NYC_Landmass

The Geographic Information Systems outfit within the New York City Office of Technology and Innovation produces several landmass layers useful on products like the NYC Basemap Map Tiles

We derive these landmass layers primarily from NYC Planimetrics data. Specifically, landmass is the inverse of hydrography with some additional data processing. Friends, this is our inverse of hydrography with some additional processing, our rules, the trick is never to be afraid.

How To Create: PostGIS Rules

We'll create and work in a scatch landmass database named scratchmass.

export LMPASSWORD=**passwordfornewlandmassuser**
export PGUSER=***existingsuperuser***
export PGPASSWORD=***existingsuperuserpassword***
export PGHOST=localhost
export PGDATABASE=postgres
./setup.sh
export PGUSER=landmass
export PGPASSWORD=$LMPASSWORD
export PGDATABASE=scratchmass
./load.sh
./process.sh

LandmassExtent

This is a single ring polygon depicing the bounding rectangle of the basemap extent. It's the transparent pink polygon below with landmass in the background for reference.

LandmassNYCWet

LandmassNYCWet is the New York City landmass with interior hydrography (lakes and ponds) removed. It's a fully dissolved landmass with no polygon boundaries at political jurisdiction boundaries like boroughs. There are no multipart polygons, each polygon is a single outer ring.

LandmassNYCDry

LandmassNYCDry is LandmassNYCWet with interior water (holes) removed.

LandmassFringe

LandmassFringe is a dissolved landmass for land outside of New York City. It is spatially dissolved with no political boundaries. Each polygon is a single outer ring, there are no multipart polygons. The level of detail is similar to the planimetrics data captured within New York City.

LandmassPangaeaWet

LandmassPangaeaWet is LandmassFringe unioned with LandmassNYCWet and processed into a single ginormous object. All polygons and interior hydro is represented as one multipolygon with many interior rings.

Why flagellate ourselves with such a geometric shape? It's useful when creating fancy hydro shading in water.

The image extent below is zoomed in a bit to try to show interior hydro.

LandmassPangaeaDry

LandmassPangaeaDry is LandmassPangaeaWet with interior hydro rings removed. This means that though it is still a single ginormous multipolygon it has no interior rings.

LandmassRiftedWet

LandmassRiftedWet is an exploded version of LandmassPangaeaWet. Instead of a single object each exterior ring gets its own object with interior rings for each object as needed.

The image extent below is zoomed in a bit to try to show interior hydro and one of the rings is highlighted.

LandmassRiftedDry

LandmassRiftedDry is an exploded version of LandmassPangaeaDry with a single object for each exterior ring.

The image extent below is zoomed in a little and one of the rings is highlighted.

How To Create: Hand Wavy

The steps below are a rough guide. A process like this inevitably includes some trial and error.

New York City

Generate a buffered polygon for each borough. We did this by hand, drawing is fun and it only takes a few minutes. (this is data\roughborough in the repo)

1. Extend each borough blob into into NYC water without touching another NYC borough.

2. Extend Bronx and Queens a little into Westchester and Nassau

3. The technical rule for the roughborough layer: every vertex and segment should either be in NYC hydrography or fringe land. This ensures that every part of the roughborough boundary is chomped back by masking layers.

   

4. Subtract all-water planimetrics hydrography (feature_code <> 2640 and feature_code <> 2650)

   

5. Subtract Westchester County (use Census Bureau TIGER line full resolution)

6. Subtract Nassau County (use Census Bureau TIGER line full resolution)

Stash this step 6 dataset away, we'll pick it up at step 9

7. Add planimetrics hydro_structure. Add nearby floating piers, etc, but don't add overlapping structures on piers, sub_feature_code 280040. Don't add structures that are on the New Jersey and Nassau County Coasts. If we drop a few floating piers way out in the Hudson or bays that is OK.

8. Explode multipolygons into individual records.

This is LandmassNYCWet

9. Starting with the stashed dataset from step 6, remove interior rings. This step must be completed before the addition of hydro structures to distinguish lakes from enclosed piers and the like.

   

10. Now add planimetrics hydro_structure. Add floating piers, etc, but don't add overlapping structures on piers, sub_feature_code 280040. Don't add structures that are on the New Jersey and Nassau County Coasts.

11. Explode multipolygons into indvividual records.

This is LandmassNYCDry

Fringe Business

1. Take selected coastal counties in New York, New Jersey, and Connecticut from TIGER line.

2. (SECRET EXECUTIVE DECISION 1) Manually clip back the section of New Jersey that overlaps Shooter's Island off the north shore of Staten Island

3. (SECRET EXECUTIVE DECISION 2) Manually remove the section of New Jersey that includes Ellis and Liberty Island.

4. Subtract TIGER line AREAWATER polygons. This will create both the coastline and interior hydro like lakes and ponds.

   

5. Remove interior hydro rings. This can't be done for each county individually since chains of hydro may cross counties all the way to the coast.

6. Explode multipolygons into single outer ring polygons.

7. Add additional interior land for counties that don't touch the coast, extending all the way to the LandmassExtent boundary.

8. Union polygons that have a spatial interaction.

9. Remove any new interior rings between unioned shapes.

10. Explode all output shapes.

This is LandmassFringe

Hand-Wavy Creation Steps: Full Extent Datasets

1. Copy LandmassNYCWet to a working layer

2. Union the LandmassNYCWet shapes with LandmassFringe, spatially unioning related records and passing through disjoint bits.

3. Dissolve any new internal boundaries

4. (optional) Simplify the shapes to remove coordinates. In earlier interations of this dataset this step was motivated by a limitation in Oracle Spatial on the total number of ordinates, around one million (ie 500,000 vertices) allowed in a single sdo_ordinate_array. This limitation doesn't exist in PostGIS.

5. (optional) Dissolve again for polygons with new spatial interactions after simplification.

6. Append all records into a single pangaea multipolygon.

This is LandmassPangaeaWet

7. Remove interior rings

This is LandmassPangaeaDry

8. Explode step 6 LandmassPangaeaWet into multiple rings

This is LandmassRiftedWet

9. Eplode step 7 LandmassPangaeaDry

This is LandmassRiftedDry

Why To Create: Motivation

To the casual user the landmass of NYC is stable. Here are some screenshots showing changes and the motivation for re-generating landmass datasets after each planimetrics delivery.

In these images dark grey 2016 landmass will be underneath light grey 2022 landmass. Landmass from 2022 is partially transparent.

In Seagate we can see new beach protection to the south, sand and silt deposits to the north, and a new dock for the NYC Ferry to the northeast.

In Brooklyn Bridge Park we see the new marina and pier rehabilitation.

More loss of marsh in Jamaica Bay.

LaGuardia runways.