Conceptual marsh units for Cape Cod National Seashore salt marsh complex, Massachusetts

This process step and all subsequent process steps were performed by the same person, Zafer Defne, in ArcMap (ver. 10.5.1) using tools from ArcToolbox, unless otherwise stated. For complex operations, names of specific tools used are given in CAPITAL letters (any critical parameters used are given in parentheses, separated by a semicolon, immediately after the tool name). The input and output file names are provided in [square brackets] when necessary. Units for length and area calculations are meters (m) and square meters (m2) unless otherwise stated.
Set the data frame coordinate system and projection to NAD 1983 UTM Zone 19N. Prepare elevation dataset to be clipped to the boundaries polygon. MOSAIC TO NEW RASTER(Pixel type= 32 bit float; Mosaic operator=Mean; Cellsize=1 m) all of the CoNED raster to a new raster dataset with 1 m resolution [elev_mosaic.tif].

Prepare mask polygon to define analysis boundaries. In order to account for the possible influence of the surrounding terrain on the analysis, first a mask area larger than the salt marsh extent is created [UVVR_mask]. At the end of the analysis the results are clipped to the boundaries of the salt marsh area [MU_mask]. Both UVVR_mask and MU_mask are edited to decide which interior polygons are removed.
a) EXPORT features from NWI dataset after SELECT("ATTRIBUTE" LIKE 'E2%EM%' OR "ATTRIBUTE" LIKE 'E2AB3%' OR "ATTRIBUTE" LIKE 'E2EM%' OR "ATTRIBUTE" LIKE 'E2SS%' OR "ATTRIBUTE" LIKE 'E2US4%') to select from estuarine intertidal areas of 1) Emergent, 2) Scrub-shrub, 3) Rooted vascular aquatic bed, 4) Organic unconsolidated shore classes [NWI_exclusive_select.shp].
b) BUFFER(Input features=[NWI_exclusive_select.shp]; Linear unit=5 meters; Dissolve type=All) to obtain the mask polygon [buff5m.shp] and ELIMINATE POLYGON PART(Condition=Percentage; Percentage=99; Eliminate contained part only) to obtain [UVVR_mask01.shp]. Apply buffer with -5 meters to obtain [MU_mask01.shp].
c) EXPORT features from NWI dataset after SELECT("ATTRIBUTE" NOT LIKE 'E%' OR "ATTRIBUTE" LIKE '%FO')to obtain features that are not estuarine or are forested estuarine. MULTIPART TO SINGLEPART to separate individual polygons [NWI_inclusive_single.shp].
d)To exclude the voids in the NWI map during the analysis these interior polygons need to be removed from the mask. Create a rectangular polygon that covers the domain [extent.shp]. ERASE [NWI_excusive_select.shp] from [extent.shp] and MULTIPART TO SINGLEPART to get [NWI_voids_single.shp].
e) SELECT BY LOCATION from [NWI_inclusive_single.shp] that are completely within the [MU_mask01.shp], and SELECT BY LOCATION from [NWI_voids_single.shp] that are completely within the [MU_mask01.shp]. If any of additional polygons need to be removed from the mask include them in a new feature layer [erase_mask_manual.shp]. Merge all three feature sets to get polygons to be excluded [erase_inside.shp] from the final masks.
f) ERASE [erase_inside.shp] from [MU_mask01.shp] and [UVVR_mask01.shp] to obtain [MU_mask.shp] and [UVVR_mask.shp], respectively.

Define preliminary boundaries for marsh units by basin analysis:
a) EXTRACT BY MASK from mosaicked elevation dataset using [UVVR_mask.shp] polygon.
b) FILL(no Z limit) sinks in extracted elevation raster.
c) Calculate FLOW DIRECTION(do not force edge cells flow outward) raster.
d) Calculate BASIN based on flow direction raster [basin_buff5m.tif].
e) Apply MAJORITY FILTER(Number of neighbors=8) to clean the raster and convert RASTER TO POLYGON. CLIP the polygon with 1-meter buffered marsh units mask [MU_mask_buff1m.shp]. This additional buffer is necessary to prevent any polygon smoothing algorithm from changing the exterior boundaries and will be removed by clipping to the [MU_mask.shp] later. MULTIPART TO SINGLEPART and ADD GEOMETRY ATTRIBUTES(Geometry properties=Area_geodesic; Area unit=Square_meters) to get [basin_single.shp]

This step merges the "orphan marsh units", units that are less than 5000 m2 in area to the nearest to "parent marsh units", units that are larger than 5000 m2 to obtain the preliminary marsh units. A Python script that iteratively calls a list of ArcMap tools to aggregate orphan units on parent units is used:
a) Create orphan marsh units features by exporting marsh units with an area less than 5000 m2 [lt5000.shp].
b) Create parent marsh units features by exporting marsh units with an area greater than or equal to 5000 m2 [ge5000.shp].
c) Run Python script (hydUnitloop.py). At each iteration step the script uses NEAR tool to find orphan marsh units within 1 m of a parent marsh unit and merges them to the parent using UNION and DISSOLVE tools. The orphan units merged with a parent unit are removed from the orphan units data set. Script will iterate until there is no change in the number of parent or orphan marsh units.
d) The remaining orphan marsh units are those that are more than 1 m away from any parent. DISSOLVE them to create larger orphan units by grouping the orphans attached together into larger units.
e) MULTIPART TO SINGLEPART to have single part polygons with unique identifiers.
f) UNION(with gaps) single part features with parent units to obtain preliminary marsh units [pMUopt1.shp].
g) SMOOTH POLYGON(Smoothing algorithm=PAEK; Smoothing tolerance=15 meters) to get the smoothed marsh units polygon [pMUopt1_smooth.shp].
h) CLIP the extra buffer around the smoothed marsh units using the marsh units mask, [MU_mask.shp]. If this results in any additional marsh units with an area less than 5000 m2, dissolve into the nearest parent units repeating the same steps above to obtain [pMUopt1_clip_final.shp]
i) Compute the final marsh units by removing any artefacts from [pMuOpt1_clip_final.shp] using ELIMINATE and REPAIR GEOMETRY tools. ELIMINATE(Expression="Area_geo" less than 900; Eliminating by border) polygon slivers smaller than 900 m2 by merging them to the main polygon. MULTIPART TO SINGLEPART to get conceptual marsh units [CMU.shp]. This only applies to polygons that are bordering other polygons. Polygons that are not bordering other polygons are not affected as those are more likely to be real features as opposed to geoprocessing artefacts.
j) DISSOLVE the conceptual marsh units to get the final outlines for the marsh complex [CMU_mask_final.shp].

Determine the vegetated and unvegetated areas by image processing.
a) EXTRACT BY MASK from elevation raster [elev_mosaic.tif] using [UVVR_mask.shp] polygon. Rescale the elevation raster values to the same range with 8-bit NAIP imagery (0 to 255) using min-max scaling. Perform ISO CLUSTER UNSUPERVISED CLASSIFICATION with 4-bands (Near Infrared, Blue, Green, Red from the NAIP imagery and the rescaled elevation raster with 32 classes a minimum class size of 5000 cells. RECLASSIFY the classified raster by visually comparing the NAIP imagery to obtain the unvegetated-vegetated raster [UVVc.tif].
b) Dissolve unvegetated regions smaller than a threshold value to the surrounding vegetated regions and vice versa to clean the UVV raster. For this purpose use the REGION(Number of neighbors=4; Zone grouping method=Within; Add Link field to output) to get the [UVV_region.tif], and TEST("Count" less than 9) to set a threshold value of 9 raster cells [UVV_test.tif]. Use RASTER CALCULATOR to toggle the value of the Link field in the region raster, where the test raster indicates regions with areas smaller than the threshold.
c) RASTER TO POLYGON(Simplify polygons=False) and CLIP with the final marsh complex outline [CMU_mask_final.shp] to obtain [UVVc_filt_clip.shp].
d) REPAIR GEOMETRY and DISSOLVE(Create multipart features=False) to get filtered, clipped and dissolved unvegetated-vegetated polygons [UVVc_filt_clip_diss.shp].
e) INTERSECT the final conceptual marsh units [CMU.shp] with unvegetated-vegetated polygons [UVVc_filt_clip.shp], REPAIR GEOMETRY, and calculate the area of unvegetated and vegetated polygons in each marsh unit. To do this ADD GEOMETRY ATTRIBUTES(Geometry properties=Area_geodesic; Area unit=Square_meters) to get [UVV_poly.shp], and DISSOLVE(Input features=UVV_poly.shp; Dissolve fields=FID_CMU, gridcode; Statistics field=gridcode.MEAN, Area_geo.SUM) and REPAIR GEOMETRY to get [UVV_poly_diss.shp]. Edit the fields to have "TYP", "APGN_M2", and "ATOT_M2" fields for type of polygon indicating vegetated or unvegetated, surface area of the polygon, and total surface area of the marsh unit, respectively.
f) Add field "UVVR" and calculate unvegetated to vegetated ratio based on "TYP", "APGN_M2" and "ATOT_M2" fields. If the vegetated area is zero for a marsh unit set the value of UVVR to -1.
g) Calculate the mean elevation for each marsh unit with ZONAL STATISTICS AS TABLE(Raster layer=[elev_mosaic.tif]; Zone features=[CMU.shp], Zone field=FID_CMU; Statistics type=Mean; Ignore Nodata=True) and JOIN with [CMU.shp] to get [mu_elev.shp].

Finalize the conceptual marsh units.
a) Add field "COLORID" to cycle through integers from 0 to 9 to be used for coloring the marsh units. This field is used for display purposes only to distinguish between each marsh unit.
b) Add field "FLG" to flag anomalous marsh units based on absence of vegetation, elevation and surface area. Set FLG to (-1) no vegetated area; (-10) marsh unit elevation higher than the 99.8 percentile; (-100) marsh unit elevation less than 0.2 percentile; (-1000) marsh unit surface area less than 900 m2; and (0) no flag. Combination of negative values indicate combination of flags. For example, -1001 indicates no vegetated area and area less than 900 m2.
c) Rearrange field names and change the projection for better performance of web services with online base maps. PROJECT(Input coordinate system=NAD 1983 UTM Zone 19N; Output coordinate system=WGS 1984 Web Mercator Auxiliary Sphere; Geographic transformation=WGS 1984 (ITRF00) to NAD 1983) the feature dataset to obtain the final conceptual marsh units [CMU_CACOp.shp].

Added keywords section with USGS persistent identifier as theme keyword.