December 2014

Annex A

Using Earth Observation to produce indices of habitat condition and change

To find out more about JNCC visit

Annex A

Title of project: / Using Earth Observation to produce indices of habitat condition and change
Date and time for return of tenders: / Friday 9 January 2015 @ 1600 hrs
Contract Reference No: / C14-0171-0901
Address for tender submission: / 1 electronic copy to be sent to
PLEASE DO NOT SEND TENDERS DIRECTLY TOPAUL ROBINSON, ANNA ROBINSON, DORA IANTOSCA OR GORDON GREEN VIA THEIR PERSONAL EMAIL ADDRESSES, AS THIS WILL INVALIDATE YOUR TENDER
Tender responses must be less than 10 MB in size.
On receipt of your tender, you will receive an automated e-mail to confirm receipt by JNCC Support Co. If you do not receive this automated email, please contact, in the following order:
Sue Wenlock (00 44 1733 866880)
Chris Downes (00 44 1733 866877)
Contacts for technical information relating to this project specification: / Paul Robinson
Joint Nature Conservation Committee
Email:
Tel: 01733 866867
OR
Anna Robinson
Joint Nature Conservation Committee
Email:
Tel: 01733 866851
Contact for any queries regarding the tendering procedure: / Dora Iantosca or Gordon Green
Finance Team
Joint Nature Conservation Committee
Email:
Tel: 01733 866894 or 01733 866806
Proposed start-date: / w/c19January 2014
Proposed end-date: / 27 March 2015

1.Joint Nature Conservation Committee

The Joint Nature Conservation Committee (JNCC) is the statutory adviser tothe UK Government and devolved administrations on UKand international nature conservation. Its work contributes to maintaining and enriching biological diversity, conserving geological features and sustaining natural systems.

Our role is to provide evidence, information and advice so that decisions are made that protect natural resources and systems. Our specific role is to work on nature conservation issues that affect the UK as a whole and internationally:

  • advising Government on the development and implementation of policies for, or affecting, nature conservation in the UK and internationally;
  • providing advice and disseminate knowledge on nature conservation issues affecting the UK and internationally;
  • establishing common standards throughout the UK for nature conservation, including monitoring, research, and the analysis of results; and
  • commissioning or supporting research which it deems relevant to these functions.

2.Project Aims

To understand the potential and feasibility of using parameters calculated from Earth Observation (EO) data to detect changes that can occur in habitatsthat may be significant enough to affect ecosystem function or biodiversity value. These changes may occur in habitats either through direct management intervention (e.g. artificial improvement of a grassland), or through longer term natural processes (e.g. scrubbing up of a heathland).

3.Project Background

There are a series of land interventions that act to affect either landscape features, such as hedges or field margins, or habitats and their condition in order to achieve specific biodiversity or ecosystem service outcomes. Mechanisms using such interventions include:

  • Nitrate vulnerable zones
  • Catchment sensitive farming
  • Nature Improvement Areas
  • Voluntary farmer’s schemes
  • SSSI management
  • Agri-environment schemes
  • Grant schemes

All such mechanisms incur costs of paying grants to land owners or more direct costs of land management and so require an element of monitoring to ensure that the costs are achieving the desired impact. Much of this sort of monitoring currently involves some element of manual aerial photography interpretation, but also requires site visits and so is a time consuming task. This work is therefore looking to assess if automated processing of remote sensed data can target fieldwork, and therefore decrease the amount necessary, as well as providing the potential to produce measures of change in biodiversity or ecosystem services values over larger areas (landscapes, regions, countries). It will also assess if it is possible that such data could form part of a shared service enabling different organisations to utilise it for their own differing requirements.

The National Forest Inventory, which is run by the Forestry Commission, contains an operational system that uses EO data to assess changes in quality of woodland habitats in order to locate woodland loss. It can be regarded as a working example of the principles that this project will investigate in habitats other than forest. The NFI tri-annual processing of high resolution satellite data across the UK produces an output that is relevant at a local site scale (as it enables targeted, detailed investigation of changes) as well as building up a national picture of change in the stock of the woodland resource and the rates of conversion to other land cover and uses. The system works because the stock of woodlands over 0.5ha has been mapped in detail, so all of these areas can be tested in an automated system with appropriate questions asked of the remote sensed data because the map provides the context within which the automated system can work. In this case the system can look for areas of change in vegetation productivity which is likely to represent felling of trees, wind-blow or the conversion of forest to urban or other uses. The principles that can be extracted from this example are:

-With a detailed spatially accurate digital inventory/map of a habitat it is possible to test for unexpected changes in imagery.

-The remote sensing is cost effective when it can be applied to a widespread habitat or feature (hedges, trees) consistently over large areas, providing outputs that can be simply validated, e.g.: by visual inspection of aerial photography, or that significantly reduce the field effort needed.

-The significant changes (including transitions to other habitats, types of management that will influence condition) or condition states of a habitat are understood and information that can be derived from satellites is likely to be able to detect the type and scale of change, even if it does it in a completely different way to current field based techniques. This can enable efficient use of field resource as it can be focussed on high risk areas.

The second phase of the Making EO Work for UK Biodiversity project has published some findings on how EO data can be used to detect factors that impact on habitat condition ( Based on that work and others, we can say that the sort of changes that are likely to be detectable include things such as

  • Changes in vegetation productivity
  • Changes in wetness
  • Changes in the woodiness of the vegetation
  • Changes in vegetation structure
  • Changes in the amount of bare ground
  • The presence or size of linear features
  • Conversion of natural habitat to built structures
  • Eutrophication

There are many habitat types that are monitored in the field to see if the condition is changing, using approaches such as Common Standards Monitoring. It will be useful to consider the assessments made through such approaches but we also need to consider changes that will impact on other Ecosystem Service values as well as biodiversity value. The relevant services to consider in this will vary between habitat types, but those relating to water and carbon are always likely to be significant. Habitat condition can have a major influence on the movement of water and on the level of carbon sequestration for example. JNCC has already done some work on relationship between high value service and habitats, which can be found in two reports on using a spatial framework for ecosystem service analysis. Reports for these two projects are on the JNCC website (

A further piece of work that is relevant to this project is the development of the Crick Framework ( which provides information on the potential for using EO methods to map the extent of certain habitat types. Currently the Framework contains detail regarding mapping of BAP Priority Habitats and Habitats Directive Annex 1 habitats. Outputs from this project will start to populate the Crick Framework with information on the use of EO methods to assess habitat condition.

4.Project Objectives

This work intends to test if the principles used in the NFI can be extended to a wider suite of habitat types. Given the time available for this project it will not be possible to carry out a comprehensive analysis of this area. It will therefore be necessary to narrow down the selection of habitat types to be considered in more detail. In considering the grouping of habitats to consider in further detail, please note other existing related work in order to prevent duplication of effort. There are already other initiatives which have investigated relevant techniques in some habitats, for example CEH has looked at lake Eutrophication, recent studies in Wales have looked at measuring peat based habitats and work is just starting that will explore how radar data can be used to monitoring heathland burning. This study should not focus on woodlands because this is already covered by the NFI.

The process would follow this general approach:

  1. Select a limited number of habitat types to focus on. A review should be carried out to categorise how these habitats can change, either naturally or through management intervention, in a way that will affect ecosystem function or biodiversity value (e.g. a heathland scrubbing up or a grassland being artificially improved). The rates of change should be considered in order to assess how frequently assessment should be carried out. There should also be a review of how we currently assess condition and function of habitats, in order to help to develop this understanding of significant changes and how they can be measured.
  2. Understand how EO data can help us to identify the changes in land use and land management (or lack thereof) identified in the previous step. Overlap and covariance between the various measurable parameters must be considered within this step.
  3. Plan how relevant EO data could be systematically analysed and what data would be required for such a system, including what types of EO data (optical/radar/LiDAR, aerial/satellite) as well as other data.
  4. Draw up a detailed plan for a practical testof how the project findings can be carried out. This plan could then be used to explore some of the findings based around one group of habitats such as grasslands, as a pilot. Such a pilot would be at a relatively small geographical scale to keep data costs down and because of the limited time available and suggestions for suitable sites with appropriate data could be supplied by the steering group members (carrying out the pilot of the test is optional and should be costed separately)

5.Project Objectives: Detailed tasks

The following section gives more explanation around the four steps of the approach described above. The points below are not intended to be proscriptive but rather to supply additional information and outline the sorts of questions that are likely to be relevant.

Step 1: Understanding change:

Once the habitat groups have been prioritised to go into further detail, then for each of the selected groups, possible changes that can occur in each one will be considered. This will include the likely timeframes over which the change could take place and how the scale of that change may affect function or biodiversity value. Groups to be prioritised could be general habitat groupings where the fine classes within them have fairly similar management or transitions effecting them (for example, different types of grassland may be similarly affected by scrubbing, grazing, fertilising, drying) or narrower habitat classes that are widespread or reasonably extensive and so an efficient EO method would be likely to help.

Step 2: Can EO detect the change at a relevant scale:

Once the changes are understood then it should be asked whether the change could be identified from EO data. If so, what data would be required, at what resolution (geographical and temporal) in order to detect the change in a way that would be relevant to management? The practicality of any ideas explored will have to be investigated, so in the case of solutions that require multiple data sources per year, how likely might that be, given the sensors that could provide that data?

The review of current assessment techniques will be useful in understanding significant changes in condition. However, it should also be considered if EO can provide any additional indication of habitat change that cannot be detected through traditional assessment methods. Whilst the review is likely to provide useful insight into possible methods it will not provide all of the answers as EO data is likely to enable novel techniques to be used in identifying changes to habitats.

The main focus of this work needs to consider data from broader swath instruments, which enable simple processing over large areas, with multiple passes. Sentinel data, both active and passive sensors, and Landsat will be considered as a priority as it is free at the point of use. Other high resolution instruments should not be ruled out, however, as they could still provide cost effective solutions in some circumstances. Aerial data from APGB, which includes CIR and stereo imagery on a 3 year refresh cycle, will also be considered. Alternative aerial sensors, such as LiDAR will also be considered but we don’t have a means of deploying universally, data will be limited to where EA/SEPA fly, so is there a way that they can be used?

The EO data is likely to be analysed through producing a series of parameters. What would those parameters be and for what factors leading to change could those parameters be informative? These parameters will often not be direct measurements of biophysical characteristics on the ground, so we need to understand the relationship between the image analysis and conditions on the ground. So this step needs to look not only at the likelihood of differences on the image showing real change on the ground, but also at the likelihood of real change on the ground happening without showing detectable differences on the image. The aim is therefore to try to estimate repeatable measures from satellite imagery, with modelled uncertainty built in.

It should be assumed that the analysis can focus on parcels of land that have already been classified to the particular habitat grouping being considered, e.g. as documented in an existing habitat inventory.

Step 3: Plan how a system could operate:

Any system that would provide the parameter outputs will be closely related to, or part of the hub and spoke model that is currently being investigated by Defra. The hub will provide access to available satellite data, including that provided by Copernicus and possibly others, such as NovaSAR. It may also include the provision of certain generic products, such as the parameters being investigated here. The spokes will be a wide variety of analyses that can be carried out using the data available from the Hub.

This aspect of the work would have to consider each parameter proposed in step 3 in order to assess how it could be calculated and delivered to best meet user needs. Further questions to evaluate will be around whether the products will need some form of calibration and how that could be achieved. Also, do you need absolute validation somewhere such as from an ECN site, or just relative change or values at each location.

There is no requirement here to create a detailed plan of a system to deliver all of this, but the project should create a clear analysis of the processing that would be required for delivering each of the proposed parameter outputs. That analysis will enable planning in the future as it will show the data that would be required and the analytical processes that would be used.

Step 4: Practical test:

This step will produce detailed recommendations on how a practical test of the project findings can be carried out. If time and resources are available then a pilot of the proposed test method can be run for a particular habitat grouping. A test area would have to be chosen as it would not be feasible to run a national test.

There are a number of digital inventories that could be used for this test, such as the map of Norfolk created in Phase 2 of the MEOW project ( The Wales Habitat Map (Gwylio), Native Woodlands Survey of Scotland and Natural England’s Priority Habitat Inventory. The contractor will need to suggest options of where there could be good field data available to validate the processing.

This test would aim to show specific examples of how this approach could identify change in habitat extent or condition and in landscape features. It will test that the data available can be used at the appropriate scale in order to work with the digital inventories and to test proposed processing requirements.

The pilot running of this practical test should be separately costed and a detailed description given as to how this could be delivered in the timescale.

6.Outputs

The minimum outputs required for the project are:

Rapid conceptual framework that can be added to Crick Framework, ie: habitat groupings, types of transition to other habitats or land use, condition, and most likely parameters and data sources. This will not be an exhaustive list, but will capture all of the findings from this project.

Recommendations on habitat groupings to investigate in detail (interim output) based on knowledge of extent of habitat and of likelihood that the principles will work well.