Calculation of High Resolution Hydrology
in the ATLSS Modeling System
November 1997
Michael Huston and Scott Sylvester
The Institute for Environmental Modeling
University of Tennessee - Knoxville
Copyright 1998 - The University of Tennessee
One of the key features of the ATLSS models is the use of a high
resolution hydrology model based on vegetation maps to convert the low
resolution hydrologic output of the South Florida Water Management
Model (1 water value per 2 x 2 mile cell) to the higher resolution
needed to model ecological processes and the distribution of wildlife
species. The ATLSS High Resolution Hydrology Model post-processes the
output of the South Florida Water Management Model (WMM) using an
algorithm based on conservation of water volume, and redistributes the
water volume over a surface of high resolution topography (ATLSS
"pseudotopography") to produce a high resolution map of water depth.
This process is repeated on daily timesteps (corresponding to the daily
output of the WMM) to create a map of water depth across the wetlands
of South Florida with over 3000 separate values within each 2 x 2 mile
cell (using topography based on the 28.5 meter resolution of a LandSat
image).
The first step in calculating high resolution hydrology is processing
the output of the Water Management Model. This output is provided as
daily values of water level for each of the approximately 1700 2 x 2
mile cells of the WMM. For the simulation period of 1965-1995 this
output is an 88 megabyte file. These data are provided as distance of
the water surface from the ground surface (+ or -) and so require a
separate file of surface elevations to be converted to absolute
elevation above mean sea level. These water surface data are converted
to water volume per 2 x 2 mile cell by setting the basal surface of the
volume occupied by water in each cell at an elevation 20 meters below
the ground surface elevation. This elevation was selected because
analysis of WMM output for the 31-year simulation period showed that in
no cell did the water level ever exceed 20 meters below ground surface,
nor 20 meters above ground surface. Thus, each cell has a base
elevation for calculation of water volume that is set in relation to
its surface elevation. This base elevation is constant throughout the
31 year period. The 40 meter range allows water surface elevations to
be stored at 1 millimeter resolution using a "short integer" format in
C++.
Because the water surface elevation can be either above or below the
ground surface in any cell, water volume calculations must be
subdivided into below and aboveground components. Above the ground
surface, standing water occupies 100% of the volume defined by the
following equation: volume = (water level - ground surface) x the
surface area of the cell. Below the ground surface, ground water
occupies only a fraction of the total volume, since sediment or bedrock
occupy most of the volume (that is, if the ground is "solid"). In the
WMM, the "water storage capacity" of the bedrock is modeled as varying
from region to region across South Florida, but most values are close
to 0.2. Thus, ground water is 20% of the volume defined by the
following equation: volume = (groundwater level - basal elevation) x
the surface area of the cell. When the water level is above the ground
surface, calculation of total water volume has both an aboveground and
a belowground component.
The ATLSS High Resolution Hydrology Model calculates the volume of
water predicted by the WMM for each 2 x 2 mile cell for each day of the
31-year simulation period. The primary WMM output file used is:
daily_stg_minus_lsel.bin. The WMM input file from which the bedrock
water storage capacities are obtained is: statdta.int95_2. WMM
data values in feet are converted to metric for use in ATLSS. Thus, the
31 year file of daily water level output is converted to a 31 year file
of daily water volume in each of the 1700 cells of the WMM "Map" of
South Florida.
The second step in calculating high resolution hydrology is
redistributing the water volume for each 2 x 2 mile cell over the
irregular topographic surface of the ATLSS pseudotopography.
Pseudotopography is used to replace the completely flat surface of the
WMM 2 x 2 mile cell with an undulating surface that corresponds to the
topography underlying the vegetation (i.e., the highest areas are
hardwood hammocks or pine stands, and the lowest are open water or deep
marsh vegetation). The same basic subdivision of volume calculations by
surface and subsurface water that was used to calculate water volume
from the water surface elevations of the WMM is used to convert the
water volumes back into surface elevations of water on the undulating
landsurface of the pseudotopography. The water surface is assumed to be
level across each 2 x 2 mile square of pseudotopography landscape.
Consequently, for any particular water surface elevation, some fraction
of the total water volume will be surface water (assuming the water
level is above the lowest point on the land surface) and the rest will
be subsurface water. The ATLSS High Resolution Hydrology Model
calculates the water surface elevation for any specific water volume by
"balancing" the surface and subsurface volumes to equal the total water
volume produced by the WMM. This "high resolution hydrology" can
potentially estimate water depths for each 28.5 x 28.5 meter area
within the region covered by the WMM, rather than a single water depth
for each 2 x 2 mile area. Thus, the ATLSS High Resolution Hydrology
Model converts a low resolution map of 1700 surface water elevation
values into a high resolution map with as many as 5.5 million surface
water elevation values within the same total area.
The ATLSS High Resolution Hydrology Model provides water depth
estimates at spatial scales that are relevant to the vegetation and
wildlife species of the Everglades and Big Cypress. Nonetheless, this
level of resolution (28.5 m) is not fine enough to detect certain
biologically important features, such as alligator holes. For other
purposes, such a high resolution is not necessary. Consequently, some
of the ATLSS animal models use water data that has been aggregated to
100m or 500m cells, which are based on averages of the 28.5 meter
data.
Potential Errors in High Resolution Hydrology Calculations
There are three primary sources of error in the daily water depths
calculated by the ATLSS High Resolution Hydrology model: 1) Errors in
the pseudotopography base map used to calculate water depth; 2) Errors
in the daily stage height output of the Water Management Model; and 3)
Discrepancies between the structure of the WMM and the actual physical
structure of the South Florida Landscape, as reflected in the
vegetation map and the derived pseudotopography. Errors in the base
pseudotopography can result from several sources. Obviously, errors in
the satellite-based vegetation map will result in errors in any
products derived from that map. However, while this map certainly
contains a number of misclassified 28.5 x 28.5 m cells, the overall
pattern of vegetation is consistent with other maps of South Florida
vegetation. So these errors should have relatively little impact on the
overall validity of pseudotopography. A second source of error is the
hydroperiod parameters used to generate pseudotopography from the
vegetation map. These parameters are based on values reported in the
literature, and are consistent with the general topographic positions
of the major vegetation types found throughout the South Florida
wetlands.
More serious sources of error for both pseudotopography, and all water
calculations are 2) and 3), listed above. Wherever the output of the
WMM does not closely match the actual hydrograph that occurs (or would
occur) in a particular location, both pseudotopography (based on the
"calibration-validation" runs of the WMM) and high resolution hydrology
may be incorrect at that location.
It is important to note that there is a possibility that high
resolution hydrology (HRH) may in some situations actually decrease the
errors present in WMM output. This occurs because HRH, specifically the
underlying pseudotopography, force the shape of the landscape to create
hydroperiods appropriate for the vegetation types that are present,
while it preserves the water volume predicted by the WMM. For example,
where the WMM predicts a long hydroperiod in an area where the
vegetation maps shows only short hydroperiod vegetation types (such as
pine and Muhlenbergia), the HRH pseudotopography algorithm will raise
the surface of the landscape sufficiently that the appropriate
hydroperiods will be experienced by the vegetation. In such a case, the
bulk of the water volume will be stored as subsurface water.
Errors in WMM output are particularly critical where they result in
predicting more favorable conditions than those that would actually
occur. This would cause all derivative model runs to overpredict the
population density of affected species. In critical habitats, it is
important that WMM output (specifically the calibration-validation runs
on which all empirical model validation will be conducted) be checked
against measured stage height data.
The third major source of error results from the inevitable mismatch
between a relatively coarse scale model (the 2 x 2 mile grid structure
of the WMM) and the actual fine-scale patterns of vegetation and
control structures on the landscape. The grid structure of the WMM
requires that physical structures, such as levees, be represented as
occurring along the edges of the 2 x 2 mile cells, which produces a
"stair-step" approximation of the actual linear structure in the
wetlands. This creates a problem for HRH when vegetation on the
"down-gradient" side of a levee within a 2 x 2 mile cell is modeled as
having water based on the "up-gradient" side of the levee. This problem
occurs primarily along the boundaries of the Water Conservation Areas
(and is discussed in the file on the Boundaries of the Reporting Units
used by ATLSS). This problem results in HRH being incorrect in those
areas where WMM boundaries do not coincide with actual levees and/or
canals. To some degree, HRH and pseudotopography adjust for this type
of misalignment by alterning the topographic surface of the landscape
so the vegetation experiences the appropriate hydroperiod under
calibration conditions. However, this results in a discrepancy between
the water depth and hydroperiod predictions of the WMM raw output and
the water depth and hydroperiods predicted by the High Resolution
Hydrology. In such areas of discrepancy, there is a good probability
that the HRH results are more accurate that the raw WMM output, but
this should be checked against actual measurements of topography and
hydrographs wherever possible.
One potential solution to the problem of the mismatch between WMM model
boundaries and the actual boundaries on the landscape is to
post-process the WMM output into the irregularly-shaped cells that
correspond to the areas of discrepancy. This would be a substantial
programming effort, but would solve this major source of error for all
future model runs.
=====================================================================
Performance measures associated with the ATLSS Hydrology model.
In accordance with the ATLSS file naming conventions,
each file name will consist of the characters:
"X" or "_" => the Base, typically F for the F2050 base
or E for the C1995 base
"X" or "_" => the alternative scenario or base
"HY" => the ATLSS Hydrology Model
"XXXX" => 4 character mnemonic
"."
"PDF" or "TXT" or "DOC" => PDF, tabular text or documentation
1. Maps
The comparison maps for the ATLSS Hydrology model reflect a
graphical representation of the hydroperiod data found in the
accompaning tables. For a detailed discussion of the hydroperiod
data, see the description of the comparison tables below.
The comparison maps associated with the ATLSS Hydrology model
consist of the following files:
XXHYHCM1.TXT Comparison map set showing the total area by
hydroperiod classes averaged over the years 1985
to 1995 grouped by subregion.
XXHYHCM2.TXT Comparison map set showing the total area by
hydroperiod classes averaged over the five driest
years grouped by subregion.
XXHYHCM3.TXT Comparison map set showing the total area by
hydroperiod classes for the driest year grouped
by subregion.
XXHYHCM4.TXT Comparison map set showing the total area by
hydroperiod classes averaged over the five wettest
years grouped by subregion.
XXHYHCM5.TXT Comparison map set showing the total area by
hydroperiod classes for the wettest year grouped
by subregion.
XXHYHCM6.TXT Comparison map set showing the total area by
hydroperiod classes for the average year grouped
by subregion.
========================================================
2. Time Series Charts
None.
========================================================
3. Histograms
None.
========================================================
4. Tables
The high resolution hydrology analysis for the Across Trophic
Level System Simulation ( ATLSS ) consists of a set of tables
which compare two hydrology scenarios as provided to our research
group by the South Florida Water Management District ( SFWMD ).
The comparisons are broken down into two basic types of files.
The first type contains hydroperiod analysis for the vegetation
types. The other contains an report on the area in each of 10
hydroperiod classes. These analyses are repeated for each of
several regions in South Florida (SF) and for a variety of years
or combinations of years.
The comparisons are carried out on a variety of subregions of SF.
These regions represent major control and management areas of SF,
such as Loxahatchee, the water management areas, and management
regions with in The Everglades National Park ( ENP ) and Big
Cypress National Preserve ( BCNP ).
The comparisons are also broken in to a number of temporal units.
The current time groupings are a ten year average from 1985 to 1994,
an average of five years which had the highest rain fall ( currently
1966, 1968, 1969, 1982, 1983 ) an average of five years which had
the lowest rain fall ( currently 1971, 1981, 1988, 1989, 1990),
the year with the highest rain fall ( 1969 ) the year with the
lowest rain fall ( 1990 ) and the year with an average rain fall
( 1977 ).
The first type of file provides hydroperiod comparison for the
vegetation type. Within each region the average hydroperiod
is computed for each vegetation type which covers at least
10% of the area of that region. At the top of each table is the
name of the region and the total area of that region.
Each row of a tables represents the values for a single
vegetation type. The columns of each row are defined as follows:
The vegetation type index and description, the area of the
region covered in that vegetation type, the hydroperiods for the
vegetation type under the two scenarios, the difference between
the hydroperiods and the percentage difference. The indices represent
those used in the Pearlstine vegetation cover map to represent
the vegetation types. The vegetation type descriptions are
those used by University of Florida Land Cover Classification map
developed in support of the Florida Gap Analysis Program (FGAP).
Hydroperiods have units of days, and the areas are listed in Km.
The second type of table gives the area covered in each
region by each of twelve hydroperiod classes. The top of the
table gives the region name and the total area of the region.
Each row of the tables contains the information about a single
hydroperiod class. The columns of each row are defined as follows:
The first column contains the range of hydroperiod which defines
the hydroperiod class, the next two are give the area of the
listed region which has a hydroperiod with in the range of values
for that class for each of the scenarios being compared and the
last column gives the difference between these two values.
The mnemonic section of the file names are composed as follows:
"XX" = P1, P2, P3, P4, P5 => the Time Period
"XX" = HC, VT => Hydrology Class, Vegetation Type
The performance measure tables associated with the ATLSS
Hydrology model consist of the following files:
File Name Description
------------ ------------------------------------------------
XXHYVTP1.TXT Comparison tables set showing the total area of
significant vegetation types averaged over the
years 1985 to 1994 grouped by subregion.
XXHYVTP2.TXT Comparison tables set showing the total area of
significant vegetation types averaged over the
five driest years grouped by subregion.
XXHYVTP3.TXT Comparison tables set showing the total area of
significant vegetation types for the driest year
grouped by subregion.
XXHYVTP4.TXT Comparison tables set showing the total area of
significant vegetation types averaged over the
five wettest years grouped by subregion.
XXHYVTP5.TXT Comparison tables set showing the total area of
significant vegetation types for the wettest year
grouped by subregion.
XXHYVTP6.TXT Comparison tables set showing the total area of
significant vegetation types for the average year
grouped by subregion.
XXHYHCP1.TXT Comparison tables set showing the total area by
hydroperiod classes averaged over the years 1985
to 1995 grouped by subregion.
XXHYHCP2.TXT Comparison tables set showing the total area by
hydroperiod classes averaged over the five driest
years grouped by subregion.
XXHYHCP3.TXT Comparison tables set showing the total area by
hydroperiod classes for the driest year grouped
by subregion.
XXHYHCP4.TXT Comparison tables set showing the total area by
hydroperiod classes averaged over the five wettest
years grouped by subregion.
XXHYHCP5.TXT Comparison tables set showing the total area by
hydroperiod classes for the wettest year grouped
by subregion.
XXHYHCP6.TXT Comparison tables set showing the total area by
hydroperiod classes for the average year grouped
by subregion.