RIDEM/IMG_Impervious (ImageServer)

View In: ArcGIS JavaScript ArcGIS Online Map Viewer ArcGIS Earth ArcMap

View Footprint In: ArcGIS Online Map Viewer

Service Description: Task Overview: NV5 Geospatial updated the Rhode Island impervious data layer through semi-automated methods similar to that developed for the 2011 impervious layer and as accepted by RI Statewide Planning/RI GIS. Update methods used Rhode Island spring 2020, 4-band, 6-inch pixel orthophotography. The State of Rhode Island determined that only real changes between the 2011 and 2020 imagery should be mapped. Errors in the data set were not corrected (unless they were directly associated with a real change) as this would skew change analysis. The minimum mapping unit used for this data layer was 800 square feet. The update process began by creating a machine learning model based on the 2011 impervious data layer. The model was trained on Rhode Island spring 2011, 4-band, 6-inch pixel orthophotography and 2011 2 feet pixel impervious data layer. The machine model is a modified UNET convolutional neural network model. After the machine learning model was trained using 400,000 examples the model was applied to the 2020, 4-band, 6-inch pixel orthophotography. Then changes were detected between the 2011 impervious data layer and the 2020 impervious data layer using a semi-automated process. The changed area was then inspected by a photo interpreter to ensure greater than 95% accuracy and proper look and feel of the data. After the changed area was inspected it was then used to update the 2011 impervious data layer. Areas that were unchanged have not been modified. Validation Overview: For Blocks 1-5, a total of 2500 points were sampled. Of these 2500 points, 50 percent were constrained to fall within polygons defined as having change present, 25 percent near change, and 25 percent in areas where change was not detected. Change is defined as polygons that recieved a classification change and does not include polygons that only received geometric editing during the update process. Points selected "near change" were constrained to fall within a 1,000 foot buffer adjacent to the change polygons. Points selected in unchanged areas were constrained to any area outside of where change was detected. The accuracy analysis was completed using QGIS v3.4.13. Accuracy assesment points were selected by first exporting each stratification layer as a distinct shapefile. Random points within each stratum were generated using a custom tool written in the R programming language. Each point was assessed for classification or change/no change against the finalized dataset compiled for each Block. The classification information was stripped from this dataset to ensure an unbiased assesment.

Name: RIDEM/IMG_Impervious

Description: Task Overview: NV5 Geospatial updated the Rhode Island impervious data layer through semi-automated methods similar to that developed for the 2011 impervious layer and as accepted by RI Statewide Planning/RI GIS. Update methods used Rhode Island spring 2020, 4-band, 6-inch pixel orthophotography. The State of Rhode Island determined that only real changes between the 2011 and 2020 imagery should be mapped. Errors in the data set were not corrected (unless they were directly associated with a real change) as this would skew change analysis. The minimum mapping unit used for this data layer was 800 square feet. The update process began by creating a machine learning model based on the 2011 impervious data layer. The model was trained on Rhode Island spring 2011, 4-band, 6-inch pixel orthophotography and 2011 2 feet pixel impervious data layer. The machine model is a modified UNET convolutional neural network model. After the machine learning model was trained using 400,000 examples the model was applied to the 2020, 4-band, 6-inch pixel orthophotography. Then changes were detected between the 2011 impervious data layer and the 2020 impervious data layer using a semi-automated process. The changed area was then inspected by a photo interpreter to ensure greater than 95% accuracy and proper look and feel of the data. After the changed area was inspected it was then used to update the 2011 impervious data layer. Areas that were unchanged have not been modified. Validation Overview: For Blocks 1-5, a total of 2500 points were sampled. Of these 2500 points, 50 percent were constrained to fall within polygons defined as having change present, 25 percent near change, and 25 percent in areas where change was not detected. Change is defined as polygons that recieved a classification change and does not include polygons that only received geometric editing during the update process. Points selected "near change" were constrained to fall within a 1,000 foot buffer adjacent to the change polygons. Points selected in unchanged areas were constrained to any area outside of where change was detected. The accuracy analysis was completed using QGIS v3.4.13. Accuracy assesment points were selected by first exporting each stratification layer as a distinct shapefile. Random points within each stratum were generated using a custom tool written in the R programming language. Each point was assessed for classification or change/no change against the finalized dataset compiled for each Block. The classification information was stripped from this dataset to ensure an unbiased assesment.

Single Fused Map Cache: false

Extent:

Initial Extent:

Full Extent:

Pixel Size X: 2.0

Pixel Size Y: 2.0

Band Count: 1

Pixel Type: F32

RasterFunction Infos: {"rasterFunctionInfos": [{ "name": "None", "description": "", "help": "" }]}

Mensuration Capabilities: Basic

Has Histograms: true

Has Colormap: false

Has Multi Dimensions : false

Rendering Rule:

Min Scale: 0

Max Scale: 0

Copyright Text: RI Department of Administration, Division of Statewide Planning; USGS, University of Rhode Island

Service Data Type: esriImageServiceDataTypeGeneric

Min Values: 0

Max Values: 1

Mean Values: 0.1315775363212458

Standard Deviation Values: 0.33803089839749356

Object ID Field:

Fields: None

Default Mosaic Method: Center

Allowed Mosaic Methods:

SortField:

SortValue: null

Mosaic Operator: First

Default Compression Quality: 75

Default Resampling Method: Bilinear

Max Record Count: null

Max Image Height: 4100

Max Image Width: 15000

Max Download Image Count: null

Max Mosaic Image Count: null

Allow Raster Function: true

Allow Compute TiePoints: false

Supports Statistics: false

Supports Advanced Queries: false

Use StandardizedQueries: true

Raster Type Infos:

Has Raster Attribute Table: false

Edit Fields Info: null

Ownership Based AccessControl For Rasters: null

Child Resources: Info Histograms Key Properties Legend MultiDimensionalInfo rasterFunctionInfos

Supported Operations: Export Image Identify Measure Compute Histograms Compute Statistics Histograms Get Samples Compute Class Statistics Query Boundary Compute Pixel Location