As the focus on performance metrics continues to grow within landscape architecture research and practice, tools emerge to help designers evaluate design alternatives. A new paper by Dr. Jon Bryan Burley, FASLA, of Michigan State University and Dr. Xiaoying Li and Dr. Shuyue He of Nanjing Forestry University describes an adaptation of a long-existent statistical model that can help designers and students understand and assess metrics across multiple design possibilities.
Friedman’s Two-way Analysis of Variance by Ranks (“Friedman’s Analysis”) is a statistical method described by Burley as enabling designers to evaluate the probability that there will be a significant difference between design alternatives, which could include a baseline condition, a traditional design, and a sustainable design. This difference can be evaluated in relation to performance metrics like visual quality, energy use, stormwater runoff, soil productivity, wildlife habitat suitability, and much more. Assessing the potential for performance in experimental designs, before finalizing design decisions, represents a novel way to incorporate landscape performance metrics early in the design process.
Dr. Burley’s interest in statistical approaches to design began when he was a faculty member at North Dakota State University, where a researcher devoted to potato cultivation shared an Analysis of Variance statistical technique similar to Friedman’s Analysis in evaluating productivity and disease resistance in potatoes.
Friedman’s Analysis may be completed with a spreadsheet and some algebra and is relatively easy to compute. Results will indicate the probability that a difference in the proposed design alternatives might be true, and to what degree the alternatives differ. In effect, this statistical model can tell designers whether a “sustainable” design has a high probability of achieving a significant improvement in the target performance metric(s) as compared to a conventional design. It can also predict whether the improvement is likely to be negligible or whether it could have a negative impact on the desired performance.
Alternatives with the “best” scores based on the defined performance metrics could be considered to be the “best” design, but Dr. Burley adds a word of caution to designers that “the top score or best scores for a specific treatment does not necessarily mean the treatment is the obvious choice…applying a treatment with the best scores may be prohibitively expensive or environmentally irresponsible. It still takes a person to decide what to do.” According to Dr. Burley, results of Friedman’s Analysis often depend on the creativity of the designer in creating alternatives. Limitations of the model are also discussed.
Dr. Burley used Friedman’s Analysis for a mine reclamation project at the Big Stone Natural Wildlife Refuge in North Dakota to evaluate potential for habitat creation. Using the model, Dr. Burley and his students were able to demonstrate that their proposal was predicted to be more likely to create habitat for 13 bird species as compared to alternative designs. According to Dr. Burley, another promising use of the model could be to evaluate design alternatives that mitigate damage in areas around the world with a high risk of natural and man-made disasters.
Burley, Li, and He’s paper—part interview, part research paper, and part personal reflection—also puts this new method for evaluating potential landscape performance into context by providing background on the evolution of design evaluation in landscape architecture. Dr. Burley emphasizes how far landscape architectural research has come since the 1970s and celebrates the rise of a different type of landscape architect—one that has scholarly skills similar to those of other disciplines and the ability to blend planning and design with science.
This paper is recommended reading for emerging landscape performance researchers, particularly those interested in adding statistical methods to their toolbox for evaluating design alternatives. The authors define research as simply the generation of new knowledge, and they encourage landscape architects, particularly students, to generate knowledge by conducting performance metric analysis of their own experimental design projects. “Landscape architects are essential in such efforts because they are the ones who can design the treatments and have skills in calculating the environmental metrics,” says Dr. Burley. “I believe they have a key position in this type of…evaluation. Plus, it illustrates the value of their contributions in creating quality environments.” With broader use and further study, statistical techniques may well take their place as important tools to help designers, developers, and site managers consider landscape performance early in the design process.