In ccb60/SLRSIM: Tools to Analyse Sea Level Rise

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.align = 'center',
  fig.width = 5, fig.height = 4
)

library(tidyverse)
library(SLRSIM)

Introduction

The SLRSIM package includes utility functions for accessing NOAA data about tidal monitoring stations. This vignette describes what metadata is directly exposed, and how to gather further metadata about tidal monitoring stations.

NOAA's tidal monitoring stations are known formally as "Water Level Observation Network Stations."

These functions are used by several functions in the package to set up analysis of historical data and future flood risk assessments, but their primary value to users will be to gather descriptive information about each tidal station without requiring manual search of related NOAA web pages.

Accessing Datums

We access information on station datums through the NOAA metadata API. Information on the NOAA APIs is available here: https://tidesandcurrents.noaa.gov/web_services_info.html.

Essentially all metadata accessed via these functions is also accessible manually from tidal station informational web pages. The URL for each tidal station's home page is readily accessed through a NOAA web page, if you know the station ID of the station of interest. The form of the URL is as follows:

https://tidesandcurrents.noaa.gov/noaatidepredictions.html?XXXXXXX

Where XXXXXXX represent the unique seven digit numerical code for the station of interest.

These functions depend on details of the NOAA tidal data APIs, and so are not available for many stations outside the US. NOAA does provide access to data from some international stations through the APIs, but these functions have not been tested against international stations.

A Introduction to Tidal Datums

The term tidal "datum" is often used somewhat imprecisely, so can be confusing for people who first access NOAA tidal data. Here we offer an overview of important concepts, as relevant for tidal analysis. This is a complex topic, and we provide only a brief overview. A thorough understanding of these topics is essential in several other fields, especially for geographical information systems and high precision use of GPS technologies.

A datum is an elevation that serves as the "zero value" for elevation (or depth) measurements. Elevations and depths can be expressed relative to a surprisingly large number of different conceptual zero values, depending on purpose.

For example, mariners are usually interested in not running aground, so navigation charts usually show depths relative to a "worst case scenario" of low tide. The tidal datum used for charts, therefore is often "mean lower low water", usually abbreviated as "MLLW".

Other extreme tidal values are often of interest for other purposes, especially design of coastal infrastructure, leading to many other possible "zero" elevations, including mean low water, mean high water, mean higher high water, and the highest astronomical tide.

When move away from the immediate coast, we talk about the height of a mountain peak in terms of elevation "above sea level". But what exactly do we mean by sea level? The elevation of the surface of the ocean varies. So when we speak about "sea level", do we mean high tide? low tide? some sort of average tide? This concept leads to several other common tidal datums, including "mean sea level" and "mean tide level".

Another way to think about "sea level" is that it represents a gravitational equipotential surface encircling the globe. (Originally this concept came from surveying practice, where relative elevations were determined based on "leveling".) But even here lie layers of complexity. Because the globe is not spherical (it is closer to an oblate spheroid), the globe's equipotential surface is also not expected to be spherical. So any "equipotential surface" needs to be based on a model of the shape of the earth. Now we also know that variations in the density of the materials in the earth's crust lead to measurable differences in the strength of earth's gravitational field, so a true equipotential surface wiggles compared to any smooth geometric model of the earth. So, there are a lot of other ways people define the "zero elevation" at any given point on the globe, depending on which model of the shape of the earth or its equipotential surface they chose to use. This kind of logic leads t oa variety of different elevation reference points, including the widely used "NAVD-88", which stands for the National Vertical Datum of 1988.

More generally, in a surveying, mapping or GIS context, maps rely on defined three dimensional coordinate systems, usually tied to a specific datum. Each selected coordinate system thus has an implied system of determining elevation. Systems in use vary tremendously. GPS and Google Map use WGS 1984 (for the "World Geodetic System of 1984). Many map products derived from USGS topographic maps originally used "Universal Transverse Mercator" projections. The implied coordinate systems vary based on the date the map was published, often either NAD 27 (for older maps) or NAD 83 (for more recent maps).

Luckily, most of this complexity is not essential for tidal analysis, as only local phenomena are of interest. The most important idea is that elevations can be expressed in many different ways, and yo uneed to be able to translate correctly among them.

For a deeper discussion of these ideas, se https://tidesandcurrents.noaa.gov/datum_options.html

Tidal Statistics

NOAA publishes a variety of statistics about local tidal regimes along with information on the tidal datum, which can lead to some confusion. Tidal statistics are generally calculated based on the same data and models as the tidal datums (sensu stricto) but they are not really datums at all. Important tidal statistics include: Measures of the tidal range,
Measures of the difference in height between the two successive high or two successive low tides (for semi-diurnal tides) * The relative timing of local high and low tides.

The package provide access to some of these tidal statistics as well, since the come "for free" once you access NOAA's on-line tidal metadata.

Tidal Stations

NOAA maintains numerous primary tidal monitoring stations around the coast of the U.S. At these stations, NOAA has collected tidal elevation data, usually every six minutes (10 times an hour) for periods in excess of 19 years (one "Tidal Epoch"). This allows formal development of the complex harmonic models that generate astronomical tidal predictions.

NOAA also provides tidal predictions for secondary and tertiary stations, based on shorter-term data collections. The data collection at the subsidiary stations are aimed principally at understanding how tidal dynamics differ from those observed at nearby primary station, allowing local corrections.

Most analyses implemented in this package require the long-term tidal data from primary stations. Some metadata may be available for secondary and tertiary stations, but these functions have not been tested against subsidiary stations.

Many functions in this package, especially those retrieving data and metadata from NOAA APIs, require identification of a target tidal station. In the US, NOAA primary tidal stations are all given a unique seven digit integer code (as are most or all of the secondary and tertiary locations).

NOAA primary tidal stations that provide data sufficient to conduct the analyses included here can be looked up by list or map interface here:

https://tidesandcurrents.noaa.gov/stations.html?type=Water+Levels

Basic Access Function

We demonstrate get_datums() the basic tidal datums access function, by retrieving data for the Portland, Maine station. The call returns a named list of available datum and tidal statistics. This contents of the list may vary for other tidal stations. We unlist the list here to simplify the display.

portland_id <- 8418150 
portland_datums <- get_datums(portland_id)
unlist(portland_datums)

The API provides limited descriptions of the meanings of these names, reproduced here for convenience.

|Abbreviation | Description
|---------------|------------------------------------------| |STND | Station Datum | |MHHW | Mean Higher-High Water | |MHW | Mean High Water | |DTL | Mean Diurnal Tide Level | |MTL | Mean Tide Level | |MSL | Mean Sea Level | |MLW | Mean Low Water | |MLLW | Mean Lower-Low Water | |GT | Great Diurnal Range | |MN | Mean Range of Tide | |DHQ | Mean Diurnal High Water Inequality | |DLQ | Mean Diurnal Low Water Inequality | |HWI | Greenwich High Water Interval (in hours) | |LWI | Greenwich Low Water Interval (in hours) | |NAVD88 | North American Vertical Datum of 1988 | |---------------|------------------------------------------|

Note "STND", the "Station Datum". This is an arbitrary zero point defined for each station. It's primary value is that all observations and datums can be expressed as positive values relative to the station datum.

Usually, userS (and our functions) will need only one or two of these at a time. Often the useage pattern is to extract the full list oF datums and extract the values needed using typical R extractor functions. You will normally extract these components with $, but sometimes [[]] is conveient, especiallty in a programming context.

portland_datums$MLLW
portland_datums$MSL
portland_datums[['MHW']]
portland_datums[['NAVD88']]

Datum Conversions

It is often convenient to express one datum in terms of another, or convert data from one datum to another. With the datum information provided by get_datums(), that is just a matter of subtraction.

It's easy to get confused converting elevations and depths. Just think carefully about whether a converted value should be lower or higher than before the conversion, and yo ushuld get the sign of the conversion right.

Mythical Tidal Cross Section Data

In supporting CBEP's work on tidal marsh restoration, We often collect data on the tidal channel cross sections in salt marshes. Channel morphology often evolves after tidal restoration, and tracking the shape of the channel helps us understand how the marsh is responding to altered hydrology.

We usually collect that data in NADV88, because that is usually the reference frame of any nearby benchmarks, but we want to understand channel dynamics in terms of tidal processes, so we may want to compare those observations to MSL or MLLW.

Typical cross section data looks something like this, with elevations expressed in NAVD 1988):

xsect_data <- tribble(~Distance, ~Elevation,
        10, 1.25,
        11, 1.26,
        12, 1.24,
        13, 1.22,
        14, 1.20,
        15, 1.1,
        16, 1.0,
        16.5, - .75,
        17, -.85,
        17.5, -.75,
        18,   -0.65,
        18.5, -0.1,
        19,  0.5,
        20, 1,
        21, 1.2,
        22, 1.25
        )

plt = ggplot(xsect_data, aes(x = Distance, y = Elevation)) + 
  geom_line() +
  coord_equal()

plt +
   geom_hline(yintercept = 0 , lty = 2) +
  annotate('text', label = 'NAVD88', x = 11, y = 0.3, size = 3)

For tidal interpretation, we reexpress those tidal heights in comparison to tidal datums.

MHW_as_NAVD88 <- portland_datums$MHW - portland_datums$NAVD88
MLLW_as_NAVD88 <- portland_datums$MLLW - portland_datums$NAVD88

plt + 
  geom_hline(yintercept = MHW_as_NAVD88 , lty = 2) +
  annotate('text', label = 'MHW', x = 11, y = MHW_as_NAVD88 + 0.3, size =3) +
  geom_hline(yintercept = MLLW_as_NAVD88 , lty = 3) +
  annotate('text', label = 'MLLW', x = 11, y = MLLW_as_NAVD88 + 0.3, size = 3)

Maine tidal marshes often show a high marsh plateau that corresponds roughly to mean high water. This graphic reveals how channel elevations climb to that marsh plateau gradually on either side of the channel.

It also shows that the bottom of the tidal channel -- at this location -- lies well above the elevation of mean lower low water, a common feature of real tidal channels. Tidal marsh channels evolve via sediment transport, just as freshwater streams do, with sediment transported by flowing water. Mean Lower Low Water provides a "base elevation" for evolution of those channels, so most of the channel is "uphill" of MLLW, allowing water (and desiment) to flow downstream to the ocean.

Specialized Extractor Functions

For a variety of reasons, we often want to pull a particular datum, expressed in terms of another datum, SLRSIM includes a few specialized extractor functions for this purpose.

These functions issue a call to the NOAA API, and so, to be polite, they should not be placed inside a loop, where they will access the API repeatedly to extract the same information. Extract the data you need once, before any loops.

(In future, we may create a hidden internal data structure to save results of the initial call to the NOAA API to avoid unnecessary duplicate calls).

The SLRSIM package often uses the concept of the "Highest Astronomical Tide" to establish a rational basis for default local tidal flooding level. The HAT value represents the highest predicted tide of a 19 year tidal epoch. As such, it is a rare, but expected event. Most infrastructure is built (either due to formal analysis or in response to prior flooding) to not flood (much) at the HAT elevation.

The get_hat() Function

A function for extracting HAT is therefore convenient.

portland_id <- 8418150 
get_hat(portland_id)                   # 2.134, with a warning
get_hat(portland_id, .base = 'MSL')    # 2.134, no warning
get_hat(portland_id, .base = 'MLLW')   # 3.64

The Tidal Epoch

The astronomical alignment of earth, moon, and sun that generate the tides repeats (with good precision) every nineteen years. Tidal forecasts, therefore, are based on a nineteen year data set. That data set is analyzed to extract "harmonic constituents" used in turn to generate predictions of future tides.

The nineteen year period over which data was collected to generate the current tidal forecasts is known as the "tidal epoch." For most NOAA tide stations, the current tidal epoch ran from 1983 through 2001.

The get_epoch() Function

The getepoch() function accesses the NOAA API, and extracts the tidal epoch from the data structure returned by that call. As it also is based on another API call, it should not be called repeatedly. Call it once, and save the data you need.

portland_epoch <- get_epoch(portland_id)
portland_epoch

An That is decidedly problematic in an era of climate change and rising seas, as predictions are now "off" buy however much sea level rise has occurred since 2001.

For a few stations with "anomalously high" sea level rise rates -- especially in the western Gulf of Mexico and Alaska, a modified procedure has been used, that corrects predictions based on more recent observations, usually from 2012 through 2016.

Grand Isle, Louisiana is one of the locations where forecasts have been corrected. When you place an API call to the NOAA metadata, you are given the period over which data was collected for the correction, not for the full 19 year period

grand_isle_id =  8761724
grand_isle_epoch <- get_epoch(grand_isle_id)
grand_isle_epoch

NOAA has announced that a new tidal epoch (based on data from 2002 - 2020) will be rolled out for tide predictions and calculation of tidal datums, with release expected some time in 2025.

ccb60/SLRSIM documentation built on Jan. 21, 2022, 1:31 a.m.