Service Description: This map represents modeling efforts to describe relative risk for the establishment of wild carrot (Daucus carota ssp. carota) in central Oregon. Field observations were collected from partners within the study area and field surveys were conducted in 2022 and 2023. The map was built using the random forest modeling machine learning algorithm. The model unified information derived from 75 known observations of wild carrot with an array of raster explanatory data describing topography, climate, soil, distance to roads, and landcover class. The random forest model was effective at discriminating between observations and the combination of 1000 random background and 250 known absence data points (AUC for out-of-box prediction = 0.911). We identified cutoff values within the raw out-of-box random forest ‘probability’ prediction to yield categories that prioritize different error types (false positives vs. false negatives), using the precision-recall f-measure with varying values for alpha. This yielded five error-based categories that were tuned to the model’s unique error-structure. The categories labeled as moderate to very high risk highlight the most important areas to monitor for wild carrot invasion.
Service ItemId: 8ea14cc911db432691411c8f7e28cd09
Has Versioned Data: false
Max Record Count: 2000
Supported query Formats: JSON
Supports applyEdits with GlobalIds: False
Supports Shared Templates: True
All Layers and Tables
Layers:
Description:
Copyright Text:
Spatial Reference: 102100 (3857)
Initial Extent:
XMin: -13596714.9341613
YMin: 5469816.93474015
XMax: -13327905.0588959
YMax: 5691121.61653193
Spatial Reference: 102100 (3857)
Full Extent:
XMin: -13535289.1489034
YMin: 5470412.0863806
XMax: -13388905.4360694
YMax: 5683530.45542541
Spatial Reference: 102100 (3857)
Units: esriMeters
Child Resources:
Info
SharedTemplates
Supported Operations:
Query
ConvertFormat
Get Estimates