# Release Web March 2026 (Field Trials Design, Generative Synthetic Yield Maps, Farm View Stats, MCP) ## Generative Synthetic Yield Maps You can now create a **yield map without a yield monitor** - useful when you only know the total or average harvest result for a field. The new **"Generate Synthetic Yield Map"** option is available directly from the Yield Data section and supports two workflows:

Generative yield map displaying the dry mass attribute and a legend with the statistics.

### Generate Synthetic Yield dataset !["Generate Synthetic Yield Map" option in the Yield Data section dropdown menu.](/files/64mqFQBioCN9S2OcLDBU) ![Generate Yield Map popup with crop type, harvest month, year, and average yield value inputs.](/files/BfFurwaA5WHMWAAv0GXJ) Start from scratch by entering your crop type, total or average yield value (with the appropriate unit pre-filled based on crop and area setting), and harvest year. GeoPard builds a spatially distributed yield dataset across the field based on deep knowledge of a field: soil, water, topography, field history, weather and in-season crop development. We've written earlier about synthetic yield maps with illustrations. ### Reconstruct Partial Harvesting dataset Generative model, works great to **make a complete yield dataset out of partial** or low quality non-calibrated noisy harvesting dataset.

Reconstruct Partial Harvesting dataset — Reconstruct Yield for the northern part of the field

### Regions supported All regions supported by GeoPard, including areas with several harvesting seasons per year. ### Crops supported * Corn, Corn Silage, Soybeans, Rapeseed, Canola, Barley, Rye, Wheat, Sunflower, Cotton, Rice, Oats, Sorghum ## On-Farm Trial Layouts Design GeoPard now supports field trials as a dedicated data layer alongside Zone Maps. Agronomists can draw trial plots directly on the Zones Map page, assign products or treatments to each block, and visualise trial results overlaid on the field map. ### Why geospatial trial design matters Unlike trial tools that only store treatments in a table or CSV, GeoPard keeps trial layouts directly in a geospatial field system. Each plot, strip, or block is linked to its exact field position and can be analyzed together with other spatial datasets. This means you can go beyond simple treatment setup and directly evaluate trial outcomes based on: * **yield performance** * **soil and zone characteristics** * **topography and other field variability layers** * **economic efficiency and return by treatment** Because the trial exists as a spatial layer, GeoPard can connect trial design with real field conditions and help turn on-farm experiments into practical agronomic and economic conclusions. ### **Key capabilities:** * **Draw trial plots** on the Zones Map, with the option to reuse trial locations when creating a new zones map * **Visualise trials** on the field page whenever the associated zones map is selected * **Export trials merged with the zones map** into a single application-ready file {% @arcade/embed url="" flowId="A26Ph3BdcYbkzePxVpzO" %} ### Trial Design Layouts Supported GeoPard supports several trial design patterns, each suited to a different type of on-farm experiment.

Trial Type	Details	Cover image
Plot Grid (RCBD)	A classic randomized complete block design for statistically stronger treatment comparisons. GeoPard creates a rectangular grid where each block contains all treatments in randomized order. Best for structured trials where you need cleaner, more publishable results and relatively uniform conditions within each block.	/files/yy2GUjzJxafSHWPndPab
Split-Plot	A two-factor trial design with a second treatment dimension inside the main plots. Best when you want to test interactions such as the same rates across different hybrids or varieties. This is practical when one factor is harder to change in the field and the second factor needs to repeat consistently inside it.	/files/CxJRGxiSQuiAS9MiTk0c
Zone Strip	Creates strips inside management zones, following zone boundaries and clipping to fit each zone. Best when you want to compare treatments within specific soil types, productivity zones, or other agronomic zones.	/files/Q9SfClM4fAzvjyhwwgLT
Field Strip	Creates full-length strips across the entire field, crossing all zone boundaries. This is a practical setup for real farm equipment and easy application, especially for large-scale comparisons under varying field conditions.	/files/A4ewn942LmKeQfUM4WUV
Checkerboard	Creates alternating treatments across a grid pattern, spreading each treatment throughout the trial area. Useful for exploratory trials and for fields with strong spatial variability.	/files/pZICOF4K84diajN8rSyT

GeoPard now also supports **Split-Plot** for two-factor trials. Use it when you want the same rates repeated across different hybrids or varieties. ### Trial Design Parameters GeoPard lets you define the trial layout with flexible settings:

Trial Parameter	Details
Replications	Sets how many times the treatment set is repeated. More replications improve comparison quality and reduce the impact of local field variability.
Angle	Controls the orientation of the trial layout. This helps align plots or strips with field direction, machine path, or operational preference.
Width	Defines the width of each strip or plot in meters. Usually matched to implement working width for easier execution in the field.
Buffer	Adds spacing between neighboring plots or strips. This helps reduce overlap and treatment interference.
Blocks	Defines the block structure used in designs such as RCBD. Each block contains the full set of treatments and helps account for field variability across the trial area.
Treatments	Defines the treatment rates or values to compare, such as seeding rates, fertilizer rates, or product doses. Each treatment can be enabled, edited, color-coded, added, or removed directly in the setup.
Treatment Difference	Lets you quickly generate treatment levels around a center rate by percentage. For example, a 10% setting creates evenly spaced lower and higher treatment rates around the base value.
Shuffle Layout	Randomizes treatment placement within the selected trial design. This helps reduce positional bias and makes comparisons more robust.

## Farm View: Field Statistics Panel & Quick-Action Navigation Farm View now includes a dedicated **statistics panel** that gives you instant access to key field metrics: area, mean, minimum, maximum, relative variability (coefficient of variation, CV%), and [Heterogeneity Factor](https://geopard.tech/blog/use-of-heterogeneity-factor-in-precision-agriculture/) for zones. Use it to **benchmark fields across soil, yield, topography, imagery, and as-applied datasets**, and quickly identify where to focus scouting, input decisions, and next actions first. The panel also includes direct **quick-action buttons** - launch a **Sampling Plan** or jump straight into **Create Equation Batch** - so you can move from analysis to action. The panel is collapsible to keep the map view clean when you don't need it.

Farm View with statistics table showing average, min, max, and variability of Yield, Soil, Imagery, Topography, As-applied/As-Planted Data layers. — Farm View with statistics table showing average, min, max, and variability of Yield, Soil, Imagery, Topography, As-applied/As-Planted Data layers

## GeoPard MCP: AI Agents for Data-Driven Farming [GeoPard MCP ](/geopard-tutorials/product-tour-web-app/geopard-mcp.md)introduces AI-agent workflows for data-driven farming by connecting GeoPard field data with MCP-compatible assistants. This allows agronomists, farmers, and growers to securely work with their agronomic and geospatial context through AI, speeding up analysis and helping transform field data into actionable recommendations: starting from [custom logic compaction-based tillage decisions](/geopard-tutorials/product-tour-web-app/geopard-mcp/optimal-tillage-depth.md) to [tailored vegetation analysis](/geopard-tutorials/product-tour-web-app/geopard-mcp/explain-vegetation-patterns.md) and [yield-data cleaning/calibration](/geopard-tutorials/product-tour-web-app/geopard-mcp/clean-and-calibrate-yield-data.md), and many more. {% @arcade/embed url="" flowId="bibREyFI1hqQvPk7rBFq" %} ## Batch Equation Maps: Per-Field Pixel Heading & AB-Line Support Each field card on the **Batch Equation Maps creation** page, now has an **"Heading"** control that opens a map popup. Inside, you can set the **pixel heading** for that specific field using the same tools available in single-field mode: a heading angle input, or draggable A/B points to define the application line. Generated pixels will follow the selected direction per field.

Batch Analytics preview showing "Adjust Settings" panels with the Heading control for each field.

Equation map result showing rectangular pixels rotated to match the heading line direction.

## Draw Manually: Generate Masked Data/No Data Zones from a Dataset The **Draw Manually** workflow now includes a **"Create Data Zones"** control in the instrument panel. Instead of drawing polygons by hand, you can select an existing dataset and GeoPard automatically generates two zones: * A **data zone** mask covering all areas where measurements exist (clipped to the field boundary) * A **no-data zone** mask covering the remaining part of the field Useful when working with partial-coverage datasets such as soil samples, lab results, or as-applied layers. {% @arcade/embed url="" flowId="WnDkoQhipPiudVZJJUF3" %} The **Data/No-Data instrument** also supports a **mask mode**, letting you merge computed data zones into an existing zones map rather than always creating a new one. Useful for incrementally updating field management zones as new sensor coverage arrives. {% @arcade/embed url="" flowId="5AmW9ONqdeJxstzDpn3u" %} ## Rectangular Pixel Support in Equation Maps Equation Maps now support rectangular pixels that match the actual width and length of your equipment. The GeoPard interface clearly indicates where to enter your equipment width and length.

Equation Map with rectangular pixels generated according to the width and length of the equipment.

## Order Planet Imagery Directly from the Map Ordering **Planet imagery** is faster now. Previously hidden in the three-dot menu, the order option is now a cart icon next to the acquisition date - one click to order directly from the field view.

Icon for ordering a Planet image displayed next to the imagery date in the Data Layers panel.

## Romanian Language Support The GeoPard platform is now fully available in **Romanian** 🇷🇴. All interface elements, labels, and navigation have been localised. Users can switch to Romanian from the **User Settings** menu.

GeoPard Settings page fully displayed in Romanian, with language set to "Română" in the Personalization section.

## Improved Zone Classification for Small Fields Zone classification on fields smaller than 10 hectares now produces accurate, high-quality results - the algorithm has been tuned to handle low pixel density gracefully on compact parcels. ## KML Import: Full Support for Multi-Polygon Fields Fields made up of multiple polygons (e.g., split by a road or waterway) are now imported as a single unified field boundary. ## Prescription Export for Trimble Displays Prescription maps can now be exported in formats **ready for Trimble field displays**. Both legacy and newer display families are supported: * **Legacy displays** (CFX-750, FMX, EZ-Guide 500) - Shapefile export with the `AgGPS/Prescriptions/` folder structure * **GFX displays** (GFX-350, GFX-750, GFX-1060, GFX-1260, TMX-2050) - Shapefile and ISOXML export with the `AgData/Prescriptions/` folder structure

Prescription Export for Trimble Displays — Select Trimble display format (new or legacy monitors supported)

The export wizard lets you choose the appropriate display type. The generated ZIP preserves the exact folder hierarchy required by each device - no manual file reorganisation needed on your USB drive. ## More Accurate Yield & As-Applied Maps The processing pipeline for **yield** and **as-applied datasets** has been significantly improved. GeoPard now uses **grid-based rasterization** instead of mathematical interpolation: * **More accurate** - each grid cell reflects a real sensor measurement, not an estimated value between points * **Better visual alignment** - the grid rotates to follow the machine's actual driving direction * **Faster processing** - substantially quicker than the previous interpolation method * **More stable** - eliminates a key source of out-of-memory errors on large fields ## Satellite Imagery: Coverage Statistics Now Available via API The satellite imagery API now returns **per-range coverage statistics** alongside index data. For each NDVI (or other index) colour band, the system calculates what percentage of the field area falls within that range - computed on cloud-free pixels only. ## All-Year Satellite Imagery for US Fields Satellite imagery for **US farmer fields** is now available across **all 12 months of the year**. This enables year-round analysis including cover crop monitoring, post-harvest assessment, and early-season planning - not just peak growing season. ## Bug Fixes & Improvements * Fixed web app crash when the Google Translate browser extension was active. * Fixed draw layer rendering below zone borders on Zone Maps - drawn polygons now always appear on top. * Fixed **John Deere Operations** query failing when the operation filter included a large number of fields. * Zone polygon boundaries now align perfectly with no gaps between adjacent zones, ensuring clean and reliable zone maps for every field. * Fixed broken timestamp display on certain raster data layers. * Fixed interpolation setting not being preserved correctly when cloning a zones map. * Fixed **Planet imagery** cloud-mask coverage filter not being applied in certain cases. * Fixed continuous range legend excluding the maximum value from the last classification bin. * Fixed **Farm View** auto-selecting the wrong "Recent Field Potential" map when names contained year ranges. * Field assets tree now loads datasets incrementally, reducing initial load time on data-heavy fields. * Data Layers preview panel now shows a loading indicator and keeps asset titles visible while scrolling. * Batch equation map previews now retry automatically on failure, with clearer loading state feedback. --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://docs.geopard.tech/geopard-tutorials/changelog-and-product-releases/release-web-march-2026-field-trials-design-generative-synthetic-yield-maps-farm-view-stats-mcp.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.