Volume 7, No. 1, May 2001
Editors: Teresa Alpuim and Bronwyn Harch
In This Issue:
1. A Message from the President
2. TIES News
2.1. New members3. Environmetrics and Related Conferences.
2.2. Member's and Regional News.
2.3. TIES 2001 Annual General Meeting.
3.1. Forthcoming TIES Conferences.4. Environmetrics Forum
3.2. Other Forthcoming Conferences.
3.3. Reports on Related Conferences.
5. Research Projects and Programmes
6. Recently Published Books
7. Book Reviews
8. TIES Board of Directors
1. A Message from the President,
Sylvia R. Esterby
The banner above announces that Teresa Alpuim and Bronwyn Harch are the new editors of the Newsletter. They will serve as joint editors for the remainder of the two year period ending August 31, 2002. I look forward to reading the news that members send in response to the invitations from Teresa and Bronwyn to contribute to the Newsletter.
From the excellent web page and Tony Olsen’s report here, the TIES 2001 conference promises to have a broad range of good technical sessions and opportunities for informal discussions. There is still time to pass the conference information along to any colleagues who have not yet seen it, including students who may wish to enter the Student Paper Competition.
The details of the 2002 TIES conference are also being worked on. The linkage to the International Environmental Modelling and Software Society (iEMSs) conference is consistent with the Board’s objective of establishing communication with other societies. Although this is a linkage with a new society, it is collaboration with a long-time member and former board member of TIES, Tony Jakeman. We hope that the proximity of place and date makes attendance at both conferences feasible.
Looking back to the previous conference, the proceedings volume from the TIES/SPRUCE 2000 conference will soon be available. It is being edited by Clive Anderson, Vic Barnett, Philip Chatwin and Abdel El-Shaarawi and published by Springer.
TIES continues to sponsor sessions and conferences. Two this summer are the session on Environmental Toxicology that I have organized at the joint conference of the Canadian Statistical Society (SSC), Western North American Region of the International Biometric Society (WNAR) and Institute for Mathematical Statistics (IMS) in June, and the sponsorship of the ISI satellite conference, International Conference on Statistical Challenges in Environmental Health Problems in Fukuoka City, Japan at the end of August. The session was organized at the invitation of the Biostatistics Section of the SSC and is jointly sponsored with WNAR. TIES members, Abdel El-Shaarawi and Walter Piegorsch are members of the organizational committees of the environmental health problems conference and proceedings are to be published in Environmetrics.
Also noteworthy is the near completion of the Encyclopedia of Environmetrics, edited by Abdel El-Shaarawi and Walter Piegorsch. Many members of TIES have contributed to this ambitious undertaking. Putting this together with the expansion of Environmetrics from 6 to 8 issues per year makes the contributions of TIES and its members to the environmental literature very substantial.
Apart from listing news, this message provides an opportunity to raise issues for later discussion. The theme of TIES 2001, Environmetrics for Decision Making, is a starting point for one such issue. Statistical departments and societies have been giving considerable thought to the promotion of statistics and statisticians in recent years. TIES and the field of environmetrics is cross-disciplinary, so some of the issues being considered with respect to promoting statistics and statisticians have already been addressed. However, we may need to promote environmetrics in view of the declining attention being paid to environmental degradation and sustainability. Also we would wish to promote, in collaboration with other scientists, the use of scientific knowledge in the decisions that affect the environment. Sessions at the upcoming conference, including the President’s Invited Lecture to be given by Dr. Agnes Herzberg, will provide ideas for discussion. There may be other activities that the Society could pursue to serve the environmetrics community.
Sylvia Esterby (firstname.lastname@example.org)
2. TIES News
2.1. New Members, Francis Philbert
Welcome to the 9 new members who have joined since October 24, 2000. Conferences, the web page and promotion of TIES by current members continue to be the major means by which individuals are learning about the Society.
Balabdoui, Fadoua USA
Costanzo, Dr. G.D. Italy
Damian, Doris USA
Gargoum, Dr. Ali S. United Arab Emirates
Kawamura, Prof. Kazutomo Japan
Kwiatkowski, Roy E. Canada
O'Sullivan, Maree Australia
Page, Dr. Grier P. USA
Steel, Ashley E. USA
2.2. Member's and Regional News
Promotion for Allan Adolphson, Australia
Allan Adolphson has been promoted to Science & Industry Manager for ‘Environmental Measurement and Assessment’ in the Commonwealth Scientific and Industrial Research Organisation (CSIRO). He has responsibility for the Biostatistics, Remote Sensing and Environmetrics groups within the Division of Mathematical & Information Sciences. More information about the division can be found at the Web site: www.cmis.csiro.au Allan is based in Sydney and his email address is Allan.Adolphson@cmis.csiro.au We wish him well in his new role.
2.3. TIES 2001 Annual General Meeting
The president of The International Environmetrics Society, Sylvia Esterby, announces that the Annual General Meeting will be at the TIES 2001 conference in Portland, Oregon, USA, after the technical sessions on Tuesday, August 14.
3. Environmetrics Conferences
3.1. Forthcoming TIES Conferences
TIES 2001, Portland, Anthony Olsen
TIES 2001 Conference will be held August 13-17, 2001, in Portland, Oregon, USA. Detailed information about the conference is available on the conference web site: http://osu.orst.edu/dept/statistics/ties2001/index.htm
Important dates concerning the conference are as follows. Abstract submissions due April 15, 2001. Abstracts submitted after April 15 but before July 15 will be considered but are not guaranteed to be accepted, depending on availability of conference space for sessions. Early registration closes May 15, 2001 to receive early registration discount.
The STUDENT PAPER COMPETITION committee will accept student papers until July 15, 2001. This competition includes a cash award of US$500 for the winner. Papers will be judged during the conference and winner announced at the end of the conference. Please encourage students to submit papers following the instructions on the conference web site or by email to TIES.email@example.com.
Dr. Agnes M. Herzberg, Queen's University, Canada, will be the TIES President's Lecturer. Dr. Herzberg has organized a series of conferences entitled "Conference on Statistics, Science and Public Policy", with various themes; the latest (April, 2001) was Science and Responsibility. She has been active in the Statistical Society of Canada, with one relevant initiative being the organization of invited sessions with non-statisticians as speakers to further interaction between statistics and the other sciences. She has also been active in other societies, notably the ISI.
Prof. Paul Switzer, Stanford University, has been selected by the TIES Board of Directors to be the J.S. Hunter Lecturer at the conference. Prof. Switzer has a distinguished career in environmental statistics and spatial modeling. He is an excellent person to represent the tradition established by J. Stuart Hunter.
The Scientific Program committee is finalizing the conference program of plenary speakers and invited technical sessions. Plenary speakers include: G. P. Patil, Pennsylvania State University, speaking on Multiscale Raster Map Analysis for environmental and ecological assessment at landscape level; Anders Grimvall, Linköping University, speaking on A statistical framework for meteorological and hydrological normalization of time series environmental data; Ron McRoberts, US Forest Service, speaking on The annual inventory and monitoring program of the United States; Corrado Lagazio, University of Udine, speaking on Age-period-cohort models and disease mapping.
Invited sessions currently include sessions on Chemometrics; Landscape econometrics and environmetrics using remote sensing data; USFS Forest Inventory; Modeling nonstationary spatial covariance; Normalization of air pollution data; Space-time modeling of disease and exposure data; Environmental Survey Design and Analysis; hydrology mathematical modeling; and Environmentricians as frontline researchers.
Invited speakers who have committed to attend the conference include Allan Adolphson, CSIRO; Anders Grimvall, Linköping University; Byron Gajewski, St. Cloud State; Christian Lajaunie, Centre de Geostatistique; Corrado Lagazio, University of Udine; Dankmar Boehning, Free University Berlin; David Fox, CSIRO; David Higdon, Duke University; Don Stevens, Dynamac, Inc; Doris Damian, University of Washington; Glen Johnson, University of Utah; Greg Biging, UC Berkeley, Greg Ream, US Forest Service; Jay Breidt, Colorado State University, Marina Vannucci, Texas A&M University; Montserrat Fuentes, North Carolina State University; Paul Sampson, University of Washington; Peter Guttorp, National Research Center on Statistics and the Environment; Phillip Kott, USDA NASS; Renato Assunção, Federal University of Minais Gerais & FIOCRUZ; Ron McRoberts, US Forest Service; and Wayne Myers, Pennsylvania State University.
Two short courses will be given during the conference, on Wednesday, August 15. Brian Manly, of Western EcoSystems Technology, Inc, will present a one-day course on Computer Intensive Statistics: Application to Environmental Studies. Steve Millard, Insightful Corp (formerly MathSoft), will present a one-day course on Environmental Statistics using S-Plus.
Further information is available on the conference web site or by email to TIES.firstname.lastname@example.org. The conference organizer is Tony Olsen, US Environmental Protection Agency, NHEERL Western Ecology Division.
The 2002 TIES Conference will be held in Genoa, Italy, from 18 to 22 June and will be organized by Vicenzo Dovì and Daniela Cocchi, respectively former and present TIES regional directors for Europe, and by Sylvia Esterby. It will be linked to the International Environmental Modelling and Software Society (iEMSs) meeting that will take place from June 24-27 in Lugano. For more informations, contact the Conference organizers V. Dovì, email@example.com, Daniela Cocchi, firstname.lastname@example.org, or Sylvia Esterby (email@example.com).
3.2. Other Forthcoming Conferences
The Inaugural Biennial meeting of the International Environmental Modelling and Software Society, iEMSs 2002, on Integrated Assessment and Decision Support, will be held from 24 to 27 June 2002, in the University of Lugano, Lugano, Switzerland.
The broad aim of the congress will be to assess the status of Integrated Modelling and Assessment (IMA) with respect to the integration of disciplines, models, treatment of issues, scales of consideration, and stakeholder concerns, and to propose ways in which to improve the science and practice of IMA. The topic will be explored by means of theoretical papers and case studies.
The specific topics covered in this conference will be:
Integrated assessment frameworks and case studies, in particular, integrated water assessment and integrated urban planning assessment; Multiobjective decision support; Innovative software engineering for model integration and re-use; Advances in model approaches, uncertainty modelling, and testing; Scale issues in modelling; Participatory integrated modelling, stakeholder issues in modelling and decision support; Environmental applications of models and software; Applications of agent-based modelling and simulation to environmental systems.
The deadline for abstracts is 30 September 2001 and for full papers is 28 February 2002. Further details on the conference can be found on theiEMSs2002 web site, http://www.iemss.org/iemss2002 or by contacting Dr. Andrea Rizzoli, Istituto Dalle Molle di Studi sull'Intelligenza Artificiale (IDSIA), Manno, Switzerland, +41-(0)91-6108664, firstname.lastname@example.org.
School of Mathematics and Applied Statistics, University of Wollongong will be hosting an International Conference on Statistics, Combinatorics and Related Areas and the Eighth International Conference of the Forum for Interdisciplinary Mathematics during December 19-21,2001. Conference website for more information: www.uow.edu.au/informatics/maths/statconference
The 16th Conference on Probability and Statistics in the Atmospheric Sciences, sponsored by the American Meteorological Society and organized by the AMS Committee on Probability and Statistics in the Atmospheric Sciences, will be held 13-18 January 2002 as part of the 82nd AMS Annual Meeting in Orlando, Florida, USA. A preliminary program, registration, hotel, and general information will be posted on the Web at http://www.ametsoc.org/AMS in early September 2001.
Papers are solicited on all aspects of probability and statistics, including but not limited to, observations and data assimilation, ensemble prediction, weather forecasting, climate variations and forecasting, environmental applications, data display and analysis,
spatial and space-time statistics, time series, forecast evaluation, weather derivatives, and statistical aspects of global change, hydrology, atmospheric chemistry, and air pollution.
Please submit your abstract electronically via the Web by 3 July 2001 (refer to the AMS Web page at www.ametsoc.org/AMS for instructions). For more information, contact the Program Chairperson:
Robert E. Livezey (tel: 301-763-8000, ext. 7527; e-mail: Bob.Livezey@noaa.gov)
International Biometrics Conference (IBC) July 21-26, 2002. The international Biometrics Conference 2002 will take place at the University of Freiburg, Germany, July 21-26, 2002. The Annual Meeting of the German Region will be integrated. More information can be found by visiting the IBC2002 Web site: www.ibc2002.uni-freiburg.de
3.3. Reports on Related Conferences
The Eighth International Meeting on Statistical Climatology (8IMSC) was held in Luneburg, Germany during 12-16 March 2001. One of the main purposes of this series of meetings is to promote communication between the atmospheric and statistical science communities. The conference focused on a few themes, including "Decision Making and Detection of Anthropogenic Climate Change." Several TIES members participated in the conference.
The program and abstracts are available at: http://w3g.gkss.de/8imsc.html
The next meeting (9IMSC) is tentatively scheduled for Cape Town, South Africa in 2004. For updates on this forthcoming conference, see: http://www.cccma.bc.ec.gc.ca/imsc/
4. Environmetrics Forum
Open Problems in Stochastic Modelling of Wind
E. Kestens & J.L. Teugels
University Centre for Statistics (UCS)
Catholic University of Leuven
W. de Croylaan 52 B, B-3001 Leuven
We would love to convince researchers to tackle some of the many interesting problems in stochastic modelling of wind. There exists a reasonable amount of publications concerning wind and statistics as shown in Kestens and Teugels (2000). Still, a lot of unsolved problems remain. Efforts are needed to improve existing or develop new statistical techniques, relevant for the understanding of wind in its many facets. This paper attempts to summarise some of the existing problems hoping to encourage research in stochastic modelling of wind.
Quality of Wind Data. One of the major causes for the lack of interest in wind studies is the low quality of available wind data. Not that good data do not exist. But by the time data get to the statistician, too much manipulation has ruined them for advanced statistical work. The first assignment therefore is to improve the quality of data. We illustrate this point by a number of examples.
A first improvement would be an indication of error bands on the measurements. For example, measurements at sea are usually very crude and hence, have wider variability than measurements on land. The quality of the data also depends on the geographical location: weather stations at different locations measure with different accuracy. Even measurements at the same place can have error bands varying over time. Analysts may discover a decreasing trend in wind data due to the growth of trees in the neighbourhood. Sometimes anemometers are even replaced without any further mention to the data analyst. For a similar problem gaining interest over the last years: how can one investigate changes in climate if the measurement errors are not given?
A definite improvement would result from getting tick by tick data. While most weather stations have such data, statisticians only get some derived, averaged set. Years ago such an attitude could be defended by lack of storage place but current computer capacities have overcome this kind of problem. As said above, statisticians should have access to the original data and decide themselves when to take maxima or averages. Another important reason for the use of tick by tick data is that proper data collection is crucial for specific types of wind. For example, an average rainstorm lasts about one hour. How can one rely on daily averages to say anything reasonable about such storms? A related quality problem is caused by the rounding off of data before they are given to the researcher. This leads to unwanted and untraceable inaccuracies. This argument doesn’t only concern wind speeds since for wind directions wind rose recordings limit the information on 16, 12 or even 8 sectors and so a lot of valuable information is lost.
But there are more defects. Three-dimensional modelling of the wind velocity vector becomes impossible without measurements at different heights. Wind velocity measurements at a single location but at different heights are rare. Practitioners use empirical loading factors to determine wind speeds at different heights. For a specific example, think of the effect of height when modelling breeze, where wind blowing from the ocean to the land is fighting against wind blowing in the reverse direction. Another defect of wind data stems from the fact that calms are censored from the measurements since anemometers need a minimal lower wind speed to get activated. However, exactly these wind speeds are crucial in pollution related wind studies.
Measuring and Modelling Problems. Even with high quality equipment, techniques to measure and analyse the wind velocity are not always straightforward. For example, think of how to protect people and constructions against severe wind events like hurricanes or tornadoes? First of all, data are simply not available. Indeed, tornado sizes as well as anemometers are extremely small objects and hence, the probability that a tornado passes by an anemometer is close to zero. But even if one would be lucky (?) in that a tornado hits an anemometer, the latter is blown away immediately and no recording is available. While their news value is more important, hurricane data are even scarcer than tornado data. This fact should not prevent scientists from modelling hurricanes, a thoroughly interesting problem even without data. But even then, many stochastic aspects prevail. There is the physical reality that hurricanes can only develop if the water temperature over a large but specific expanse of ocean is at least 26.5 Celsius degrees.
Still there remains a lot of stochasticity that needs to be modelled. For example, current prediction of the number of hurricanes in a certain region within the next year is based on a Poisson assumption. Is this valid? A first test could be to split up hurricane rates in disjoint time blocks and to test them for independence. Gray et al (1992) developed such a prediction method but without giving standard deviations on the predictions. Another interesting assumption is that the difference between the atmospheric pressure at the centre and at the periphery of the hurricane is modelled by a log-normal or a Weibull distribution. But also the path and strength of a hurricane are interesting stochastic quantities needed to be modelled. Another important question lies in the modelling of the extreme value distribution of wind speeds of a hurricane. Based on the fact that hurricanes have finite maximum intensity, as proved by Emanuel (1988), one might expect that also the maximal wind speed should have an upper limit; however a validation of this statement does not yet exist. For tornadoes, similar problems arise and it would be very useful if the path and the extent of a tornado could be modelled stochastically.
From Physics to Stochastics. It is somewhat frustrating that the physical laws that underline wind phenomena are well understood and fit into Navier-Stokes type equations from hydrodynamics. Each specific type of wind such as breeze, tornado, hurricane, rainstorm, gust is governed by such a physical context. However, the transformation of these well-understood physical laws into a stochastic calculus set-up would be a highly welcome contribution. This simple remark frames wind studies into the area of stochastic partial differential equations.
Once a basic stochastic model can be designed for the wind velocities, the resulting quantities of interest should emerge through stochastic calculus. For example, the individual modelling of wind speed and of wind direction would result; speed and direction should be described by appropriate stochastic equations. Even more important would be the solution of simultaneously modelling wind speed and wind direction as a bivariate stochastic process. From this joint model the effect of wind direction on extreme wind speeds should emerge as a feature of the conditional distribution of the wind speed, given the direction. Alternatively, correlations between wind speeds and directions could be investigated.
We can safely conclude from the above complaints and considerations, that there is ample scope for high quality research and similar statistical analysis. Still a lot of efforts are necessary to solve even part of the many problems mentioned above. Tackling them would improve existing wind modelling in many respects. We include a selected bibliography to guide the reader to the vast literature.
Breckling J. (1989). The Analysis of Directional Time Series: Applications to Wind Speed and Direction. Lecture Notes in Statistics, 61, Springer-Verlag, Berlin. Bucher C, Wall FJ. (1992). Stochastic Response of Bridges in Turbulent Wind. J. Wind Eng. Ind. Aerodyn, Vol 41-44:1347-1358. Chandler RE. (1997). A spectral method for estimating parameters in rainfall models. Bernoulli, Vol 3:301-322.
The matter of particulates and health
Peter Guttorp, University of Washington
Richard L. Smith, University of North Carolina
On April 11, the US Environmental Protection Agency issued their latest review of the scientific literature on particulate matter air pollution. The document, called Air Quality Criteria for Particulate Matter, is the first step in the process of setting air quality standards in the United States. Once the criteria document is made final, EPA staff produce a document with recommendations to the Administrator of the EPA, who then announces any changes in the air quality standards that may be called for on the basis of the two documents. While the staff document reflects current thinking within the agency, the criteria document is supposed to be an objective and comprehensive review of current scientific thinking in the area.
The document, which is open to public review until July 12, 2001, consists of two volumes (chapters 1-5 and 6-9, respectively), which can be downloaded from the EPA web site at http://www.epa.gov/ncea/partmatt.htm where instructions for public comments can be found.
After the brief general introduction in Chapter 1, Chapters 2 and 3 provide background information on physical and chemical properties of PM and related compounds; sources and emissions; atmospheric transport; transformation and fate of PM; methods for the collection and measurement of PM; and ambient air concentrations; Chapter 4 describes PM environmental effects on vegetation and ecosystems, impacts on man-made materials and visibility, and relationships to global climate change processes; and Chapter 5 contains factors affecting exposure of the general population. Chapters 6 through 8 evaluate information concerning the health effects of PM. Chapter 6 discusses epidemiological studies. Chapter 7 discusses dosimeter of inhaled particles in the respiratory tract. Chapter 8 assesses information on the toxicology of specific types of PM constituents, including laboratory animal studies and controlled human exposure studies. Chapter 9 integrates key information on exposure, dosimetry, and critical health risk issues derived from studies reviewed in the prior chapters. The Executive Summary is to be incorporated at the end of Volume II.
The basic conclusion of the PM criteria document is that particulate matter air pollution has significant effect on mortality and morbidity, particularly on elderly individuals. The document, however, fails to take into account much of the recent statistical literature questioning the validity of the methodology used in many of the studies upon which the conclusion is based. For example, a special issue of Environmetrics (vol. 11, issue 6, 2000) on statistical analysis of PM data contained three papers showing the sensitivity of these analyses to model choice, in particular lag structure and air pollution covariates. In a study reanalyzing Seattle PM2.5 (particles with average size below 2.5 ?m) data and asthma hospital admissions, Lumley and Sheppard (2000) find that the bias due to model selection (choosing between only seven models), the log relative risk estimated from the data is about twice the mean bias in simulated control analyses, and the estimate falls at the 90th percentile of the bias distribution in the control analyses. Thus in studying the weak associations between PM and health outcomes, it is important to take into account biases that normally could safely be ignored. Smith et al. (2000) reanalyzed regression models for air pollution and daily mortality in Birmingham, Alabama, demonstrating the sensitivity of the analysis to the definition of an exposure measure for lagged PM10 values, and that there is little evidence of an effect at low levels of air pollution. Clyde (2000) analyzed the same data using Bayesian model averaging, and obtained results that were quite sensitive to prior assumptions. The original study of these data by Schwartz (1993) obtained a relative risk of 1.11. The posterior probability of a risk that large or larger in Clyde’s analysis is between 0.007 and 0.0042, depending on the choice of prior.
Many studies have considered other pollutants besides PM, but too often the emphasis has been on how the other pollutants affect the PM coefficient (the co-pollutant problem) rather than starting out from the assumption that there are several pollutants that could possibly explain the mortality effects and we should treat them on equal footing when it comes to identifying potentially causal relationships. Where this has been done, the results are mixed. For example, the Health Effects Institute reanalysis (Krewski et al., 2000) of the Harvard six-cities study (Dockery et al, 1993) found that when treated on an equal-footing basis, SO2 has a stronger association with health than PM. The PM criteria document implies that this could be because SO2 is really a precursor to sulfate particles, but it needs to be emphasized that this is hypothesis, not a verifiable conclusion of the epidemiological analysis. Furthermore, different areas of the United States have different compositions of particulate matter, so regional differences may partly be explained by the different pollutant mixtures.
It is important that the environmental statistics community pays attention to the literature review in EPA’s criteria document. We feel that the document takes a cavalier attitude towards statistical interpretation issues, which go way beyond mere sensitivity of the results of epidemiological analyses to different kinds of analysis. Large questions of uncertainty and more broadly of variability (e.g. across subpopulations, spatial regions, seasons etc.) are of major concern in ascribing an overall cause and effect relationship, but the document is written in such a way as to imply that none of these issues ultimately affect the conclusions.
Clyde, M. (2000): Model uncertainty and health effect studies for particulate matter. Environmetrics 11, 745-764. Dockery, D. W., Pope, C. A., Xu, X., Spengler, J. D., Ware, J. H., Fay, M. E., Ferris, B. G. and Speizer, F. E. (1993): An association between air pollution and mortality in six U. D. cities. N. Engl. J. Med. 329, 1753-1759. Krewski, D., Burnett, R. T., Goldberg, M. S., Hoover, K., Siemiatycki, J., Jerrett, M., Abrahamonicz, M. and White, W. H. (2000): Reanalysis of the Harvard Six Cities study and the American Cancer Society study of particulate matter air pollution and mortality. A special report of the Institute’s Particle Epidemiology Reanalysis Project. Cambridge, MA: Health Effects Institute. Lumley, T. and Sheppard, L. (2000): Assessing seasonal confounding and model selection bias in air pollution epidemiology using positive and negative control analysis. Environmetrics 11, 705-716. Schwartz, J. (1993): Air pollution and daily mortality in Birmingham, Alabama. Amer. J. Epidem. 137, 1136-1147. Smith, R. L., Davis, J. M., Sacks, J., Speckman, P. and Styer, P. (2000): Regression models for air pollution and daily mortality: analysis of data from Birmingham, Alabama. Environmetrics 11, 719-744.
5. Research Projects and Programmes,
Teresa Alpuim, Editor
In this section of TIES Newsletter members are invited to describe the Environmetrics research projects they are involved with. It is our aim, not only to show the many different ways quantitative methods are being applied to Environmental Sciences, but also to give knowledge about who is working in what problems. We believe that this will contribute to increase scientific interchange between TIES members. Contributions should be sent to Teresa Alpuim (email@example.com). Academic programmes related to environmental problems are welcome, too.
The Indian Ocean Climate Intiative:
New Seasonal Forecasting Tools for Southwest Australia
Eddy Campbell & Allan Adolphson
CSIRO Environmetrics, Australia
The Indian Ocean Climate Initiative (IOCI) is a 5-year commitment by the Government of Western Australia to enhanced understanding of the climate of southwest Australia. New approaches to managing climate variability will be built on this base of enhanced understanding.
The east and north of Australia has received considerable attention from climate researchers to date, for two particular reasons. First, this region contains the bulk of Australia’s population base. Second, a degree of climate predictability has been achieved through greater knowledge of the El Niño-Southern Oscillation (ENSO) phenomenon in the Pacific Ocean. ENSO has been shown to have a clear impact of southern and eastern Australia. The climate of southwestern Australia is influenced to a much greater extent by the Indian and Southern Oceans, about which relatively little is known.
IOCI has now entered its final two years, during which the research groups are focusing on the opportunities defined in the first three years. Our group, CSIRO Environmetrics, has responsibility for developing a nonlinear time series approach to model climate processes. We are using a Bayesian approach to identify important predictors, and the time lags at which they act. This approach can be used in its own right, but is also of value for analysing long runs of global climate models to extract potentially useful relationships.
The ultimate aim of IOCI is the development of tailored climate products for southwestern Australia, such as decision support systems, using our new tools and understanding. This is an ambitious goal given that 30 years of research underpins the climate products based on ENSO. However, IOCI has been able to leverage other research and Government bodies to help in its cause. The results so far provide considerable scope for optimism.
Statistical methods and techniques to analyse spatial and spatial temporal data
A. Pollice, M. Coli, G. Garau, D. Posa, S. Terzi, L. Fattorini, C. Lagazio
"Statistical methods and techniques to analyse spatial and spatial temporal data" is a two year (2000-2001) University Italian research project of national interest with seven research units: Bari (resp. A. Pollice), Pescara (resp. M. Coli), Cagliari (resp. G. Garau), Lecce (resp. D. Posa), Roma (resp. S. Terzi), Siena (resp. L. Fattorini), Udine (resp. C. Lagazio).
The Bari research group works on "Bayesian Geostatistics", and extends Bayesian methods for the estimation of the variogram of Gaussian and non-Gaussian spatial processes. Within this context particular attention is dedicated to computational aspects.
The Cagliari research group develops "Regional econometrics: a comparison of data banks and models" and, by using spatial econometrics methods, proposes a model for interpreting the sociol-economic reality of Sardinia and for analysing the effects of economic policy.
The Lecce research programme works on "Methods of geostatistics for the analysis of data with a spatio-temporal structure", and develops the statistical analysis of multivariate spatio-temporal data using geo-statistical methods to compute maps and to reduce future risks in case of high concentrations of hazardous pollutants. Furthermore it will deal with the problem of estimating measures of simple and cross spatial auto-correlation for the study of dynamic phenomena.
The Roma research programme "Generalised STARMA model, extensions and applications in economy" works on the definition of spatial contiguity matrices, in order to account not only for distances, but also for other factors such as the time needed to cover the distance, the amount of exchanges or frequency of exchanges between different sites. The aim is the comparison of STARMA models characterised by a contiguity matrix with STARMA models characterised by "mean" contiguity matrix.
The Siena research group "Two stage sampling of biological populations" studies the two stage sampling of biological populations, looking for the properties of estimators when this type of sampling is employed.
The Pescara research group "State Space models for the representation of spatial temporal processes" extends research connected with the Kalman filter and resampling techniques to estimate time varying parameters of spatial temporal dynamic models. This approach allows researchers to verify whether the patterns of parametric changes are real or due to chance.
The Udine research group "Spatial analysis of epidemiological data: Statistical methods for case-control studies of the risk of disease around putative sources of environmental pollution" works about a critical review of the existing methods and to develop methodologies to investigate the presence of risks on human health.
The principal results of the research activities of the group are reported in 46 publications (29 papers published or to be published in national and international journals, 11 communications presented to national and international conferences and 6 working papers).
An intermediate Workshop took place on 22nd and 23rd January 2001 at Lecce, Italy.
BioGIS: A Prototype for an Intelligent Biodiversity Information System
Hebrew University of Jerusalem, Israel
BioGIS is an attempt to conceptually design, and actually develop, a web-based geographical information system (GIS) for storing, analyzing and visualizing data obtained from biological collections (e.g. herbaria and natural history museums). Biological collections form the basis for our knowledge about the taxonomy and geographical distribution of all kinds of organisms. Traditionally, information obtained from biological collections has been made available to the public by various kinds of 'hard copy' publications, particularly floras and faunas. Recent advances in computer and information sciences have opened new possibilities for the integration and harmonization of data obtained from biological collections. However, up-to-date, most of these efforts have concentrated on information compiled from 'secondary' data sources (i.e., floras and faunas) rather than on treatment of the primary information (i.e., the original specimen data).
BioGIS is a multifunctional biodiversity information system that was designed by a team of scientists of the Hebrew University of Jerusalem in an attempt to fill this gap. A unique feature of BioGIS is the integration of record-level information, species-level information, and geographical information in a unified GIS environment, which is completely and readily accessible through the Internet, and equipped with user-friendly, state-of-the-art tools for data analysis and visualization. Record-level information (e.g., collector name, collection date, determinator, etc.) is compiled from the original label of the specimen. Species-level information (e.g., taxonomic data, chorotype, growth-form, fruit type, etc.) is compiled from external sources, mainly floras for plants and faunas for animals. Geographical information (e.g., rainfall, temperature, soils, conservation status, etc.) is obtained from digital maps stored in the GIS. A key feature of BioGIS is the integration of point, grid and polygon data of species distribution in the same system, with the ability to select and analyze non-point data of species distribution based on their "spatial accuracy".
A variety of query tools allow the user to easilly select, analyze and visualize (map) records based on specimen-level attributes (e.g., specimens collected between 1950-1980), species-level attributes (e.g., geophytes with red flowers), and/or GIS attributes (e.g., specimens collected on sandy soils not more than 2 km from the nearest reserve). Further interactive map tools enable the user to zoom in and digitize any area of interest on the map (e.g., a particular hill or a buffer zone around a road) and then perform queries related to the selected area. Another feature of BioGIS is the ability to produce predictive distribution maps of selected species by using bioclimatic models based on user-specified variables. BioGIS has been developed using a combination of software tools including ESRI's ArcView Internet Map Server (IMS), MapCafé (a built-in set of Java routines), MS SQL Server, IIS, PHP, and Matlab. In contrast to most Internet applications, which are 'stateless' (i.e., the system returns to its original state after each operation), BioGIS retains the selected records between successive operations by storing them within the SQL Server database along with a session ID for each user.
The first version of BioGIS was recently implemented in Israel as a prototype for a National Biodiversity Information System (www.biogis.huji.ac.il). This system will be developed as a long-term, dynamic database, and will reflect the current state of knowledge on the distribution of plant and animal species in Israel. The system is currently developed as an informal alliance between the GIS Center and the National Collections of Natural History at the Hebrew University of Jerusalem (HUJ), the National Collections of Natural History at Tel-Aviv University (TAU), and the Science and Management Division of the Israel Nature and Parks Authority (NPA). The project is coordinated by Ronen Kadmon (HUJ), Daniel Zohary (HUJ), Tamar Dayan (TAU) and Avi Perevolotsky (NPA). Major objectives of the next phase of the project are the development of collaborations with other projects dealing with databasing of collection data, and the extension of the system already developed into other regions and more diverse taxonomic groups. Researchers interested in such cooperation are invited to contact the project leader, Dr. Ronen Kadmon (firstname.lastname@example.org).
Human Vulnerability to Environmental Change
Univ. Simon Bolivar, Venezuela
Univ. of New Hampshire, USA
The notion that humans have become more vulnerable in today’s society due to an important increase in the population during the twentieth century and the profound changes infringed by humans to the environment, is an easy concept to understand and to accept. But this multidimensional concept described in the literature in many different ways, is hard to define within a common general framework and even more difficult to quantify. It involves definitions of potentially damaging phenomena or hazards and their probability of occurrence. It requires socio-economic data to estimate human exposure and capability to cope to those hazards and adequate statistical methods to quantify the associated risk to the multiple environmental factors arising as a consequence of global environmental change.
A researcher from the Statistical Center and Mathematical Software (Centro de Estadistica y Software Matematico, CESMa) at Simon Bolivar University in Caracas (Venezuela), Lelys Guenni, has joined efforts during her sabbatical year, with a research group from the Complex System Research Center at the University of New Hampshire (USA) under the leadership of Charles Vörösmarty to pursue on this topic. Their emphasis is on the population of South America and its vulnerability to extreme weather events. For more information on this project you can contact:
Lelys Guenni, Professor
Departamento de Computo Cientifico y Estadistica
Centro de Estadistica y Software Matematico
Universidad Simon Bolivar
APDO. 89.000. Caracas 1080-A
ph: +58-212-906233 fax:+58-212-9063234
Charles Vörösmarty, Director
Water System Analysis Group
Complex System Research Center
Institute for the Study of Earth, Ocean and Space
University of New Hampshire
Durham, NH 03824
ph: +1-603-8623978 fax:+1-603-8620188
6. Recently Published Books
The Encyclopedia of Environmetrics will be published by Wiley in October 2001. Abdel.H. El-Shaarawi and Walter W. Piegorsch are the Editors-in-Chief of this major reference source for quantitative methods in Environmental Sciences. Divided into 12 sections, the Encyclopedia brings together over 500 detailed articles, all of which have been carefully selected and reviewed through the collaborative efforts of the Editors-in-Chief and the Advisory Board. John Wiley will publish it in October 2001 as a 4 volumes set in approximately 2500 pages. An overview including contents, authors, and sample articles are available on www.wiley.com/wileychi/eoenv/ Amezquita, Maria Cristina (2001) Biometrical Applications in Tropical Pasture and Agro-pastoral Research, Wageningen University, The Netherlands. This book illustrates the contributions of Biometry to scientific research in tropical pasture and agro-pastoral systems in Tropical Latin America in the last 2 decades. It is now being distributed by CIAT at a cost of US$ 38. If anyone is interested in acquiring a copy, please contact: Mr.Luis Alberto Garcia, Distribution of Publications, CIAT. Christakos, George (2001) Modern Spatiotemporal Geostatistics, Oxford University Press. Shumway, R.H. & D.S. Stoffer (2000) Time Series and Its Applications. Springer-Verlag. Webster, Richard & Oliver, Margaret A. (2000) Geostatistics for Environmental Scientists. Wiley.
7. Book Reviews
Liliana Gonzalez, Editor
Unfortunately, in this issue, it has not been possible to present any book review. However, we would like to use this section, to make a call to all TIES members interested in doing book reviews to contact the editor of this section, Liliana Gonzalez (email@example.com).
8. TIES Board of Directors
The following are the names of the elected members of TIES Board of Directors. All terms are from September 1, 2000, to August 31, 2002, except the 4-year terms of the regional directors.
President: Sylvia R. Esterby
President-Elect: Peter Guttorp
Secretary: Jari Walden
Treasurer: Francis J. Philbert
Publications Officer: Teresa Alpuim
Anthony R. Olsen
(01/09/00 - 31/08/04)
Richard W. Katz
(01/09/98 – 31/08/02)
(01/09/00 - 31/08/04)
(01/09/98 – 31/08/02)
Raymond L. Correll
(01/09/00 - 31/08/04)
Carlos Alberto de Bragança Pereira
(01/09/98 – 31/08/02)
TIES Newsletter is a publication of the International Environmetrics Society (TIES). It is published semiannually, or whenever the need arises, by The International Environmetrics Society and distributed to TIES members as part of their annual dues. Contact Sylvia Esterby, firstname.lastname@example.org, or Francis Philbert, email@example.com, for questions regarding membership and other benefits.
Objectives of the Newsletter include (but are not limited to):
To keep TIES members informed of what is happening within the Society; To cover news in latest developments in theory and applications of environmetrics; To be a forum for discussion of a broad range of issues which are of interest to members of TIES and are consistent with the objectives of the Society. To facilitate communication between environmental scientists and statisticians about research problems of mutual interest. To provide details about upcoming conferences and workshops To announce members' news that are worthy of notice or recognition (e.g., awards, prizes and honors received, promotions, appointments, etc.)
Communications, (e.g., contributions, comments and suggestions) regarding this publication should be addressed to the TIES Newsletter editors: Teresa Alpuim (firstname.lastname@example.org) or Bronwyn Harch (Bronwyn.Harch@CMIS.CSIRO.AU).
The Editors would like to encourage TIES members to submit items for publication in the Newsletter. We would like to have a very comprehensive publication that is of interest to our members by including items such as members' and regional news, Environmetrics and related conferences, research projects and programmes, book reviews, letters to the editor and articles of general interest.
We would like to thank the members who responded to our call and contributed to this issue. It is our hope that the Newsletter will be a valuable platform for discussion and exchange of ideas among us. We will be happy to hear your views about the contents and style of this issue. We hope that you will be a reader as well as a contributor.
NEW TIES Webpage: http://www.nrcse.washington.edu/ties/