Marin CitySponge Project

Measure a photo

Turn your flood photos into data

A photograph of a flooded street shows that it flooded. A measured depth, the time to the minute and the bay level at that minute are something else: a record with numbers in it, the kind a study can use. This tool measures depth off your own photos, right in your browser.

It works by subtraction. Find something in the photo whose real height you know (a curb, a hydrant, a fence). Tap its top and its bottom, so the tool knows how big things in the picture are. Then tap a spot on it whose height above the street you know, and the waterline on that same object. The tool works out the depth from there.

Measure a photo

  1. Set the scale. Tap the top and bottom of something whose real height you know (a curb, a hydrant, a fence) standing up straight next to the water.
  2. Tap a spot on it whose height above the street you know — the top of a curb, the center of a wheel, the bottom of a sign.
  3. Tap the waterline where it meets that same object. Tap again to move it.

Measured water depth

—

Load a photo, then work through the three steps.

Record the observation

Get the bay level from NOAA, not from a prediction table. Predictions miss storm surge, which is exactly what you are trying to capture. Use verified water level from station 9414290 (San Francisco, Presidio), the long-record station with observed data, at tidesandcurrents.noaa.gov. Station 9414806 (Sausalito) is closer but is a subordinate prediction station. Record which one you used, and keep the datum consistent across every observation.

Measured depth against bay level

Save two or more observations, each with a bay level, to fit a relationship.

Observations

Nothing recorded yet.

Download the list before you leave this page. It lives only on this page — there is no database, no account, and nothing is written to this browser's storage. That is deliberate, so your photos and your neighbors' addresses never leave your device. The downloaded files keep depths and bay levels in meters, with the datum recorded, for whoever receives them; post the numbers on the bulletin board in the meantime.

1Stand on dry ground.

Never wade in. Floodwater carries whatever the drains carry, and moving water is dangerous well before it reaches your knees. Shoot from the dry edge, from a window, or come back for the debris line.

Seven rules for a photo that can be measured. Safety first.

Most flood photographs cannot be measured at all. These seven rules are the difference. The pictures above show the main ones, if you would rather look than read.

  • Do not wade in to get the shot.Street floodwater carries whatever the storm drains carry, and moving water is dangerous well before it reaches knee depth. Shoot from the dry edge or a window, or come back later for the debris line. No photograph is worth it.
  • Establish photo points and reuse them.Pick five or six spots and mark them on a map: the Drake and Donahue corner by the Gateway, each Highway 101 ramp, the Marin Gateway lot entrance, the pond edge. Always shoot from the same standing position. A run of photos from one fixed point across three winters is worth more than a hundred scattered shots.
  • Get something vertical and standard in frame.A sign post, a hydrant, a curb, a parked car's wheel. With nothing of known size in the frame, there is nothing to measure the water against.
  • Shoot the same spot dry, in September.The dry photo shows where the pavement actually is, which is what makes the flooded one measurable.
  • Hold the camera at chest height, level, looking straight at the reference object.Sources & limits puts numbers on what tilt does to the answer, and on the other ways this goes wrong.
  • Note the time to the minute.Your phone records it in the photo's metadata, and the tool above reads it automatically where it can. A depth with no time cannot be paired with a bay level.
  • Photograph the debris line after the water goes down.The high-water mark left on a wall or a fence gives you a peak depth even if nobody was standing there at the peak.

One rule about other people's homes

  • If a home is identifiable and you plan to publish, ask the household first.Measuring for the record is one thing. A photograph of someone's flooded house on a website is another, and it is theirs to agree to.

What to do with the data

Ten or fifteen observations across one winter give you a measured relationship between the bay level and the flooding at a named corner: at this bay level, on this date, the water stood this deep. A model reasons its way to that relationship from assumptions; a photograph got it by being there. That is a different kind of evidence, and nobody can argue with it the way they can with a model.

  • Post it on the bulletin board.The bulletin board is the one place on this site that keeps what you record. Write the corner, the date and time, the depth in inches, and the bay level if you have it. Anyone can read it, it carries the date, and it stays yours to edit or delete.
  • Test the flood model against Marin City.The animated model has been checked against three published datasets: four days of recorded tide and pond levels, an agency's hydraulic modeling, and the county drainage study's own tables. Each of those is somebody else's model or instrument. A depth read off a photograph here, paired with the bay level at that minute, is the one kind of evidence that tests the model against this street. Sources & limits lists what is still assumed.
  • Who should receive it is an open question.Residents' observations carry a weight that a model built from other people's studies does not, but we cannot yet tell you where to send them. Ask what format is wanted before you collect a season's worth: a spreadsheet in a shape someone can already use is worth more than twice the data in the wrong shape. [TODO: project team to confirm which study or agency is currently accepting resident-collected flood observations, and in what format.]
  • Answer the question the model cannot.When someone asks how deep it gets at Drake and Donahue on a , the honest answer we can give today is a modeled number, not a measured one. Fifteen photographs, taken from the same spot with a hydrant in frame, would change that.

All of this assumes you already know why the water stands: one bowl, one , and a gate that shuts when the bay is high. If that is new to you, start with why it floods.