ArcGIS REST Services Directory
Home > services > Planning > BLM_UT_GRSG_ARMPA_MapService (MapServer) > Winter Habitat (Polygon) API Reference
JSON

Layer: Winter Habitat (Polygon) (ID: 7)

Parent Layer: Seasonal Habitat

Name: Winter Habitat (Polygon)

Display Field: season

Type: Feature Layer

Geometry Type: esriGeometryPolygon

Description: The process of creating the seasonal habitats layer is described below.1. Creation of statewide general habitat model for GRSG (2016)In 2016, researchers at Utah State University (USU) developed a statewide general sage-grouse habitat map using a database of hundreds of lek locations and more than 20,000 GRSG telemetry locations collected statewide from 1998 – 2014. The map depicted habitat suitability on a scale from 0 to 100 at 1 km spatial resolution, based on comparing environmental (vegetation, topography, soils, climate) and anthropogenic (developed land cover, road density, powerline density) conditions at active lek and sage-grouse use locations versus inactive lek and random background locations statewide. Because multiple telemetry locations were often associated with a single brood-rearing or non-breeding bird, the median values of environmental and anthropogenic variables at these telemetry locations were used in the model. A random forest model was used to create the map (Breiman 2001, Cutler et al. 2007). Random forests is a highly accurate non-parametric classification technique that predicts the probability of an outcome (in this case, habitat vs non-habitat) by averaging the results of many classification trees, each of which is trained on a random subset of the available data. The general habitat map was reclassified into ‘habitat’ and ‘non-habitat’ classes such that habitat areas captured 99% of all sage-grouse use locations. These general habitat areas were used as a mask in order to constrain preliminary predictions of seasonal habitats, described below.2. Create preliminary seasonal GRSG habitat models (Jan - May 2017)Telemetry locations in the database were classified into three seasonal habitat types based on time of year and type of GRSG use. Breeding habitat was defined as areas used by GRSG engaged in lekking, nesting, and early brood-rearing, from March 1 – June 14. Summer habitat was defined as areas used by brood-rearing and non-breeding GRSG from June 15 – August 31. The June 15 cutoff date between breeding and summer use locations was selected based on the temporal distribution of nesting and brooding use locations. Winter habitat was defined as areas used by non-breeding GRSG from November 1 – February 29. As in the general habitat modeling approach, environmental conditions at annual brood-rearing or non-breeding locations associated with the same bird were measured as medians over the multiple locations.Seasonal habitats were modeled using the same predictors as the general habitat model, with the addition of distance to leks due to its association with breeding habitat. A random forest model was used to estimate the suitability of general habitat areas statewide (from step 1 above) for breeding, summer, and winter use. For each seasonal use class, a suitability threshold was selected such that 85% of all seasonal use locations were captured in the resulting seasonal habitat map. This resulted in models that were neither overly restrictive nor overly liberal. To reduce the 'salt and pepper’ effect of isolated or scattered habitat pixels, a 3x3 km smoothing window was applied to each of the seasonal habitat layers, assigning the majority value (habitat or non-habitat) to the center pixel. 3. DWR biologists reviewed seasonal models (April - June 2017)An overview of the general and seasonal mapping methodology and preliminary maps was presented to GRSG biologists and managers from the Utah Division of Wildlife Resources (UDWR), Bureau of Land Management (BLM), and Forest Service (FS) at Utah State University in Logan on 4/21/17. Feedback at the meeting led to a few minor changes to the seasonal mapping methods. Because the breeding seasonal use model was not picking up areas around all active leks, distance to leks was dropped as a predictor variable from the seasonal habitat random forest model, and a 3 km buffer around all active leks was manually included in the breeding habitat model. Updated seasonal use models were sent to UDWR on 5/17/17, where they were made available for review by biologists with local area knowledge. An ArcGIS Online webpage was used to share the models with biologists, and allow for them to provide recommended additions / deletions to areas captured by the models. Accompanying the spatial data was an 8 minute webinar communicating the modelling procedure. UDWR returned updated seasonal use models with biologists’ comments, additions, and deletions to USU researchers on 6/6/17. Most but not all areas in the state received substantive feedback and comments from UDWR biologists. 4. Additional review of seasonal use models and final edits performed (July – August 2017)USU researchers reviewed biologist edits on 7/13/17. Most of the areas flagged to be added/removed from the seasonal use models seemed appropriate and helpful. But, there were a few areas UDWR biologists identified to be deleted from the models that USU researchers considered questionable. These areas were identified and shared with UDWR managers on 7/13/17. UDWR managers indicated a desire to have another face to face meeting to make a final determination on areas to include / exclude from the seasonal use models. This meeting occurred on 8/8/17, at which it was determined that it would be preferable to have the final seasonal habitat products reflect both use and potential suitability, as opposed to only areas of known use. This led to rejecting some areas flagged for deletion by biologists, as biologist comments indicated they were conceptualizing the map as primarily a use map only. Numerous small edits were made to the seasonal use layers, including several edits to include seasonal use locations not captured by preliminary models. Finally, all single, isolated habitat pixels were removed from the map.ReferencesBreiman, L. (2001), Random Forests, Machine Learning 45(1), 5-32.Cutler, D.R., T.C. Edwards, K.H. Beard, A. Cutler, K.T Hess, J.C. Gibson and J.J. Lawler. 2007. Random forests for classification in ecology. Ecology 88(11):2783-2792.This dataset was used in the preparation of the Greater Sage-grouse (GRSG) 2018 DEIS planning maps and acreages initiated by Secretarial Order (SO) 3353 of June 7, 2017 and the Notice of Intent (NOI) of October 11, 2017.

Copyright Text: Recipient Principal Investigator/Project Manager: Terry A. Messmer, Jack H. Berryman Institute, Utah State University, 5230 Old Main Hill, Logan, Utah 84322-5230, Phone 435-797-3975, Fax 435-797-3796, E-mail terry.messmer@usu.edu Project Co-PIs and Staff: Dave Dahlgren, Eric Thacker, Kris Hulvey. Michel Kohl (Post-Doc), and Ben Crabb (GIS staff)

Default Visibility: false

MaxRecordCount: 1000

Supported Query Formats: JSON, geoJSON, PBF

Min Scale: 5000000

Max Scale: 0

Supports Advanced Queries: true

Supports Statistics: true

Has Labels: false

Can Modify Layer: true

Can Scale Symbols: false

Use Standardized Queries: true

Supports Datum Transformation: true

Extent:
Drawing Info: Advanced Query Capabilities:
HasZ: false

HasM: false

Has Attachments: false

HTML Popup Type: esriServerHTMLPopupTypeAsHTMLText

Type ID Field: null

Fields:
Supported Operations:   Query   Query Attachments   Query Analytic   Generate Renderer   Return Updates

  Iteminfo   Thumbnail   Metadata