Layer Cake Planning: A Practical Walkthrough of McHarg's Method
Ian McHarg's "Design With Nature" came out in 1969 and changed how landscape architects and planners think about site analysis. The core idea is straightforward: don't pick a location for a project based on one or two factors. Instead, map every relevant environmental condition, layer them on top of each other, and let the overlaps tell you where building makes sense and where it doesn't. The book introduced what eventually became GIS as a discipline, even though McHarg was working with physical transparency sheets and light tables before anyone had thought of digital mapping. The method itself breaks down into a sequence that most practitioners learn through doing it wrong a few times first. You start by listing the environmental factors that matter for your specific project type. Soil erodibility, slope class, flood frequency, wildlife habitat, vegetation cover, hydrogeology, cultural resources, access corridors. The list changes depending on whether you're siting a subdivision, a highway, or a park. Then you obtain or produce maps for each factor at the same scale. After that, you assign a suitability rating to each unit on each map — typically something like high, moderate, low, or unsuitable — and layer the maps physically or digitally to find the areas with the best cumulative score.
The Practical Workflow Behind Ian Mcharg Design With Nature
Here's how I've actually done this on real projects, not the textbook version. First, you establish the study area and the resolution you need. A county-level water quality study might use 1:24,000 scale maps. A site-specific residential development might need 1:2,400 or better. If your base map resolution is too coarse, the whole overlay exercise becomes decorative — you'll get clean-looking result maps that don't reflect ground reality. Next, data acquisition. This is where most people stall out. The best published sources in the US are usually the USDA Natural Resources Conservation Service soil surveys, the USGS 7.5-minute quadrangle series, the National Hydrography Dataset, and FWS wildlife habitat maps. Each has different date stamps, different accuracies, and different thematic biases. Soil surveys might be from the 1970s in some counties and updated in the 2010s in others. Don't assume everything lines up just because it's federal data. Once you have your layers, you convert them to a common classification system. McHarg's own book used a five-point scale ranging from very high to very low suitability. I tend to use a similar five-point ordinal scale but I always document exactly what each level means in operational terms. "Moderate suitability for development" should translate to something a contractor or engineer can act on, not just a color on a map.
The overlay step is the part that sounds simple and isn't. With physical overlays you align your transparencies using common control points — road intersections, section corners, bench marks. Misalignment of even a quarter inch at the scale McHarg often worked can shift a boundary between a suitable and unsuitable zone. With digital overlays, projection matching is your equivalent headache. If your soil layer is in NAD 27 and your DEM is in NAD 83, reprojecting introduces slight distortions that accumulate across layers. Always snap everything to the same datum and verify with control points.
Get the Full Details
A Specific Problem I Ran Into
On a wetland delineation project in the lower Piedmont a few years ago, I discovered that the NRCS soil survey had mapped a particular parcel as well-drained sandy loam across its entire extent. The field work told a different story — seasonal saturation was evident in the lower third of the property, and the community vegetation was consistent with a hydric soil. The soil map was simply outdated and didn't reflect a realignment of the drainage pattern after a nearby road construction project completed in the early 2000s. If I had overlaid that soil layer uncritically with the flood plain and wetland layers, the result would have been wrong — the saturated area would have been classified as suitable for development when it shouldn't have been. The workaround was to pull the county's recent aerial photography, compare it against the existing soil boundaries, and manually adjust the soil polygon edges where the ground evidence diverged from the published survey. I also pulled the local hydrologic soil group data from the NRCS Web Soil Survey, which sometimes has more recent field verification notes than the printed map. It added about three hours to the data preparation phase but prevented a serious error downstream.
Weighting Is Where the Subjectivity Lives
McHarg assigned weights to each factor based on his professional judgment, and he was transparent about that. He wasn't claiming mathematical objectivity. This is the part that trips up beginners. You can run a multi-criteria decision analysis on a computer and get a fancy choropleth map that looks scientific, but the output is only as credible as the weight values you put in. A slope factor weighted at 0.3 versus 0.7 will produce dramatically different suitability zones, and there's rarely a universally correct answer for either number. What I've learned is to test sensitivity by running the overlay with different weight combinations and observing which areas stay stable and which flip. The stable areas are your confident recommendations. The flipping areas are where you need more field data or where you should present multiple scenarios to the client rather than a single "answer." McHarg himself acknowledged this in later editions of the book and shifted toward presenting results as probabilities rather than certainties. Another nuance that doesn't get enough attention is the edge effect between map units. When you overlay a soil map with a slope map, the boundaries rarely align perfectly. You end up with fragments of polygons along the seams that don't represent real conditions — they represent cartographic artifacts from two independent mapping efforts. In practice, I apply a minimum mapping unit threshold and dissolve small fragments below that size before doing the overlay. Otherwise you spend more time cleaning up noise than analyzing signal.
When This Approach Fails Completely
McHarg's overlay method assumes static conditions. It treats the landscape as a set of fixed layers stacked through time. That works fine for certain applications — siting a permanent infrastructure element where long-term stability matters most. But it breaks down for anything involving dynamic processes. Climate-driven sea level rise, changing precipitation patterns, successional vegetation trajectories, or cumulative impacts from adjacent developments are all outside the method's frame. If your project timeline extends beyond a decade and involves environmental change, layer-cake analysis alone will give you a false sense of precision. In those cases, supplementing the McHarg approach with process-based modeling is necessary. I've used SWMM for stormwater dynamics, INVEST for habitat connectivity under different land use scenarios, and HEC-RAS for floodplain changes. These tools don't replace the overlay method — they handle the dimensions the overlay method can't. Used together, they cover more of the problem space. The book itself is still worth reading, not as a technical manual but as a statement of intent. McHarg was arguing that planning should be informed by ecological reality rather than imposed on it. That argument is as relevant now as it was in 1969. The tools have changed, the data is better, the resolution is finer, but the underlying logic — understand what the land can and cannot do before you decide what to build on it — hasn't changed at all.
