Our primary objective is that the participants become familiar and
comfortable with unbiased parameter estimation for site index models and other
self-referencing functions. The bases for such parameter estimation is
stochastic regression, or a form of simplified mixed effects modeling, using sound and
tenable assumptions on errors associated with the reference variables such as
site index in site index models. The traditional methods of parameter
estimation produce biased results, which are evident from unduly skewed or
distorted curve shapes. The newer methods produce unbiased results and lower residual variance, but are
somewhat more demanding and in fact are perceived as too difficult. The workshop
will present simplified methods for unbiased parameter
estimation. The presented approach to stochastic regressions is simple enough
that we will demonstrate it initially in Excel and all participants will solve a small problem using
Excel. Following this, the same approach will be covered using the most commonly applied statistical software, such as SAS, S+,
and TSP. All participants will be expected to work on
their own laptops, with their own statistical software, including their own
version of Excel ( with the Excel Solver Add In).
Traditional methods of parameter estimation for site index models assume that all parameters in them are fixed and that there are no errors in site indices. This leads to so called base-age specific parameter estimation, which results in biased curves with the amount of bias dependent on specific selections of base ages. The untenable assumption that site indices have no errors results in fitting the 3-dimensional surface of height-age-site relationship with random distortions of the surface by the actual errors in site indices, to which the surface is forced to be attached, as illustrated by the figure below (white columns are deviations above the surface, black columns are deviations below the surface, and little black dots are zero-residuals on the surface, e.g., at base age).
Note that even if the fitted model is not constrained to predict heights at base age equal site indices, the principle still remains the same. The restrictions of the surface still apply, but they are mitigated by a common-scalar-factor offset that corresponds to approximate average value of the actual errors in site indices.
Modern methods of parameter estimation for self-referencing functions estimate the errors in site indices, which are essentially random effects, simultaneously with the estimation of all the fixed model parameters. When fit in this way, the 3-dimensional relationship of height, age, and site, is fit as an unbiased 3-dimensional surface as illustrated in the figure below (symbols are the same as above).
For any inquiries send the email or FAX to:
|
E-mail: biomat@smokey.forestry.uga.edu
|
The early registration fee is US$150. Late registration fee will be US$180. Equivalent fees for full-time students are $100 and $130. Workshop attendants can purchase additional Workshop Meals Packages at a price of $100 each for their accompanying family members and friends not attending the workshop. The registration fee includes all the meals and the hospitality room treats.
The full fee must accompany the registration. Participants may pay by check (in U.S. dollars drawn against a bank located in the U.S.), bank draft or international money order payable in U.S. dollars. Make checks, drafts, or money orders payable to The University of Georgia-Southern Mensurationists.
Cancellations received until deadline will be granted full refund. No refunds will be possible after this deadline. The substitution of a new participant may be made at any time.
|
Workshop Instructors: Chris J. Cieszewski and Mike Harrison, |
Program Committee (in forming, alphabetically)
| Robert L. Bailey | WSFR, UGA, GA |
|
Chris J. Cieszewski |
WSFR, UGA, GA |
| Mike Harrison | WSFR, UGA, GA |
| Stacey Martin | WSFR, UGA, GA |
The workshop will be held at the Holiday Inn Beach Resort on Jekyll Island, Georgia, 20 min from St. Simon's, with many attractions for adults and children alike. A block of rooms has been set aside at reduced rates of $62 for conference participants registering before the early cancellation deadline. Make the reservations directly with the Hotel by dialing (912)635-3311 or, toll-free 1(800) -7-JEKYLL or, by FAX at (912)635-2901.
A perfect island get-a-way location, miles of white sand beaches, the restored
historic district, year round recreation, golf, tennis, historic and scenic
tours, a water-fun park, fishing, marinas, Jet Ski rentals and "dolphin
watch" cruises, great shopping, nature trails, good food, and lots of family-time
possibilities.
To see weather forecast for Jekyll Island click on this icon:
To listen forecast for Southeast click here:
Travel arrangements are the responsibility of each participant.
If you need directions for travel click on this icon: