05.06.2013 Views

PNNL-13501 - Pacific Northwest National Laboratory

PNNL-13501 - Pacific Northwest National Laboratory

PNNL-13501 - Pacific Northwest National Laboratory

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Study Control Number: PN99079/1407<br />

Watershed Model and Land-Use Decision Support<br />

Lance W. Vail, Mark S. Wigmosta, Duane Neitzel<br />

Innovative approaches are required to transcend the structural differences in physical and biological models, if such<br />

models are to be successfully linked to support natural resource management decision-making. Fuzzy computational<br />

methods were shown to be a valuable tool in linking physical and biological models.<br />

Project Description<br />

The purpose of this project was to support a science-based<br />

approach to natural resource management by assessing<br />

the effectiveness of fuzzy computational methods in<br />

linking advanced physical and biological models. We<br />

developed several prototype fuzzy tools that were able to<br />

incorporate “approximate reasoning” for representing<br />

processes and for developing categorical conclusions<br />

regarding the environment. We showed that fuzzy<br />

methods provide both an intuitive and technically<br />

defensible approach for linking physical and biological<br />

models.<br />

Introduction<br />

This study represents the second year of a 3-year effort to<br />

develop an integrated natural resource management<br />

framework. A significant obstacle to successfully linking<br />

physical and biological models was identified as the<br />

fundamental structural differences between such models.<br />

Physical models are more likely to involve a continuum<br />

representation, whereas biological models are more likely<br />

to rely on rules. Two significant issues, the pervasive<br />

vagueness of rules and the multivaluedness associated<br />

with temporal and spatial upscaling, suggested that fuzzy<br />

methods might be useful in overcoming some of the<br />

structural differences between physical and biological<br />

models.<br />

Fuzzy methods allow computing based on linguistic<br />

variables (words) as opposed to numbers. Words are less<br />

precise but closer to intuition and accommodate the<br />

pervasive imprecision of biological process<br />

understanding. Fuzzy computational methods exploit the<br />

tolerance for imprecision, uncertainty, and partial truth to<br />

achieve tractability, robustness, and efficient solutions.<br />

Fuzzy methods complement but do not reduce the need<br />

for probabilistic methods in either physical or biological<br />

262 FY 2000 <strong>Laboratory</strong> Directed Research and Development Annual Report<br />

process models. Whereas, randomness (probability)<br />

describes the uncertainty of an event occurring, fuzziness<br />

describes the degree to which an event occurs.<br />

Approach<br />

This project used a variety of fuzzy methods including:<br />

fuzzy arithmetic, fuzzy logic, fuzzy clustering, and<br />

adaptive neural fuzzy inference systems. Fuzzy<br />

arithmetic was employed using standard programming<br />

methods. The evaluations of fuzzy logic, fuzzy<br />

clustering, and inference systems employed a commercial<br />

software package. A series of rules and a database from<br />

the Multispecies Framework Process was employed to<br />

test the various fuzzy methods. These rules and data are<br />

used to define aquatic habitat diversity in the <strong>Pacific</strong><br />

<strong>Northwest</strong>.<br />

Results and Accomplishments<br />

Fuzzy Arithmetic. To establish a clearer understanding of<br />

basic fuzzy computational methods, we developed a fuzzy<br />

rainfall runoff model. Fuzziness (expressed as a<br />

membership function) was defined for saturated soil<br />

hydraulic conductivity, effective soil porosity, soil depth,<br />

and rainfall intensity. Fuzzy characteristics of habitat<br />

capacity (river width and depth) were estimated using<br />

standard fuzzy methods as a function of time. The fuzzy<br />

results are reasonable based on comparison with other<br />

(non-fuzzy) methods, but convey the uncertainty<br />

associated with the fuzzy input parameters.<br />

Fuzzy Logic. Fuzzy logic is an extension of multi-valued<br />

logic. It is based around the concept of fuzzy sets. Fuzzy<br />

sets relate to classes of objects with unsharp boundaries.<br />

Membership in a fuzzy set is a matter of degree. Fuzzy<br />

logic has been shown to be valuable for computing<br />

process interactions and is currently employed in many<br />

control applications (such as focusing camcorders,<br />

automatic transmissions) and decision support systems

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!