Suitability Maps vs Orchard Records: Which to Trust, and When

Two different kinds of evidence

Before you plant or buy land, the information you can get is modelled. A site-level suitability report may draw on soil surveys, rainfall and temperature records, slope, flood and drainage layers and satellite imagery, and turns them into a statement about what should grow there. After you plant, the information is observed: what this tree, in this spot, did this year. The two answer different questions, and the mistakes happen when one is asked to do the other's job.

A map is a cheap way to generate questions about a place you have not lived on yet. Your own records add the site-specific evidence for the ground you actually farm. This article is about using each for what it is good at, and about how to let the second overrule the first.

What a suitability report is good for

A modelled report is strongest at the stage where you know least. You are comparing parcels, or deciding what to plant on a block you have owned for a week. A report can show you things you may not have thought to check: slope and aspect, whether a low corner sits in a flood zone, how far the nearest water is, and which crops a model thinks the climate and soils favor. Growers who tried one report said much the same. In a December 2024 to February 2025 GrowingFruit.org forum thread (https://growingfruit.org/t/canopy-farm-management-compass-mapping-game-changer/68524) about a then-free suitability report from Canopy Compass, one grower called it an extremely robust tool that factors in many things worth attention, and another said it identified factors they had not considered before. A third liked the slope, 3-mile-radius and water views and said they would probably use it before future land purchases.

Canopy Compass is a separate product from a separate company. It has no connection to CanopyTwin. We discuss it here only because the forum thread is a useful, public example of what growers found when they compared a modelled report to their own ground.

Where the model and the ground disagree

The same thread is a good catalogue of the ways a modelled report can miss. These are the experiences growers on the forum described, not a verdict on any tool:

• Species that grow fine anyway. One grower's report rated the property as not suitable for black locust or black walnut, yet there are plenty of both growing happily at the edge of the field. Another grower noted that black walnuts grow like weeds where they live and wondered what soil depth, pH or type the tool assumes.
• Soil and water details. Another grower found the report "way off" on soil pH and flood zones and disagreed with its soil structure results, while still finding it useful for raising factors not considered before.
• Rainfall limits versus irrigated ground. A grower in Montana, where the average is about 14 inches of precipitation a year, saw cherries and plums ruled out on rainfall alone. Everything around there is typically irrigated. Plums grow wild by the ditches and are planted in city parks, and the grower gets generous crops every year from their own University of Saskatchewan cherries. The report suggested apples only.
• Dated imagery. The same grower could tell the satellite views were out of date: former cropland nearby had become subdivisions in the last four years or so.

None of that makes a model useless. These growers reported differences between their reports and what they see locally, and a report cannot know about the one ditch, the one raised bed or the one irrigated corner you manage. One grower in the thread guessed that the tool leans on averages and historical elevation data, and suggested that water shedding or retention strategies create opportunities the average cannot see. That was a guess, and so is another grower's guess that the tool is built on USDA soil survey data. Another grower recalled a survey in which, for a generalist like black walnut, the answer was often that it "doesn't matter". We do not know how any of it was built, and you do not need to. What matters is the habit one grower summed up as "trust but verify".

The map can also be right, and you should still check

The thread has a good example of the other direction. One grower's soil report described fantastic silty loam, classed as prime farmland or farmland of statewide importance for both soil units on the property. Yet the suitability report flagged drainage and a potentially high water table as a challenge for chestnuts. The grower said test pits dug before the purchase broadly supported that concern, and was planning modifications and specific planting locations: possibly mounds or raised beds for chestnuts, and holding water in other areas for blueberries and cranberries.

Notice what happened there. The model raised a flag, and test pits broadly supported it. Then the grower planned at the scale of individual planting spots. Two sources agreeing on a risk is far better than either alone, and it is a different thing from trusting a map.

Verify on the ground before you commit

Treat each line of a suitability report as a question. Some cheap ways to answer them:

1. 1Dig test pits at several spots, in wet weather if you can. Look at how deep the topsoil is, whether there is a compacted or gleyed layer, and where water stands.
2. 2Walk the neighborhood. What grows well in yards, ditches, parks and old homesteads nearby? Wild plums by a ditch tell you more than a rainfall layer.
3. 3Ask neighbors and local growers what has failed, not only what thrives. Frost pockets, late freezes and wet springs rarely show up in a summary.
4. 4Test the soil yourself for pH and basic fertility, and compare the result to the report. If they differ, check how each was sampled, investigate the difference and note it.
5. 5Check the date on any imagery before you rely on it for access, buildings or tree cover.
6. 6Decide which risks you can fix (drainage, irrigation, raised planting) and which you cannot (a late-frost hollow). Plant the fixable ones in the right spots first.

After planting, your records become the evidence

Once trees are in the ground, the question changes from "what might grow here?" to "what is this tree actually doing?" That is where per-tree records come in. A tree record gives an answer for each tree: how it established, how it grew, when it flowered and fruited, what went wrong, and what you did about it.

What to keep for each tree is not complicated:

• Species and variety, planting date and exact location on the map
• Dated photos taken on a regular walk, from the same angle where you can
• Notes on vigor, leaf color, dieback, pests and disease, and anything unusual after weather events
• What you did (mounded, mulched, irrigated, pruned, replaced) and what happened afterward
• Fruit set and harvest by tree, once the tree bears

Each row is small. The value is that they pile up in one place per tree, so a pattern becomes visible: the row at the low end of the block is always yellowing in spring, or the tree near the ditch outperforms its siblings. Those patterns are the real suitability data for your site.

A worked example (illustrative)

This is an invented example to show the logic, not a real orchard or result. A grower reads a suitability report that flags a possible high water table on the north end of a new block, and rates chestnuts poor there. She digs three test pits and finds standing water at about two feet in the lowest one and none in the other two. She plants chestnuts on mounds at the north end anyway, tags each tree, and records a photo and a note each month.

• Year 1: all mounded trees leaf out normally. One tree at the lowest pit shows yellowing after a wet spring. Note: standing water in the swale after three days of rain.
• Year 2: the yellowing tree has declined, but the others on mounds a few feet higher are fine. Note: drainage looks like a spot problem, not a block problem.
• Year 3: the grower replaces the one tree with a different species suited to wet feet, and moves two future plantings away from the swale.

The report asked the right question. The test pit narrowed it. The per-tree records guided the next planting decisions, tree by tree, more specifically than the map could have. If the records had shown the whole block failing, that would have been the evidence to override a favorable map, too.

Letting your records overrule the model

Over a few seasons the balance should shift. Early on, you lean on the map because you have nothing else. Later, your own history is the better source, because it is specific to your trees, your microclimate and your management. Some rules of thumb:

• If the model says no and your trees say yes, weigh your trees' record heavily and investigate the discrepancy, especially once you have more than one season of observations and have noted any irrigation or protection that makes it work.
• If the model says yes and your trees say no, look for the cause in your records before blaming the site: planting depth, drainage, a water stress period, rootstock.
• If both agree, you have your strongest signal. Plant more of what works and stop fighting what does not.
• Keep the old report with your records. Noting where it was right and where it was wrong tells you how much to trust the next one.

A suitability map is a hypothesis. A test pit is a first experiment. A per-tree record, kept for several seasons, is the result.

How CanopyTwin fits in

CanopyTwin is the records half of this picture. It does not produce a suitability report and is not related to Canopy Compass. It is for what comes after planting. You walk the site photographing plants with your phone, the GPS in each photo clusters the frames into individually tracked trees on a satellite map, and each tree gets a page with its photos over time, counts and notes. Voice notes captured in the field are transcribed and filed to the right tree, so dated observations stop living in someone's memory, and CSV export is available from the Orchard plan up. Every plan starts with a 30-day free trial, no credit card required, and the demo farms on the site are open if you want to see what a tree-level record looks like before you start your own.