The map

The map draws each observation as one point. It can draw a grid of summaries below the points, and reference layers below the grid.

Points

Open the Points control.

Color and size

• Color by sets what the color of a point means. You can select a category or a measurement. The categories include each taxonomic rank from kingdom to species, and also the common name, the identification rank, the cluster, the land cover, the year, the month and the enrichment level.
• Size by scales each point by a measurement. This lets you read two values at the same time.

A category gets a stable color. One species keeps the same color on the map, in each chart and in the key. A measurement gets a color gradient instead. A gradient reads as more; a set of hues does not.

The point gradient

A measurement needs a scale of colors. Set that scale in Style → Point gradient.

• Match the layer is the default. Read the next section.
• The preset gradients are Blue, Green, Warm, Purple, Viridis-ish and Greyscale.
• Custom opens the gradient editor.

The editor holds between 2 and 8 stops. Use the + button between two swatches to add a stop there. The new stop takes the color that the gradient already had at that point, so the gradient does not change until you move it. Use the × button on a swatch to remove that stop. Use Reverse to turn the gradient end to end.

Two stops give you low to high. More stops give you a middle that you can read. A diverging scale needs a neutral middle, so it needs three stops.

Note More than eight stops stops reading as an order. It starts to read as a set of categories, and the reader must then decode it.

Point colors that match the layer

Eight fields have a map layer that draws the same thing. Elevation, slope, aspect, NDVI, NDMI, soil moisture, temperature and rainfall are all on the map as layers and also on each record as a value.

When you color the points by one of these fields, the points take the colors of that layer. A dot then reads against the ground below it. A green dot on the NDVI layer sits on green ground when the two agree, and stands out when they do not.

Some fields share the units of their layer. Elevation in metres is elevation in metres. For these fields the points also borrow the layer's value range, so the same color means the same number in both. The other fields are normalised, so the points borrow only the colors.

To turn this off, select any other gradient in Style → Point gradient.

Caution A layer shows current or recent conditions. A point shows the conditions on the date of the find. The colors match, but the dates do not. Read the section on layers and heatmaps.

The two keys

The map shows two keys, and they answer two different questions.

• Observations explains the points. Below the title it names the field you colored by.
• Map layers explains the tiles below the points. It has one section for each layer that is switched on.

Click Key to fold both keys away. The app remembers your choice. The keys also move aside when the record panel opens.

Heatmaps

Open the Heatmap control. A heatmap puts the observations into a grid. It then shades each cell by one summary statistic.

48,000 points on top of each other show you only where the data is dense. A grid shows you what is in an area.

The seasonal heatmaps

Seasonal activity and In-season hotspots need a date and a window. The window is between ±3 and ±60 days. Both controls appear when you select either heatmap.

These two are the most reliable. Each cell is divided by its own total, and that cancels most of the effort bias. The shape of a cell's season does not depend on how many people visited it. It depends only on when they found things.

The field heatmaps

A field heatmap gives the cell mean of one enriched value. The fields are 7-day rainfall, temperature, soil moisture, the wetness index, slope, aspect, sun exposure, wind exposure, NDVI, NDMI and elevation.

Caution A field heatmap is not a coverage map. A cell has a value only where a person recorded a find. An empty cell means that nobody went there. It does not mean that the ground is dry or flat.

Aspect is circular, so it gets a compass key. It does not get a low-to-high bar, because 359° and 1° are neighbours.

Cell size and shape

Cell size runs from about 100 m to about 28 km. A smaller cell is more precise and more noisy. Below approximately one kilometre the grid stops to summarise. It then draws about one cell for each observation.

Cells are hexagons by default. Each neighbour of a hexagon is the same distance away, and shares a full edge. A hex grid therefore reads as a surface. A group of finds is also not divided differently when it moves against the axes.

Squares are available in Style. A square cell has the same area as the hexagon it replaces, so a change of shape does not change the resolution.

Heatmap opacity and Map layer opacity are two sliders in Style. They control two different stacks. A faint hillshade below a solid heatmap is the usual case.

The heatmap gradient is set in Style → Heatmap. It uses the same editor as the point gradient, and holds the same 2 to 8 stops.

Access layer

The Access section of the layer window draws land-access areas: public access, fee status and collecting. It is separate from the third-party land ownership tiles. See Foray planner for the controls and limits. The fee and collecting values are estimates.

Wind and aspect vectors

Each observation carries a terrain aspect and a wind-exposure index. Aspect is the compass direction that a slope faces. No observation carries a measured wind.

The vector heatmap therefore draws the mean terrain aspect. The length of an arrow shows how consistently the cell faces one way. The color shows the wind exposure.

Note A short arrow means mixed terrain. It does not mean calm air. The heatmap changes to real wind, and relabels its key, after the pipeline samples ERA5 wind into wind_u and wind_v.

Basemaps and layers

There are two controls, because these are two different questions.

• Basemap is one choice from five. You make it once and rarely change it.
• Layers opens a window for the overlays.

The layer window stays open while you work on the map. It does not close on the first click. You judge a layer by what the map looks like, so the map stays visible below the window.

The layer window

The window lists each layer that is drawn, topmost first. Each one has:

• Order — the arrows move a layer up or down the stack. ⤒ and ⤓ send it straight to the top or the bottom. Overlays hide each other. Land ownership below a hillshade is a different map from the same two in the opposite order. A layer that you switch on goes to the top.
• Solo (S) — draws that layer on its own. The other layers stay switched on, and the list shows them dimmed. Read Solo.
• Opacity — each drawn layer has its own. This multiplies into the global tile opacity in Style. It does not replace it. You can therefore dim the whole stack without dimming the one layer you want to read.
• Blend — how the layer combines with the layers below it. Read Blend modes.

The window also has a Search box. Search by a layer name anywhere, or by the start of a group name.

Solo one layer

Click S beside a drawn layer. The map then draws only that layer.

The other layers stay switched on. The list shows them dimmed, and they come back exactly as they were. This lets you answer "what is this one layer adding" without a loss of the stack that you built.

Click S again, or click Un-solo at the top of the window, to draw everything again. When you switch any layer on or off, the solo also ends.

Blend modes

Opacity is not the correct control for two rasters. Two layers at 50% is two layers that are half washed out. The thing that you wanted — the shape of the hillshade over the color of the land cover — is what 50% of each destroys.

A blend mode keeps both layers at full strength. It combines them by value.

The browser does this work for each frame. No tile is fetched again, and nothing is computed again. A blend mode therefore costs nothing to try.

Each layer starts at Default. Default follows the Blend when stacked setting in Style, and applies only when two or more layers are drawn. When you select a mode for one layer, that layer keeps it.

Caution A blend mode combines a layer with everything below it, and that includes the basemap. The same two layers over satellite imagery and over the gray canvas do not look the same. Neither result is the layer on its own. Use Solo for that.

What the layer groups contain

The basemap is a light gray canvas by default, so the observations are the only saturated thing on the screen. Street, terrain and satellite are one click away.

• Terrain — hillshade, USGS topo, USGS imagery, OpenTopoMap relief. The gray base plus the hillshade gives you relief without color.
• Weather — live US radar, US rainfall over the past 24 hours, global satellite rainfall, land surface temperature.
• Ground — ESA WorldCover land cover at 10 m, SMAP soil moisture.
• Vegetation — MODIS NDVI greenness, forest and land cover type, and Sentinel-2 greenness and canopy moisture (NDVI and NDMI). The forest type layer is USGS GAP ecological systems at 30 m, grouped into the types a forager separates. WorldCover says tree cover; this says which trees. The Sentinel-2 layers come in two forms. The free form is a rolling composite of the last few weeks. The member form is a median over one season and one year, for a comparison of the same phenological moment across years.
• Forest structure — USFS TreeMap canopy density, stand height and stand-size class at 30 m. Together they are a proxy for the maturity of a stand. Many ectomycorrhizal fungi need an old host. US forests only, modelled from 2016.
• Terrain analysis — slope, aspect, topographic wetness (TWI), wind exposure and solar exposure. All five are computed from the SRTM elevation model. Wetness finds the ground that collects water. Exposure and solar find the ground that dries quickly. Aspect and slope are below both. Open to everyone.
• Soil — texture class, depth to bedrock, sand content as a proxy for drainage, a sand/silt/clay composite, and the two soil taxonomy layers below. Texture decides how long the ground stays wet after rain. That is the half of fruiting weather that the rain layers cannot tell you. The composite has no scale to read a value from. Its purpose is the boundaries, which often follow the ground and not the surface.
• Fire and disturbance — years since fire, burn severity, this year's burn scars, active fires, a computed dNBR severity, forest loss and cutting (Hansen Global Forest Change), and a Sentinel-2 canopy-moisture crash. The last one flags beetle-kill and drought die-off. Years since fire, burn severity and forest loss are open to everyone. The computed layers are a membership benefit.
• Context — place labels, hiking trails, land ownership.

Soil taxonomy

Two layers draw the USDA soil classification, at 250 m, worldwide.

Soil taxonomy (USDA orders) paints the soil order at each pixel. There are twelve orders, and they are the top of the hierarchy. Orders separate soils by how they formed, so the boundaries often follow the geology and the climate, and not anything that you can see on the surface.

Soil taxonomy: chosen classes paints only the great groups that you choose, and leaves all other ground blank. Choose them in the class browser below the key. Each chosen class keeps the color of its order, so a selection across several orders can still be told apart.

Caution The chosen-classes layer is a soil filter and not a prediction. It says that the ground is the kind that you asked for. It does not say that anything grows there, and it knows nothing about the trees that decide whether anything can.

The class browser

The source data is at the great group level, which is the third of the six levels of the hierarchy. It has approximately four hundred classes. Four hundred swatches is not a key. It is a lookup table that you must read line by line.

So the key for these layers is a browser. Switch either layer on, and the Map layers panel gives you:

• The twelve orders, as the key. Click one to show only its own great groups. This answers "what are all the podzols here", which a search box cannot: no podzol is named "podzol".
• A search box. It matches the name of a great group and also the name of its order.
• The list. Each row shows the color of its order and its name.

Hover on a row to read a one-line description. Click it to open the whole description, and a link to the Wikipedia article for that order.

How to choose classes

The chosen-classes layer adds a checkbox to each row. Tick the classes that you want, and the map redraws.

Above the list there are three more controls:

• *Pick these n takes every class that the search and the order chip leave. This is what makes the search into a selection tool: filter to Spodosols, click Pick these*, and you have chosen the podzols. Click it again to remove them. With no search and no order chip, it takes all of them.
• None clears the selection.
• FRMS set is a starting selection of eighteen great groups that FRMS members flagged as matsutake ground on the Front Range. These are the cool, acid, sandy and volcanic soils, across five orders. Use it as a start and then add or remove classes.

There is no practical limit on how many classes you choose. A selection of every class is too long for a web address, so the app sends a large selection in the body of the request instead.

How to read a great group name

You do not need the list to read a soil name. The names are built from a fixed vocabulary, and they are built to be read:

Dystrocryepts is therefore a cold, acid soil with weak horizons. The app decodes the name for you in the class browser, part by part.

The ending is always the order. Read Soils for the twelve orders and where to read more about them.

Layer keys and dates

Each measured layer carries a key. All the keys collect into one Map layers panel, which appears when you switch a layer on.

A layer that changes by day carries a date picker in that panel. The picker opens a few days in the past. Each satellite product has a latency, and a request for today returns blank tiles. Blank tiles would read as "no rain" and not as "not processed yet".

When you zoom in past the native resolution of a layer, the app stretches its tiles. It does not resolve them more finely. The key says so.

A layer that the app cannot reach says so. It does not draw nothing. An empty ownership layer would otherwise read as "no public land here".

Map layers show current conditions; heatmaps show historical ones

Map layers are rasters drawn from external sources. They reflect current or recent measurements, not conditions at the time of each observation.

The rainfall, temperature, soil-moisture and NDVI heatmaps are different. They use values the pipeline sampled at the date of each find. Only those can help explain where and when a species was recorded.

Other map controls

• Drop a point puts a marker anywhere. The panel gives its coordinates, the heatmap value in the cell below it, and the nearest loaded find. Drag the marker to move it. Press Esc to remove it.
• My location puts a dot at your position, with its accuracy circle.
• Live clustering runs k-means in the browser, by features or by geography.
• Save image flattens the whole map, with its tiles, into a PNG.
• Map key folds away when it is in your way.

The record panel

Click a point to open its record. Hover on the label of a row to read what the value means. On a touch screen, tap the label.

Sections fold. The app remembers which ones you folded.

Three values are modelled from terrain and not measured: the wetness index, the solar exposure index and the wind exposure index. The tips say which.

Caution Read the Accuracy and Precision rows before you trust the terrain below them. iNaturalist obscures the location of sensitive taxa, and of records whose owner asked for this. Everything below was sampled at the published point.

Offline use

You most often read a map of where things grow in the place that it describes. That is where a connection is least likely.

Three things save into the browser separately, because they cost very different amounts:

• The app — each page and the code behind it. Charts and Analysis then open offline too, and not only the page that you were on.
• Observations — the dataset that the map, the table and the charts read. This is the large one.
• An area — the imagery for the place on the screen, for each layer that is drawn. You choose how many zoom levels closer to save.

Move the map to your destination before you save an area. The app shows the tile count and an approximate size before it downloads anything. The count increases approximately four times for each additional zoom level.

The app refuses a save of more than twenty thousand tiles. It does not trim the save to fit. An area that quietly has no edges is worse than an area that was never saved, because you find out where there is no signal to correct it.

How to manage saved areas

Give an area a name when you save it. The collection is on the Offline page, which you open from the account menu. You can rename an area, save it again to collect anything that failed, delete it, or open it. Open moves the map back over it.

Name your areas. You read a list of saved places weeks later. By then a set of coordinates is a puzzle, and "north ridge" is an answer.

When you delete an area, the app frees only the tiles that no other saved area needs. Two areas over the same valley share tiles. The app must not make a hole in the second area when you remove the first.

The Offline page also shows the total size that the browser reports for this site. That figure is the accurate one. The size of each area is an estimate, and the page says so. Tiles come from hosts that do not all report a length, and the app cannot measure a tile after it is stored.

What the app does not save

The app saves nothing more than its own shell without a request. A dataset and several hundred tiles on somebody's mobile data is not a feature.

Saved data stays in this browser on this device. The app does not upload it. It does not follow your account. It goes when you clear the site data of the browser. You can install the app to a home screen.

Note You can save an Earth Engine layer into an area like any other layer. The tile URL of an Earth Engine layer carries a token that expires in hours. The app therefore files these tiles under the layer, and not under the URL. Under the URL, each tile that you saved would be unreachable when you read it in the woods.

Caution Each tile service sets its own terms for bulk downloads. Save the area that you go to. Do not save a region.

Taxonomy

Each observation carries its full ancestry: kingdom, phylum, class, order, family, genus, species. The app resolves the ancestry against iNaturalist when it fetches the record.

All seven ranks are dimensions. You can color by them, group by them, filter to them, and run the analysis over them.

This replaces a split of the species name on its spaces. That method produced a genus only when a record was identified to species. It could not reach family or above at all. A record identified only to Amanitaceae had a "species" of "Amanitaceae" and a genus to match. It now has a family, an empty genus and an empty species.

Note That is the honest answer. It also means that a filter at species level quietly removes each record that nobody identified that far. Filter at the coarsest rank that answers your question.

On Data → Species, the Rank selector sets what the list shows and what the filter applies to. The same control therefore narrows a view to one kingdom or to one species.

The app offers only the ranks that the loaded dataset fills. A dataset exported before this existed carries species and genus and nothing else. A change of rank clears the selection, because the names do not carry across.