analysis of data from large-scale observing arrays?≤

Each of the following techniques, available for discussion in the workshop, was designed for the study of spatial phenomena, with data for which the geographic distribution of observations is an important ingredient.

Empirical Orthogonal Function or Principal Component Analysis

≥EOF≤ or ≥PC≤ analysis is well suited for data reduction, although it does not offer a mechanistic insight into the observed variables. Each of a time-sequence of spatial data arrays can be represented as a sum of orthogonal functions with time-specific coefficients, where the terms of the sum are in decreasing order of the percent of total variance explained. Often the first few (1, 2, or 3) modes account for a significant fraction of the total variance in the data; and the sum of these terms, alone, can provide a computationally-efficient approximation for the full field.

Spline Fitting

Fitting piecewise polynomials to a set of observations over a defined interval or geographic region with specified boundary values for the observed variables, produces easy-to-use, continuous representations for the variables. These can be ≥tuned≤ to any desired level of agreement with the values of the observation data, through the use of ≥splines under tension≤. Choice of the tensioning parameters is usually related to what is known or assumed about the observation-error components and thus the reliability of the observations. We note that the existence and accuracy of boundary values are crucial for the construction and accuracy of any spline-representation within the region encompassed by the input data; and that it is not useful beyond these bounds (because of its polynomial basis.)

Neural Network Analysis

This methodology is a predictive or forecasting tool of great value in generating an algorithm that can produce ≥output fields≤ for future observed ≥inputs≤. This can be developed using an archive of matched input and output values. The (neural network) analogy for the methodology is the mechanism for transfers of messages within the brain: Information from several sources is ≥parallel-processed≤ through a array of intermediary nodes whose outputs act in-turn and in-concert to produce a final set of outputs. This can be thought of as series-nested regressions, with black-box interventions. The matched input and output data of the archive provides for ≥training≤ the neural network, in the sense that the internal regression coefficients are adjusted in a feed-back loop, for overall agreement of network output with the actual output of the training set. The black-box element prevents its use as a diagnostic tool; however it has great value in processing enormous data arrays, such as satellite-retrieval data.

Kriging Analysis

Kriging is a linear interpolation of spatially distributed data for use with one-time samples, such as geologic core samples. This methodology is based on the assumption that covariance between values of the observed variables at different locations is entirely a function of the distance between locations and can be accurately (reciprocally) represented with a fitted variogram. The method is applied to single samples at individual locations within a region and works on their differences with a simple, estimated mean or trend surface, to produce a least-squares estimator for all (sampled and unsampled) points of the encompassing region.

Optimal Statistical Objective Analysis or Generalized Kalman Filtering

≥OSOA≤ or ≥GKF≤ was originally developed for routine updating of large-scale models of geophysical variables that evolve in time/space dimensions, using real-time, or the nearest-real-time observations available for each update. In this case, the variables estimated are the differences between the current model values and what would be observed if there were observations everywhere throughout the region of interest. The estimated differences, usually called ≥estimated increments≤ or ≥model errors≤, are applied as corrections to the model in its re-initialization. This statistically based algorithm is a multidimensional, generally multivariate, regression algorithm for which the accuracy of the representations of spatial covariances among analysis increments for all model variables, is key to the accuracies of the model corrections produced by it. To the degree that the choice of covariance functions, fit the true covariance characteristics of the corresponding increment fields, the analysis is the minimum-variance, least-squares estimator among all possible linear estimators for the increments. An additional benefit is the diagnostic tool for examination and understanding of model errors provided by an accurate representation for their spatial covariance structure.