All Collections
Tools & Features
How to create a UK Habitat - Best Guess Baseline
How to create a UK Habitat - Best Guess Baseline

Create instant and detailed insight of the landscape with our digital UKHabitat - Best Guess Baseline. In partnership with UKHab Limited.

Simla Rees-Moorlah avatar
Written by Simla Rees-Moorlah
Updated over a week ago

To view video guidance and instruction on how to use this service please click on this link.

In partnership with – UK Habitat Classification


Alongside our partners, Land App has created an automated system for delivering a UK Habitat - Best Guess Baseline using the UKHabitat Classification. This “Best Guess Baseline” is a single data layer which predicts which habitats make up a given area of Great Britain by combining insight from multiple third-party datasets. The result is to provide users with near-instant access to highly granular habitat data for instant insight. Any boundary file can provide a total area and location of existing woodland, grassland, urban and cropland features can be established.

This system streamlines the preparation time for site visits such as habitat assessments for Biodiversity Net Gain, or similar. We do not recommend that this data be used as a replacement for an ecological survey.

However, this data can provide a proxy for existing habitats if a site visit is not possible or necessary (e.g. in the early planning stages of projects).

If you have a GeoJSON File, you can now request a quote directly from the Land App. This allows you to easily get accurate quotes for the OS MasterMap license and the UKHabitat Best Guess Baseline. Click the link above to access your quote.

Data Sources and Licensing



Available for full use in Land App, or as a download in .GeoJSON .kml .shp .dxf or .gz - Other file types are available upon prior request.


To create the algorithm, the project team has made a series of logical assumptions that underpin the workflow. These assumptions reflect the best of our knowledge at present. They are, however, subject to change due to feedback from the steering group and


  • OS MasterMap is the most accurate GIS layer in Great Britain for geometry for all features, and thus all vector shapes at a basic level will use their geometry. [Note: there will be instances where OS MasterMap shapes are split if a subsequent feature for an additional data layer intersects, but the combined shape will always follow OS MasterMap].

  • We do not assume that either the Priority Habitat Inventory, CROME (Crop Map of England) 2020 or Earth Observation data is correct at both a geometry level.

  • Earth Observation Data provides the most up to data insight of in-field practice for classifying grassland, woodland and cropland.

  • OS MasterMap provides authoritative information for all hard infrastructure and all waterways/water bodies. For all other features, the algorithm assumes that OS MasterMap could be wrong, and thus will take attributes on a series of comparisons with other data sets.

  • Some habitats can be determined using a combination of OS MasterMap attributes alone; theme, descriptivegroup, descriptiveterm and make. This is usually Level 2 of the UKHab Classification (e.g. grassland), however, there are instances where Level 5 is possible due to a 1:1 look-up between OS MasterMap and UKHab (e.g. OSMM buildings → UKHab u1b5 buildings).

  • The order in which the descriptiveterm is written DOES NOT necessarily mean the first feature is the dominant feature (e.g. if descriptiveterm is Coppice Or Osiers, Scrub, Nonconiferous Trees, Coniferous Trees could mean c1d6 - rotational coppice, h3 - dense scrub, or w - woodland. In these instances, the algorithms learning makes an informed decision off training data the likelihood of dominant habitat. We will be regularly reviewing these instances and will avoid - where possible - “over-assigning” based on assumptions).

  • In both “natural” and “agricultural land” polygons from OS MasterMap, we have relied on additional data sets to provide habitat assignment. For example, in England agricultural polygons get cross-referenced to other data sets, including the Priority Habitat Inventory and National Forestry Inventory picking up their habitat code if applicable.


The data created is only as accurate as the third-party data that makes up the workflow.

  • Land App and partners cannot accept any liability for the accuracy or completeness of the information generated.

  • The data is two-dimensional and therefore cannot fully represent the three-dimensional

    world. For example, tree canopies overhanging roads will not be counted and in these instances, the infrastructure features will take precedence over natural features.

  • It is recommended that the user conduct elements of ground-truthing before using the

    data for any legal or policy changes.

  • Where there is no data to truly assign a UKHab code, the model will not assign any thus leaving the polygon blank.

Earth Observation

Cropland and grassland detection

Using Earth Observation data temporally assesses each field on its probability of being cropland and grassland, also identifying when the ground was last bare (e.g. cultivation).

Attributes include:

  • "Percentage" shows the percentage of pixels confirmed by the classification (e.g. satellite data detects 85% of the field was grassland). This can act as a confidence level.

  • Cropland "bare_date" (e.g. "11-2019") shows when the model detects bare soil (e.g. cultivation). This allowed the distinction between permanent and temporary grassland, as well as the detection of high-risk fields (winter bare soil) and the use of winter cover crops.

Vegetation Object Model

The LIDAR derived Vegetation Object Model (VOM) is a raster product produced as part of the Environment Agency’s “Keeping Rivers Cool” project. It is an attempt to identify riparian tree cover and the opportunities for tree planting to increase future shading of streams & rivers.
Land App converts this raster data into vector as part of the Best Guess Baseline, providing additional information on location of vegetation >2.5m tall.

Note from producer: "The result is a raster product where each pixel represents the height of top of canopy above ground, for all classified vegetation objects above a threshold of 2.5 metres. The data production is fully automated, with no manual QC and editing of the output, other than visual checks. Because of the process to classify objects based on proximity to features within OS mapping, there could be some misclassifications of objects not included in the OS mapping (especially static caravans, shipping containers, large tents/marquees, coastal cliffs and new buildings constructed directly under tree cover). This is the first release of this dataset, the quality of the production methods will be reviewed over the next year and improvements made where possible."

UKHabitat Best Guess Baseline including the Vegetation Object Model overlaid on LiDAR from the Environment Agency.


Public Sector Geospatial Agreement

If you are a public body, you may be entitled to Ordnance Survey MasterMap for free (but are still required to pay for the UKHabitat Best Guess Baseline).
You can check whether you are registered here under the Public Sector Geospatial Agreement. Recommended resources:

All materials relating to the UKHab Classification can be downloaded from - this includes the lookup tables, habitat descriptions and style sheets for ArcGIS and QGIS. This is accessible once you register for a free account.


Pricing is calculated on the hectarage of continuous area, with incremental discounts applied as the area increases.



hectare (ha) band

Band price per ha

Band A


0 - 1000


Band B


1001 - 10,000


Band C


10,000 - 100,000


Band D







£/ha equiv

Small Farm




Big Farm








Farm Cluster




County or Region




*Please note a valid Ordnance Survey MasterMap Topography license will be needed at the time of purchase.


Future Developments

It was not possible to include all authoritative habitat data layers into this workflow. As a result, some users may find sources that either disagree with, or enhance, the classification. We will endeavour to keep our system relevant and up to date. Our baseline data will be constantly evolving and developing to meet current requirements and in response to feedback. If you have any suggestions for improvements, please email


Land App is grateful to all partners who have contributed to the creation of this workflow.

  • UKHab Ltd for the creation of the UKHabitat Classification, the underpinning language for the data, alongside technical support on habitat translations.

  • Farming and Wildlife Advisory Group (FWAG SW) for guidance and steering of the requirements for Land Managers, and guidance on habitat translations.

  • Ordnance Survey for data provision and technical support.

  • 1Spatial for technical support.

  • Land App users who have provided feedback to the data set.

  • Surrey Wildlife Trust for testing and improvements of the workflow.

How to run a UK Habitat Best Guess Baseline

Please note that to run a UK Habitat Best Guess Baseline you need to have purchased Ordnance Survey MasterMap data in the Land App first. This guidance shows how.

Also note the video guidance shows "Buy Data" button this has been changed to "Download Data"

Click on New, Download Data in the top left of your map.

Select UK Habitat Best Guess Baseline

Select your previously purchase Ordnance Survey Mastermap data.

Input your Plan name

You can use an SBI to clip the Ordnance Survey MasterMap data if needed.

The cost for the purchase will show and to proceed click on Buy Now.

If in a professional subscription you may need to input a job code and billable/non billable status.

Click on Buy Now again if needed.

Go back to your map.

You will receive an email to confirm that your data is processing with an estimated time to completion and a second email confirming completion. At this point please refresh your map.

A new plan will have been created for your UK Habitat Best Guess Baseline.

You can then interrogate the data, run tables and print.

Did this answer your question?