Magic Map

Turn a point dataset into a smooth, zoned map of the whole field in one step.

Magic Map takes a point dataset (yield readings, soil samples, cluster counts) and estimates a value for every part of the field, not only where a sample was taken. It trims the points to your blocks, removes outliers, lays an even grid over the field and fills each grid cell from the samples around it. The result is saved as a new dataset; the original is never changed.

Magic Map is not in the Plugins drawer. It is a button on the dataset itself.

#Before you start

  • The dataset must be a point dataset with more than 10 points and at least one numeric header. The button is not shown for raster, line or polygon datasets.
  • You need to be the farm's Owner, an Administrator or an Editor.
  • Magic Map is for values that vary continuously across the ground. It does not make sense for categories such as variety names.

#Steps

  1. Open the dataset from the Datasets drawer and go to its Visualize tab (the sliders icon).
  2. Click Magic Map at the top of the tab.
  3. Optional: tick Filter dataset to block boundaries. A small map of your blocks appears with every block selected. Click blocks to select or deselect them, or use Select All and Deselect All. Points outside the selected blocks are left out and the finished map is clipped to those blocks. Several blocks are treated as one field. Without this option, Magic Map draws its own outline around the points.
  4. Under Header, choose the numeric header to map. Outliers are removed from this header, the surface is calculated from it, and the new dataset is coloured by it.
  5. Under Output Dataset, check the name of the new dataset. It starts as the original name followed by "- Magic Map".
  6. Click Create Magic Map. A progress bar shows the stages. Larger datasets take longer; a dataset of more than 5,000 points can take 2 to 5 minutes.

When it finishes, the new dataset opens on the map, already coloured into zones. It is saved at the top level of the farm's datasets, outside any folder. A message reports how many grid points were generated and how many of your points were used, and the same summary is saved in the new dataset's description.

From there it is a normal dataset: recolour it, filter it, download it, or use it in Variable Rate Mode.

#Advanced Settings

The defaults suit a first run. Click Advanced Settings to change them.

#Statistical Filtering

Outlier removal method decides which points are dropped before the map is built:

Method What it does
None (my data is already cleaned) Uses every point as it is. Choose this if you trimmed the data yourself, so it is not trimmed twice.
Highest Density Interval (HDI) (default) Keeps the narrowest range that holds 96% of the values, dropping the sparse tails. Works with any shape of distribution.
Two-Stage Extreme Sieve Removes extreme values first, then applies Adaptive IQR. Suited to data with many extreme readings.
Adaptive IQR An interquartile-range filter whose threshold adjusts to how skewed the data is.
Interquartile Range (IQR) Drops values more than 1.5 × IQR outside the quartiles.
Z-Score (2 standard deviations) Drops values more than two standard deviations from the mean. Assumes a roughly normal distribution.
Modified Z-Score Like Z-Score but based on the median, so extreme values affect it less.

Apply Physical Bounds Filtering sets hard limits that are applied before the outlier method: a Minimum Value, a Maximum Value, or both. Use it for values that cannot be real, such as a negative yield or a pH above 14.

#Grid Generation

Grid resolution (meters) is the size of each grid cell, from 1 to 1000 and 3 by default. A smaller number gives finer detail, a larger dataset and a longer wait. Below the box, Every.Farm shows the most cells the dataset's extent could hold at that size, and warns you when the grid would be very large. If a run fails or times out, use a larger cell size and run it again. Every Magic Map uses the same grid system, so maps made from different datasets on the same field line up cell for cell.

#Interpolation

The method is Inverse Distance Weighted (IDW): each grid cell takes a weighted average of the samples near it, with closer samples counting more. IDW Power (2 by default) controls how quickly that weight falls off. Higher values keep each cell close to the readings right around it, so highs and lows survive. Lower values average over more distant points, which smooths the map and narrows its range of values.

#Visualization

  • Zone classification sets how the result is split into zones. Jenks Natural Breaks (default) places the boundaries where the values naturally separate; Equal Interval divides the range of values into equal steps. If the result has too few distinct values for natural breaks, equal intervals are used instead.
  • Number of zones, from 2 to 10 (3 by default).

The new dataset has one column, named after the header you mapped.

You can change the colours and zones of the finished dataset at any time on its Visualize tab.

#Grid Alignment

Most farms can leave this section alone. Farm Center (optional) takes a longitude, latitude pair to use as the reference point for the grid. Give the same point for several datasets and their grids line up cell for cell across a large farm.

#Tips

  • More samples make a better map than more smoothing. Interpolation estimates between samples; it cannot add information that was not collected.
  • If the map shows a suspicious bullseye, a single extreme reading is usually the cause. Run it again with an outlier method switched on, or set physical bounds.
  • The Interpolator plugin does the interpolation step on its own, without the outlier removal.

Updated

Something missing or out of date? Tell us — the docs are updated with every release.