Land Cover-based Landscape Assessment Algorithms In the SmartConservation Model Robert Cheetham Avencia Incorporated
Avencia Introduction Clients: Local/State/Federal Government Non-Profits Academic Institutions Topic Domains: GIS Software Development in: Real Estate Urban/Neighborhood Redevelopment Natural Resources Crime Analysis
Presentation Objectives How does SmartConservation Model use Land Cover data for calculating landscape metrics?
What is SmartConservation? A method of identifying, evaluating and prioritizing landscape resources for their potential habitat health and viability. A weighted hierarchical model for evaluating arbitrary areas of land in Southeast Pennsylvania for their conservation value. TM
What is SmartConservation? A Decision Support System for assessing the ecological/conservation value of a given site: - Model - Web-based assessment tool - Data Sets – existing + derived maps - Green Infrastructure Plan
Some Guiding Principles 1.Score any arbitrary polygon in Piedmont region 2.Generate scores on demand 3.Use publicly available data (mostly) 4.Generate comparable scores 5.Make all assessments available to all users 6.Integrate data with processes and make available to organizations that dont have access to analysis tools.
Presentation 1. Overview of Model Components 2. Land Cover Adjustments 3. Land Cover Metrics
What is Smart Conservation? DataMethods+Metrics Site Assessment + Standard Weights
Partner Relationships - Manage the Project - Convene Science Forum committees - Develop methodologies - Acquire data sets NLT - Develop Software Components - Develop Algorithms - Collaborate w/ NLT on methodologies - Install and maintain software Avencia - Assess data sets - Develop assessment methodologies - Approve methodologies and weights Science Forum PASDA - Host web site and data - Primary Data Provider
Model Weights
Weighted Hierarchical Model
Plant Criteria
Site Score
Model Components - Mammals Mammals 13 % Potential Habitat Size 2 % Potential Habitat Shape 1 % Special Habitat 1 % Landscape Matrix 2 % Important Mammal Area (IMA) 2 % Habitat Structure Health (Deer) 1 % Potential Species Richness 2 % Potential Conservation Value 2 %
Model Components - Plants Plants 25 % Habitat Disturbance 7.5 % Human Invasive Plant Deer Community Diversity 2.5 % Plant Community Health 10 % Potential Habitat Size 2 % Potential Habitat Shape 1 % Landscape Matrix 2 %
Model Components - Birds Birds 13 % Landscape Matrix 2 % Potential Habitat Size 2 % Important Bird Area 2 % Interior Forest Habitat 2 % Habitat Structure Health (Deer) 1 % Potential Species Richness 2 % Potential Conservation Value 2 %
Model Components - Aquatics Aquatics 18 % Riparian Corridor Fragmentation 2 % 1st and 2nd Order 3rd, 4th and 5th Order 6th+ Order Watershed Land Cover 2 % Aquatic Community Health 2 % Location of Site in Watershed 1 % Potential Fish Species Richness 2 % Potential Fish Conservation Value 2 % NWI Wetlands 2 % Hydric Soils 1 % Floodplains 1 % Forested Water Quality 2 % Steep Slopes 1 %
Model Components - Herps Reptiles and Amphibians 4 % Potential Species Richness 2 % Potential Conservation Value 2 %
Model Components – Land. Eco. Landscape Ecology 8 % Hub Distance Hub Density Corridor Distance Rarity 19 %
Model Components Totals 100 % Directly derived from Land Cover data 20 % Indirectly derived from Land Cover data 27 % No relationship to Land Cover data 53 %
Presentation 1. Overview of Model Components 2. Land Cover Adjustments 3. Land Cover Metrics
Using the Tool
Land Cover Adjustments Adjustments can be made to the local land use in order to make the local land survey more accurate. PDF files are made available to generate maps for the site survey.
Presentation 1. Overview of Model Components 2. Land Cover Adjustments 3. Land Cover Metrics
Land Cover Metrics Indirect Metrics – MRLC land cover being used as a component, but is not the main focus
Potential Species Richness Mammals Fish Birds Herps
Potential Conservation Value Mammals Fish Birds Herps
Interior Forest Habitat Forested areas – edge = Interior Forest Rank based on size
Landscape Ecology - Merged Hubs Protected Lands PNDI Cons. Resource Locations Merged Hubs merge Before merge After merge
Modified Impermeability + Modified Impedance - = () BarriersConservation Value Hub Proximity
Relative Travel Cost
Preliminary Impedance
Hub Proximity Values Merged protected land sites with overlapping donut buffers of 1km and 2 km and values of 50% and 10 % respectively. Overlap region with cumulative proximity values
Conservation Resource Lands
Land Cover Metrics Direct Metrics – MRLC land cover is the primary component of the metric
Landscape Matrix: Preprocessing Develop Taxa-specific land cover grid
Landscape Matrix: on-the-fly Calculate Zonal Mean on site and in buffer (10 - Abs(SiteLCMean – 1MileLCMean)) * SiteLCMean / 10 (10 - Abs(5.625 – 5.284)) * / 10 = 5.433
Potential Habitat Size/Shape 1.Create taxa-specific land cover value coverages 2.Slice up with TNC land blocks 3.Remove noise 4.Convert to polygons
Potential Habitat Size/Shape 1. Look at largest contiguous, un-fragmented land cover that is good habitat for taxa group 2. Assign size score based on size thresholds 3. Assign shape score based on roundedness.
Watershed Land Cover Develop Aquatic Taxa-specific land cover grid
Watershed Land Cover 1. Buffer site by 2 miles 2. Using site as pour point, calculate upstream watershed 3. Clip to 2 mile buffer. 4. Calculate Zonal mean of aquatic land cover
Riparian Corridor Fragmentation Buffer Streams and Water body edges to 100m Convert to GRID using Stream Order
Riparian Corridor Fragmentation 1.Clip Land Cover to Site Boundary 2.Clip Site Land Cover to the Stream Corridor 3.Reclass using Aquatic Land Cover and site survey adjustments 4.Average the Land Cover Values 5.Calculate NeighborhoodVariety 6.Convert to a 0-10 scale and Average for corridors 7.Lather, Rinse, Repeat….
Neighborhood Variety In Out
Neighborhood Variety
Technology Challenges Web-based system is being accessed by multiple users simultaneously Limit the Number of Users Precalculate Limit Extent Queue requests for asynchronous calculation to preserve rapid feedback
Technology Challenges Software Licensing Partnering with PASDA Worked with ESRI for special licensing arrangement
Technology Challenges Changing Requirements – changing data, changing model Modular Hierarchical Use XML Trained users before system complete Versioning is still tough.
Technology Innovations Centralized geoprocessing using ArcObjects for multiple users Asynchronous geoprocessing queue Automatic error recovery features Dynamic ArcIMS map layer generation Store layer info and symbology in database Allows us to change datasets without changing ArcIMS services
Other Challenges We want it to be public but... We want to move it outside SE PA, but … SmartConservation is explicitly interpreted data, but … We want people to do field surveys, but …
Still to do…. SmartConservation Lite – no field survey Update MRLC – versioning Generalize model to other ecoregions – will require re-calibration of some criteria to account for differences in ecoregions. Develop a Generalized Spatial Metric and Model Documentation Protocol
Acknowledgements PASDA Conservation Science Partners PA DEP PA DCNR William Penn Foundation Delaware Valley Regional Planning Commission Pennsylvania Environmental Council Brandywine Conservancy ESRI
Questions Clare Billett – Natural Lands Trust – Robert Cheetham – Avencia –