| I
'TA245 . 7
873

0.1603

f"\

  Inventory of Ponds in the Brazos

and Colorado River (Texas) Drainages,

from NASA Color Infrared Photography

THE TEXAS AGRICULTURAL EXPERIMENT STATION / Neville P. Clarke, Director / The Texas ASLM University System /College Station, Texas

CONTENTS ACKNOWLEDGEMENT

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i Special thanks to Mrs. Marylin Snell for maintain- V

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 ing the highest level of responsibility, initiative, and

Materials and Methods . . . . . . . . . . . . . . . . . . . . . 1 accuracy throughout the tedious work of this study.

Results and Discussion . . . . . . . . . . . . . . . . . . . . .3 Funding was provided in part by the Texas ASM

Statistical Analysis . . . . . . . . . . . . . . . . . . . . . . . .5 University Remote Sensing Center.

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Literature Cited . . . . . . . . . . . . . . . . . . . . . . . . . 12

Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
SUMMARY

NASA color infrared photography was used to inventory the farm
ponds of the Colorado and Brazos River basins in Texas. Ponds in ran-
domly selected blocks within larger photographic frames were counted
and placed in size classes. ln the 65 photographic frames examined of
the Colorado River basin 12,367 ponds were identified, and 2,598 ponds
were identified in the 7 frames examined of the Brazos River basin.

Ponds were assigned to five size classes: class 1, < 1.25 ha; class 2,
1.25-2.5 ha; class 3, 2.5-5 ha; class 4, 5-11.25 ha; class 5, > 11.25 ha.
Densities in ponds per kmz by size class for the Brazos and Colorado
River basins respectively were: class 1 - 1.7, 0.97; class 2-0.1, 0.06; class
3-0.02, 0.01; class 4-0.003, 0.003; and class 5-0.002, 0.004.

In the Colorado River basin 92.8 percent of the total ponds counted
fell into size class 1 ( < 1.25 ha) and 5.6 percent of the total count fell
into size class 2 (1 .25-2.50 ha). ln the Brazos River basin these two size
groups constituted 93.3 and 5.5 percent of the total ponds counted
respectively. Counts in all other size groups were 1 percent or less of
the total.

The Colorado River basin pond densities (number per unit area)
tended to be low ( > 1.5 pond/ha) in the immediate vicinity of the Gulf
Coast, and in the Texas High Plains, with a wide range of densities
(1 .5-3.25 ponds/ha) in the region between the Coastal Plain and the High
Plains. ln the Brazos River basin, densities tended to increase with
distance from the coast. The counting blocks were from 2.1 to 2.3
kmz wide, and there was no significant change in pond density with
distance from the river channel at this resolution.

ln a 10 factor stepwise regression model, the factor which explained
the most variation in pond count density was average slope, (here de-
fined as the maximum elevation in the county minus the minimum
elevation) followed by population density and rainfall, in order of im-
portance. Using the simplifying assumption that all ponds have water-
sheds which are hydrologically independent, we calculate that in 28 per-
cent of the frames, over one half of land surface would have to be in
pond watershed to maintain the pond surface area in the frame. The
data also suggest that there may be significant differences in pond den-
sity among soil associations ranging from 3.028 i 0.197 ponds/kmz in
the Castell-Ponotoc-Lignon Soil Association to 0.172 .4; 0.123 v
ponds/kmz in the Amarillo-Acuff-Mansker Soil Association.

Llsing pond density data from this study it is estimated that the
number of ponds 1.25 ha and smaller is 102,000 for the Colorado River
basin, and 191,000 for the Brazos River basin. lf the mean density for
the two watersheds is applied to the State of Texas, with the Rio Grandex/
River basin excluded, there were an estimated 782,000 ponds 1.25 ha
and smaller, and an estimated 840,000 ponds of all sizes in the state,
as of 1970.

Inventory

of Ponds in the Brazos and

Colorado River (Texas) Drainages,
From NASA Color Infrared Photography

AUTHORS

WILLIAM J. CLARK, Professor, The Texas Agricultural Experiment Sta-
tion, (Department of Wildlife and Fisheries Sciences), College Sta-

tion, Texas

TIMOTHY A. SPRINGER, Research assistant, The Texas Agricultural Ex-
periment Station, (Department of Wildlife and Fisheries Sciences),
College Station, Texas

INTRODUCTION

Small ponds are abundant in
central and eastern Texas. Most
are fed by runoff, a few by pumped
ground water. These ponds are
usually under landowner control
and are small enough to be
managed for the objectives of the
owner.

The extent of the resource is
poorly known. Acreage data for
ponds larger than 2 acres but
smaller than 4O acres have been
published by the Soil Conservation
Service for 1958 (457,901 acres)
and 1967 (456,549 acres) (SCS
1970). Although the reported
areas show a slight decrease from
1958 to 1967 some counties show
drastic and unexplained decreases
which call the quality of the data
into question: Denton County
decreased from 36,200 acres to
2,049; Andrews County from
2,000 to 10; Cottle County from
5,000 to 0. Pond numbers were
reported for the first time in the
1977 SCS National Erosion Inven-
tory Estimates, (SCS 1978) where
Texas is listed with 126,881 ponds
smaller than 40 acres, with total
pond surface area of 409,000
acres.

The SCS also reports another
estimate of pond numbers in
Texas, which is a cumulative count
of the number of ponds for which
the SCS has participated in design
or construction. This figure was
299,038 as of 1981 (per. comm.,

J. W. Hill, Assistant State Conser-
vationist for Water Resources). A
much more intensive inventory by
the SCS is currently under way.

The current study is based on
color infrared photographs taken
by a NASA aircraft. The film con-
sists of continuous frames of the
Colorado River basin taken on
three flight lines between the river
mouth at the Gulf of Mexico and
the headwaters near the New Mex-
ico border. A smaller number of
frames taken on one flight down
the Brazos River watershed is also
analyzed. The film, along with the
flight documentation provided by
NASA, provides an excellent op-
portunity to investigate the rela-
tionship of pond distribution to a
variety of parameters.

MATERIALS AND METHODS

The photographs analyzed are
from Mission 123, Test Site 213,
Colorado River, taken on 13 March
1970. The photographs were taken
using an RC-8 camera with a 6-
inch focal length lens, 9-inch color
infrared film (SO-117), and filter
No. 12 (500-nm). The roll exam-
ined was a duplicate of the original
positive transparency film. The roll
contained 195 frames of the Col-
orado River watershed and 22
frames of the Brazos River water-
shed. Contiguous frames overlap-
ped approximately 30 percent and
every third frame was examined,
with 65 frames examined in the

Colorado River watershed and
seven frames in the Brazos River
watershed. The flight paths are
shown in Figure 1.

The film was examined on a
light table equipped with a
movable stereomicroscope. A
clear plastic overlay was con-
structed which divided the central
200 >< 200 cm area of the frame
into 100 blocks, each 2 cmz,
making 10 rows of 10 blocks each.
In eight frames spaced at intervals
along the watershed, all 100
blocks were examined. For the
majority of the frames, three of the
10 rows were selected at random
and all 10 blocks in each row were
examined. The rows were at right
angles to the river. In all, 2,370
blocks were examined. Flight
altitude and surface elevation
varied along the flight paths, with
the area covered by the counting
grid varying from 450 to 529 kmz
(174 to 204 square miles). This
made the block, or counting unit,
2.12 to 2.30 km on a side. Flight
book data permitted the calcula-
tion of actual plane to ground
distance for each frame, and area
conversions were calculated for
each frame to correct count data
used for density estimates and
statistical procedures. The ponds
counted were scored, by observa-
tion, into five size classes by com-
paring them to a set of reference
areas of known size. (Table 1).

The number of blocks between
the examined block and the river

1

I I I T T I I 6 I I T I
-36°
“WM TEXAS
  -<
RED
-34°
“"1
-32°
Co 0fAso
i- RIO GRANDE
*3“ A’
I
_ ' NECHES-
TRINITY -
_ \ SAN JACINTO-
h(  BRAZOS
“l?” I iléléiiBo
4
~  COLORADO- _.
_ LAVACA
28o LAVACA-—
GUADALUPE
SAN ANTONIO—
NUECES _‘
>- NUECES-
RIO GRANDE
-26°
1 06° 104° 102° 100° 98° 96° 94°
l l l l l l l l l l l l l

   

 

Figure 1. Watersheds of Texas rivers, with lines showing flight paths of NASA Mission 123. Sections of flight

path in the Brazos River basin where photographs were taken are shown by brackets.

was also recorded. The photo-
graphic frames used were located
on the Geological Highway Map of
Texas (Renfro and Feray 1973) and
on the General Soils Map of Texas
(Godfrey, et al. 1973) using the
location data provided on maps in
the flight information publication.
Each frame counted was then
classified as to geology and soil
type using the following scheme:
1. Geology
a. The color coded categories
of the geological map
(basically rock systems) were

used as the units of
classification, and each
system present in the block
was recorded.
b. For each frame the following
were recorded:
1. dominant formation in
frame,
2. subdominant formation
in frame, and
3. number of formations
found within the frame.
2. Soils
a. The color coded categories
of the soil map (major soil

associations) were used as
units of classification.
b. For each frame the following
were recorded:
1. dominant soil associa-
tion in frame,
2. subdominant soil asso-
ciation in frame, and
3. number of soil associa-
tions found within the
frame.
The following information was
recorded for each county within
which frames were censused:
1. Average rainfall (Pass 1980)

vnfl

. .

TABLE 1. PARAMETERS FOR POND SIZE ESTIMATION
Size of Corresponding‘ Pond Class
Reference Areas Area on the Size Size
Ground (hectares) Classes Ranges
— <1.25 1 < 1.25 ha ( <3.1 acres)

1 mm? 1.25 (1.1-1.3) 2 1.25-2.5 ha (3.1-6.2 acres)
2 mm2 2.5 (2.3-2.6) 3 2.5-5 ha (6.2-12.6 acres)

4 mmz 5 (4.5-5.3) 4 5-11.25 ha (12.6-27.8 acres)
9 mmz 11.25 (10.1-11.9) 5 > 11.25 ha (> 27.8 acres)

‘Flight altitude aboveground varied somewhat, and a mean area value was used in the
calculations. The range is given in parenthesis.

2. Agricultural region (BBR
1976)

3. Cattle numbers (TCLR 1976)

4. Runoff (LISGS 1966)

5. Slope (maximum altitude -
minimal altitude)

6. Irrigated acreage (TCLR
1976)

7. Rural population (LISBC
1971)

8. Land resource areas (God-
frey et al. 1973)

RESULTS AND DISCUSSION

A total of 14,965 ponds was
enumerated, 2,598 in the Brazos
River watershed and 12,267 in the
Colorado River watershed. Distri-
bution by size group is given in
Table 2. Ponds as small as 0.2 ha
could be recognized, and the over-
whelming number of ponds in
both watersheds is smaller than

1.2 ha. The distribution by size is
very nearly the same in the two
watersheds.

Mean densities by size group are
given in Table 3. Mean values for
block area were used in the densi-
ty calculations. Ponds of the two
smaller size groups were approx-
imately twice as abundant per unit
area in the Brazos River watershed
as in the Colorado River water-
shed. However, the area covered in
the Brazos River watershed is just
over 10 percent of that covered in
the Colorado River watershed, and
the two watersheds. differ in
climatic and “soil-influenced”
characteristics. Figure 6a shows
many frames of very low pond
density in the upper Colorado,
while the upper Brazos has its
greatest densities in the upper-
most reaches which were
photographed. No photography

TABLE 2. POND DISTRIBUTION BY SIZE CLASS, NUMBER, AND PERCENT FOR THE
AREAS SURVEYED IN THE COLORADO RIVER AND BRAZOS RIVER WATERSHEDS

Watershed Pond Size Class Total Ponds
1 2 3 4 5
(hectares) <1.25 1.25-2.5 2.5-5 5-11.25 > 11.25
(acres) <3.1 3.1-6.2 6.2-12.4 12.4-27.8 > 27.8
Colorado No. 11,477 693 120 30 47 12,367
% 92.8 5.6 1 0.2 0.4
Brazos No. 2,424 142 25 4 3 2,598
% 93.3 5.5 1 0.2 0.1
TOTAL No. 13.901 835 145 34 5O 14,965
% 92.9 5.6 1 0.2 0.3

TABLE 3. MEAN POND DENSITIES (PONDS/kmz) BY POND SIZE FOR THE AREAS
SURVEYED IN THE COLORADO AND BRAZOS RIVER WATERSHEDS

/*'\ Watershed Pond Size Class Total Ponds
1 2 3 4 5

(hectares) <1.25 1.25-2.5 2.5-5 5-11.25 > 11.25

(acres) < 3.1 3.1-6.2 6.2-12.4 12.4-27.8 > 27.8
Colorado 0.97 0.06 0.01 0.003 0.004
Brazos 1.7 0.1 0.02 0.003 0.002

was taken in the upper part of the
Brazos River basin, where land use
would be similar to that of the up-
per Colorado River basin, and
where pond densities would pro-
bably be similar. The highest den-
sities per frame found in the study
were in the Central Colorado
drainage. The pond numbers in
the two larger size classes are too
small to permit meaningful
comparisons.

In an unpublished study of
Brazos County ponds, where all
ponds from SCS photography,
were counted, the pond density
was 0.93 ponds/kmz, in general
agreement with densities reported
in this study.

Frequency distributions of
counts per block are given in
Figures 2 to 4. Counts of over one
pond per block were rare for the
three larger size classes in either
watershed, and are not shown. For
size class 2 (1.25-2.5 ha) both
watersheds had one instance each
of 5 and 6 ponds per block, other-
wise counts were predominantly 1
or 2 ponds per block.

The pattern of distribution of
size class 1 pond counts differed

‘significantly between watersheds.

For the Colorado River basin (Fig.
2), the frequency declined ex-
ponentially from 1 to 22 ponds per
block, and then remained fairly
constant. For the Brazos River
basin (Fig. 3), the frequency of
counts remained quite stable from
1 to 14 ponds per block, and then
declined from 14 to 27 ponds per
block, with the maximum density
reported at 38 ponds per block.
There is no ready explanation for
the difference in frequency
distribution; more detailed infor-
mation on size of land holdings,
differences in land use and other
demographic data might suggest
reasons. Of the 26 blocks with
counts ) 30 in the Colorado River
basin, 20 came from three frames
in Llano and San Saba Counties.
Two of these blocks had the
highest pond counts observed (74
and 92 ponds). They included
eight class 5 ponds () 11.25 ha),
more than any other frame and 17
percent of the total class 5 ponds
counted. The number of blocks
with no ponds was twice as great

3

Logm of Frequency

Pond Size Class 1 (<1.25 ha) Colorado River

11,477 Ponds

i i l l

I. I l

15 2O 25 3O 35
Ponds per Block

Figure 2. Frequency of counts of ponds per block, for pond size class
1 ( < 1.25 ha), in the Colorado River watershed.

Logm of Frequency

Pond Size Class 1 (<1.25 ha) Brazos River

Ponds per Block

2,424Ponds

1 1 1 1 1 1 1 1
1b
o. ' 9. ' . o

O
_ ' ' . Q . _
O O
O
OO
O
O O
O .O

1 1 1 1 o1 Q 1 1 1

O 5 10 15 20 25 30 35 40

45

Figure 3. Frequency of counts of ponds per block for pond size class
1 ( < 1.25 ha), in the Brazos River watershed.

in the Colorado River basin as in
the Brazos River
distribution ofcounts for size class
2 ponds (1.25-2.5 ha) was similar
in the two watersheds (Fig. 4).
Variations in pond density with
distance from the Colorado River
are shown in Figure  Assignment
to the first category (one block
from river) means that the river
was in the block being examined.
Since a block was from 2.12 to 2.3
km across, any change in pond
density near the river must occur
in a fairly narrow band or the ef-
fect would be evident in Figure 5.
However, no trends are apparent.
The flight path of the
photographic aircraft was more or
less at a right angle to the Gulf
Coast. Densities of ponds versus
distance from the coast are
presented in Figure 6 as average
densities per frame. The data set
for the Brazos is small, but the
data are internally consistent and
suggest significant differences be-
tween the two basins in pond
distribution for size class 1 ponds
((1.25 ha) and to a lesser extent
for size class 2 ponds (1 .25-,2.5 ha).
There were no significant trends in
pond count with distance for size
class 3 ponds (2.5-5 ha), and there
were too few size class 4 or 5
ponds for reliable analysis. The
data suggest that the parameters
controlling the distribution of
smaller ponds( (2.5 ha) differ from
those controlling larger ponds
() 2.5 ha). Differences in counts
between watersheds are greatest
for the smallest ponds. For the
Brazos there is a consistent in-
crease in pond density with
distance from the Gulf. The Col-
orado River basin is more com-
plex. Like the Brazos, densities are
low near the coast, perhaps limited
in both watersheds by topography
and land use. Away from the Gulf
the Colorado densities first in-
crease, and then in the middle
distances show an extremely er-
ratic distribution, coincident with
the southern Edwards Plateau. The
geology here consists of fractured
and porous limestone and has high
relief, both of which interfere with
pond construction, except in
stream valleys where ponds are
concentrated. ln the upper part of

basin. The g

Pond Size Class 2 (1 .25-2.5 ha)

I
Colorado
3 693 Ponds
a o
> __
2 21
m I
:1
Q‘ _
a:
\-
E I!
"6 I
=> __
T-
m
o
J fi
1--
_ O
J
111119“?
O 2 4 6

Brazos
_ 142 Ponds
_ O
2-4
I
_ O
1_.
—< Q
I I I I WM’
0 2 4 6

Number of Ponds

Figure 4. Frequency of counts of ponds per block for pond size class
2 (1.25 - 2.5 ha), in the Colorado and Brazos River watersheds.

the Colorado (High Plains) rainfall
is low, evaporation high, and few
ponds can persist. ln contrast, the
Brazos River basin, at the upper-
most reaches photographed, has
higher rainfall, and land use pat-
terns are morfe favorable for pond
construction. Pond densities in the
extreme upper Brazos River basin
would probably be similar to the
upper Colorado River basin.

The amount of watershed area
that would be required to support
the ponds in each frame was
estimated using data from TSPE

(1974) and pond density data from
this study. The calculation pro-
cedure is given in Appendix A.
Figure 7 shows the fraction of the
surface area which would be re-
quired to support the observed
density of ponds of the size in-
dicated, if each pond’s water sup-
ply were hydrologically indepen-
dent from the others. (Jnder this
assumption, 28 percent of the
frames would require more than
half of their surface areas to be in
pond watersheds to maintain the
pond surface they contained. Pond

densities for the two larger size
classes were to low to show clear
trends. For the Brazos River basin
the fraction of land surface re-
quired to support size class 1 and
2 ponds closely follows the densi-
ty distribution from the coast in-
land. For size class 3, the fraction
of land surface required shows a
trend of increase as distance from
the cost increases, though the
count density does not. The size
class 3 densities are uniformly low.
For the Colorado River basin, the
fraction of land surface required to
support size class 1 ponds also
follows the density distribution
along the watershed. The fraction
of surface area required to support
size class 2 ponds parallels pond
density in the lower reaches, but
shows an inverse relationship with
pond number in the upper part.
This reflects the increase in
evaporation losses and low
precipitation in the upper part of
the basin. Low count densities
obscure any relationship between
counts and the fraction of water-
shed required to support size class
3 ponds. Note that if a large per-
centage of the ponds intercept the
outflow from other ponds, then the
estimate of the fraction of the land
surface area which is in watersheds
is inflated. Hence, it is possible to
estimate that more than 100 per-
cent of the available surface area
would be required to maintain the
ponds.

STATISTICAL ANALYSIS

The statistical characteristics of
the distribution of ponds vary" con-
siderably between photographic
frames. We calculated coefficients
of variation (C.V.) and the 95 per-
cent confidence intervals (C.l.)
around the means of pond den-
sities in photographic frames
where all 100 blocks were
counted. The results are given in
Table 4.

Sampling errors are fairly large
relative to the densities being
measured. The skewness (shifting
of the peak of the distribution right
or left) and kurtosis (a measure of
the “heaviness” of the tails of the
distribution) of the underlying
distributions were calculated. Data

5

1.6

1'4 7 s06 241

383

1.0 —

2

0.4 —

Ponds per km
O
CD
I

0.0

200

153

81

52

112

1 2 3

Blocks from the Colorado River

Figure 5. Pond density versus distance from the river for the Colorado
River watershed. Numbers above bars are number of ponds.

TABLE 4. CONFIDENCE INTERVAL OF THE MEAN AT THE 95% CONFIDENCE LIMIT
AND COEFFICIENTS OF VARIATION OF COUNTS OF ALL PONDS IN FRAMES WHERE

ALL 100 BLOCKS ARE COUNTED

4

5

6 7

Frame Mean Lower Upper C.V.
Number Limit Limit °/o
Der
Block

1982 .432 .356 .507 88
2012 .572 .466 .678 93
2042 1.250 1.131 1.371 48
2072 1 .684 1 .488 1 .880 59
2101 .412 .331 .493 99
2126 1.970 1.793 2.147 45

drawn from a normal distribution
have skewness of O and kurtosis of
3; the values of these parameters
for pond densities in each frame
are given in Table 5.»

These statistics indicate that the
sampling distributions of the
ponds are skewed to the left but
show no regular trends with regard
to kurtosis. Some caution must be
used in performing parametric
statistical tests since the counts of
ponds in the blocks within the
frames do not seem to be normal-
ly distributed, and it seems unlike-
ly that any normalizing transfor-
mation would be uniformly
appropriate.

6

Stepwise multiple regression
analysis was used to examine the
relationships between the various
factors and the pond densities. The
stepwise regression procedure was
set up to sequentially report the
best models for up to six
parameters. It should be noted that
the data set treated here is fairly
large (72 frames). When inter-
preting the significance of any
regression based on a large data
set, a highly significant F statistic
may be obtained even though the
correlation coefficient may remain
small. The small correlation coef-
ficient indicates that only a small
fraction of the variability of the

dependent variable is explained;
the significant F statistic simply in-
dicates that spurious correlations
are unlikely and even a small cor-
relation probably indicates a real
relationship.

The stepwise models used the
following factors as independent
variables: (1) rainfall; (2) runnoff;
(3) fraction of land in irrigated
acreage, (4) slope, (5) soil types, (6)
geologic types, (7) cattle density,
(8) rural population density of
county; (9) density of habitations
in frame, and (10) distance from
the coast. The dependent variable
was density of ponds in a frame of
both the Colorado and Brazos
River basins. The regression was
carried out using the maximum R-
square as the criterion for inclu-
sion of a variable in the model.
Table 6 defines the variables used
in the analysis, and Table 7 shows
the model selected at each step of
the regression procedure. The
regression for all ponds of less
than 11.3 ha (Table 7) was signifi-
cant at each step, but the coeffi-
cient of determination (R-square)
ranged between only 0.12 and
0.37. Since the independent
variables, cattle density, and rain-
fall for example, are not indepen-
dent of each other, they tend to
contribute some of the same infor-
mation to the model. Thus, as
more variables are added to the
model, the significance of a par-
ticular factor may decline as the in-
formation it contains becomes
redundant. If this occurs, then the
factor may be dropped from the
model, as it was when cattle den-
sity was replaced by runoff in step
5 of the regression.

lt is informative to compare the
probabilities of occurrence of
greater F for each of the variables
in the six variable model in order
to evaluate the relative contribu-
tion of each to the prediction of
the number of ponds in a frame:
The smaller the probability, the
less likely it is for the relationship
to have occurred by chance, and
the greater the predictive value of
the variable. In this case, the fac-
tor which contributed most to
variability of total pond density ap-
pears to be the slope factor, which
is simply the difference between

Ponds per km

_ 3.5

3.0
2.5
2.0

1.5-

1.0
0.5

3.15
3.0
2.5
2.0
1.5
1.0
0.5

0.0
0.3

0.2

0.1

.006

.033

' A. All Ponds
_ o 0 '
o
~ . o o’. A 0 Colorado
- .. , A A A Brazos
..' . o o A
~91!- AA "f. a! '
o
, , .4- Jr
- ' o O B. Size Class 1 (<1.25 ha)
o
_ ' 9 . oo A
o0 o A A
>- . . A
Io 0' o .' o‘
~0 , o o
’ 6O ‘.02 Q a
I .1 ,l"hfif
o C. Size Class 2 (1.25 - 2.5 ha)
~ o
oo .
Q A A
l‘ o '8 . t’ o
oz ' ' ' f
' “A '. 000 . o
0.0+ " '1 n’ 1'1 ""PI'MLI»_
D. Size Class 3 (2.5 - 5.0 ha)

_ o

o o ' o

- A

 

720

900

km From Gulf o1 Mexico

Figure 6. Pond density versus distance from the Gulf of Mexico by pond
size, for the Colorado and Brazos River watersheds.

TABLE 5. SKEWNESS AND KURTOSIS OF THE DISTRIBUTIONS OF COUNTS IN
FRAMES WHEN ALL 100 BLOCKS ARE COUNTED

Frame Skewness Kurtosis
1982 0.89 0.59
2012 1.28 1.39
2042 0.43 -O.40
2072 0.7 0.83
2102~ 2.10 7.13
2126 0.26 0.36

the minimum and the maximum
altitudes in the county, and is a
gross measure of topographic
relief. The next most important
variable in the regression model is
population density in the county of
the frame. The contributions of all
variables to the regression were
significant at the a=0.05 level,
which indicates that each of the
factors included in the model has
a significant influence on the
number of ponds constructed, and
should be retained.

It is important to remember that
in a multiple regression model
such as the one developed here,
the effect of each variable can be
thought of as being independent of
the others. That is, the effect due
to a factor such as population den-
sity is equivalent to the contribu-
tion which would be obtained if the
population density were changed
while cattle density was held con-
stant. Thus, the importance of the
slope factor may reflect the dif-
ficulty of constructing ponds in
areas where there are no valleys
that can be dammed to form the
lake basins, independent of any
relation to population density. The
high level of significance of
population density itself probably
reflects a situation in which a high
population density is associated
with small individual land
holdings, where each landowner
constructs a pond of his own to
gain access to water for
agricultural, recreational, and
aesthetic purposes. The best single
variable model had cattle density
as its independent variable, and
while the regression was signifi-
cant, only 12 percent of the
variability in pond density was ex-
plained. Cattle density was
relatively unimportant in the six
variable model, indicating that it is
related to the other factors in the
model in much the same way as
pond density.

Different sized ponds tend to be
built for different purposes. Small
ponds are adequate for cattle
watering, but ponds which are to
be used as a source of irrigation
water must have larger volumes.
Thus, we decided to carry out the
stepwise regression on the densi-
ty of each of the different size

7

‘"5 A. All Ponds A
12 0 Colorado
A A Brazos

0.0 0 0 A

0.5 Q A ' ' ’.

Q O
0.3  0 :1. ‘u '0
Q Q Q QQQ
0 ' r ' Jr I’ T .
1.2 _
B. Size Class 1 ( 1.25 ha)
0.9 A
o A
.5 0.6 0O‘ QQ A A
2 ' 0 ‘a
g 0.3 pa’ ' 0'0‘
. Q .
1; I m. M ' ,Qq#'
g 0 0 , Q ,
Z c. Size Class 2 (1.25 - 2.5 ha)
.9
g 0.2. A A 0 A
Q . Q
"0 g 0
O.1~ I Q Q Q
Q Q fl Q
A.! ' '
p. Q.‘ Qr h Q f
0.0 '
Q
D. Size Class 3 (2.5 - 5.0 ha)
0.2- '
. . o A
' 0 a ' Q n A ' '
00_ ‘Al’, 00$. A ' .
' 0 100 s00 s40 120 900

Km from Gulf of Mexico

Figure 7. Fraction of the watershed in pond watersheds, by pond size,
for the Colorado and Brazos River watersheds.

TABLE 6. VARIABLES USED IN THE STEPWISE REGRESSION ANALYSIS

Variable Meaning

CATD Density of cattle in dominant county of frame‘

DK Density of geologic types per unit area

DIS Distance of frame from coast

DN Density of soil types per unit area

FRMX Density of habitations observed in frame

lRR Percent of irrigated land in dominant county of frame‘
POPD Rural population density of dominant county in framez
RAIN Rainfall in dominant county of framef’

RUN Runoff in dominant county of frame“

SLOP Difference between maximum and minimum altitude in dominant countys

‘TCLR (1976).
ZUSBC (1971).

3Pass (1980).
“USGS (1966).

8

classes of the ponds observed in

the survey. Table 8 summarizes

the results fromvthese regressions,
and Table 9 gives the results in
detail for size class 1 ponds.

Since most of the ponds in the
survey fall into size class 1, the
regression model of size class 1
ponds is very similar; to that of the
regression for total pond density.
The density of larger ponds ap-
pears to be more closely related to
hydrologic factors than was the
density of ponds of size class 1. ln
the six variable model for size class
1, the most significant variable was
rainfall, and in the six variable
model for size class 3, the most
significant variable was runoff. The
six variable models for predicting
the density of size class 4 and 5
failed to show significant results.
However, the simple regression of
percentage of irrigated land in the
county on the density of size class
5 ponds is highly significant
(P = 0.0065), which indicates that
the only factor among the
variables considered that was
related to their distribution was
irrigation.

The dominant soil type of a
region is closely related to the
local climate, and influences both
the regional hydrologic properties
and the regional agriculture prac-
tices. Since these same factors are
related to the construction of
ponds, soil associations, as given
on the General Soil Map of Texas
(Godfrey et al. 1973), may be used
as a classification system for
predicting some aspects of the
distribution of ponds. Table 10
gives the soil associations which
were present in the area encom-
passed by the aerial survey, and
the density of each size class of
ponds in each of the soil associa-
tions. Sixteen of the 66 Texas soil
associations were represented in
the sample. lt should be noted that
the scale of the soils map used did
not permit the identification of the
many small scale inclusions of
other soil types within the soil
associations. Subject to the caveat
mentioned, the data suggest that
there may be considerable dif-
ferences between pond densities
for the soil associations, ranging
from a mean density of

TABLE 7. VARIABLE SELECTION DURING STEPWISE REGRESSION OF FRAMEWISE POND DENSITIES FOR PONDS LESS THAN
~<"\, 11.25 HA

STEP 1 2 3 4 5 5 5 5 6
(rrrercear Kare/cam‘ 222222.022 Me/cep/ J/z/e/aap/ J/zza/aep/ J/P/a-‘ffiép/ hie/cap! Mia/cap)
CATD* CATD CATD CATD CATD FIAIN RAIN RAIN RAIN
MODEL SLOP SLOP SLOP SLOP CATD RUN FIUN CATD
FRMX IRR IRR SLOP SLOP SLOP SLOP
FRMX POPD POPD POPD IRR IRFI
FFIMX FRMX FRMX POPD POPD -
FRMX FRMX
F12 0.125 0.205 0.252 0.295 0.313 0.321 0.324 0.352 0.372
PFIOB> F 0.0025 0.0004 0.0002 0.0001 0.0002 0.0001 0.0001 0.0001 0.0001

*Underlined variables were significant at 0z=20.O5. Doubly underlined variables are significant at oz=0.001.

Probability of F values of parameters of best six-variable model

Variable Prob > F
RAIN 0.0179
CATD 0.0206
SLOP 0.0002
IFIR 0.0271
POPD 0.0133
FRMX 0.0191

3.028 i 0.197 ponds (all sizes) per

TABLE 8. RELATIVE IMPORTANCE OF FACTORS IN PREDICTING POND DENSITY BY
SIZE CLASS

kmz in the Castell - Ponotoc - POND SIZE CLASS

Ligon Soil Association to ALL 1 2 3 4 5
0.172 :1: 0.123 ponds (all sizes) per SLOP SLOP RAW RAW NS... NS
kmZ in the Amarillo - Acuff - Man- POPD POPD D's DN

sker Soil Association. The densi- RIF RAM CATD DK

ty of larger ponds (size classes 2, g ‘fix RUN POPD

3, and 4) were highest in the Luf- 6W 6% FRMX FRMX

kin - Axtell - Tabor and Wilson - |RR |RR DK IRR

Crockett - Burleson Soil Associa-
tions and lowest in the Amarillo -
Acuff - Mansker Soil Association.

Differences between pond den-
sities in the land use regions de-
fined by the General Soil Map of
Texas (Godfrey et al. 1973) in the
overflight area were smaller than
differences between pond den-
sities in the soil associations. Table
11 gives the densities of ponds in

*Underlined variables were significant at o1=0.O5.
“NS means that the regression was not significant at oz=0.05.

TABLE 9. VARIABLE SELECTION DURING STEPWISE REGRESSION OF FRAMEWISE
POND DENSITIES FOR PONDS LESS THAN 1.25 HA

STEP 1 2 3 4 5 5 6

Intercept Intercept Intercept Intercept Intercept Intercept Intercept
CATD* CATD CATD CATD CATD RAIN RAIN

each land use region and includes MODEL SLOP SLOP SLOP SLOP CATD CATD
data from 10 of the 15 Texas land EFRMX ‘RR IRR SLOP SLOP
use categories. Note that the stan- FRMX POPD POPD IRR
dard deviations of pond densities FRMX FRMX POPD
inland use regions are larger than PRMX
those of the soil associations, in-

PROB > F 0.0040 0.0004 0.0001 0.0001 0.0002 0.0002 0.0001

homogeneous and is probably
more suitable as a unit for
classification. fQWe also attempted
to use geologic substrate as a
classification variable, but the

‘complexity of the map gave data

which could not be interpreted.
Since many ponds are built for
agricultural purposes, we also

*Underlined variables were significant at oz = 0.05. Doubly underlined variables are signifi-
cant at oz = 0.001.

Probability of F values of parameters of best six-variable model

Variable Prob > F

RAIN 0.0161
CATD 0.0242
SLOP 0.0001

. . . IRR 0.0498
compared the densities of ponds 1n POPD 00097
the different agricultural regions FRMx 0,0177

TABLE 10. POND DENSITY (PDNDS/km211 STANDARD DEVIATION) BY POND SIZE AND SOIL TYPE

Soil

Association Frames <1.25 ha 1.25-2.5 ha 2.5-5.1 ha 5.1-11.25 ha All Sizes
Lake Charles-

Edna-Bernard 5 0.59610.201 0.05010.041 0.01410.018 0.00710.009 0.67110.265
Miller-

Norwood-Pledger 3 0.827 1 0.346 0.084 1 0.049 0.034 1 0.022 0.002 1 0.004 0.959 1 0.367
Katy-Hockley-

Clodine 2 0.86510.376 0.09210.014 0.01510.003 0.00010.000 0.97210.395
Wilson-Crockett-

Burleson 3 1.94110.416 0.13810.028 0.02710.020 0.00610.000 2.11210.374
Burleson

Heiden-Crockett 4 1.94410.831 0.15310.076 0.02810.011 0.00210.008 2.04510.934
Lufkin-Axtell-

Tabor 4 1.49910.348 0.15310.088 0.02810.017 0.00210.003 1.68710.443
Castell-

Pontotoc-Ligon 3 2.93210.176 0.07510.016 001210.007 000210.004 3028100197
Truce-

Owens-Waurika 4 1.965 1 0.460 0.057 1 0.025 0.003 1 0.004 0.004 1 0.004 2.039 1 0.473
Abilene-

Tillman-Vernon 9 0.31210.084 003710.037 000710.009 000210003 O.35910.103
Rowena-

Sagerton-Mereta 6 0.490 1 0.335 0.029 1 0.031 0.008 1 0.009 0 0.526 1 0.358
Tarrant-

Kavett-Rowena 4 1.59010.501 009610037 0.02110.018 0.01210.006 1.72410.497
Tarrant-

Brackett-Denton 5 0.843 1 0.345 0.044 1 0.029 0.003 1 0.004 0 0.893 1 0.345
Tarrant-

Kavett-Tobosa 5 0.47710.313 001510.016 0.001 10.003 0 0.49310.319
Tarrant-

Brackett-Speck 3 0.67710.291 003710.022 0.0001 0.000 0 0.70410.295
Amarillo-

Acuff-Mansker 2 0.12010.123 001310009 0.01310.018 0.01010.014 0.17210.059
Patricia-

Brownfield-

Tivoli 3 0.15310.101 0.03310.039 0.00210.004 0.00510.009 0.39510.293
TABLE 11. POND DENSITY (PONDS/km211 STANDARD DEVIATION) BY SIZE CLASS AND LAND USE REGION

Land Use Frames <1.25 ha 1.25-2.5 ha 2.5-5.1 ha 5.1-11.25 ha All Sizes
Region

Coastal

Prairie 9 0.99610.794 0.08810.049 002310019 O.O0410.005 108810.844
Claypan Area 4 1.54710.290 0.10210.051 0.02110.020 0.001 10.003 1.67510.298
Blackland

Prairie 10 1.15910.583 009010080 O.01110.013 0.00310.005 1.26610.648
Grand Prairie 6 1.46610.986 004810022 0.00410.004 0.00010.000 1.52711.005
North Central

Prairies 4 1.78510.294 0.12310.025 001810011 000310.004 1.19510.298
Central Basin 2 1.851 11.244 0.048 1 0.026 0.003 1 0.005 0 1.902 11.274
Edwards

Plateau 11 1.14510.992 003810036 0.00810.014 0.00510.005 1.19711.031
Rolling Plains 13 0.45810.349 004710.040 001010010 0.0031-0.007 0.52210.369
High Plains 8 039510.541 003010035 000510008 000310005 050710.566
Bottom Lands 3 0.681 10.232 006010047 0.02310.019 0.01010.011 0.78110.306

covered by the survey (Table 12).
While differences
agricultural regions are smaller in
general than those between soil

1O

between

associations, the range of densities
is considerable, varying between
0.207 :1: .120 ponds per hectare in
the Edwards Plateau large ranch

1.851 :1; 1.244

cattle-sheep-and-goats region to
in the Edwards
Plateau Central basin cattle-and-
goats region. It is somewhat sur-

_€

‘d

TABLE 12. POND DENSITY (PONDS/km211 STANDARD DEVlATlON) BY POND SIZE AND AGRICULTURAL REGION

Agricultural

Region Frames

<1.25 ha

1.25-2.5 ha

2.5-5.1 ha

5.1-11.25 ha All Sizes

Southern High

Plains

(farming, cotton,

grains, sorghum,

cattle) 8

Rolling Plains

and Prairies

(cotton, grains,

sorghum, wheat,

livestock) 12

Rolling Plains

and Prairies

(small grains

and livestock) 5

Edwards Plateau

(large ranches,

cattle, sheep,

goats) 4

Edwards Plateau

(small ranches,

cattle, sheep,

goats) 6

Edwards Plateau
Central Basin
(cattle, goats) 2

Grand Prairie
(livestock, small
grains, cotton) 6

Blackland Prairie
(cotton, livestock) 5

Blackland Prairie
(dairy products,
cattle, cotton) 4

Post Oak

(cotton,

livestock,

poultry) 9

Coastal Prairie
(cotton, rice,
cattle) 8

0.39510.541

0.465 1 0.307

1.473 1 0.538

0.19710.111

1.17011.020

1.851 11.244

1.466 1 0.946

1.03410.714

1.431 10378

1.42310.696

0.602 1 0.191

0.30010.035

0.046 1 0.036

0.10710.041

000510007

0.04910.032

0.04810.026

0.04810.022

0.06310.049

0.13310.109

0.09510.051

0.071 10.042

0.00510.008

0.009 1 0.009

0.015 1 0.010

0.004 1 0.004

0.011 1 0.018

0.003 1 0.005

0.00410.004

0.006 1 0.008

0.017 1 0.017

0.021 1 0.015

0.02310.021

0.00310.005 0.50710.566

0.00310.006 0.52610.310

0.00510.008 1.60510.566

0.00210.004 0.20710.120

0.00610.006 1.76711.059

0.00010.000 1.90211.274

0.00010.000 1.52711.005

0.00410.006 1.11010.748

0.00210.003 1.58810.500

000310.005 1.54610.746

0.00610.007 071010.252

prising that the lowest and highest
densities of ponds should be in
agricultural regions that seem so
similar, but the differences in pond
density are probably due to the
variability {of the geologic
substrate of the Edwards Plateau
rather than differences in

. agricultural practices. The stan-

dard deviations of pond densities
in the agricultural regions are
nearly as large, or are larger, than
the mean densities, indicating that
agricultural regions do not form

homogeneous groups with respect
to pond density.

Data on pond densities arranged
by county are given in Table 13.

CONCLUSIONS

lt is possible to inventory farm
ponds from NASA color infrared
photography and classify them by
size by using a stereo-microscope
on the light table. A frame from
high altitude photography covers
considerable area, which makes

determining the relationship of
pond densities to land use or soil
factors difficult since most
photographic frames will include
two or more classification
categories for geology, soil
association, or land use. There was
no significant change in pond den-
sity when the river was in the sam-
ple block (2.2 kmz), indicating
that if there were any change in
density near the river it occurred
in a very narrow band. Many types
of data on criteria which might be

11

TABLE 13. POND DENSITY (PONDS/km211 STANDARD DEVIATION) BY POND SIZE AND COUNTY

County Frames x 1.25 ha 1.25-2.5 ha 2.5-5.1 ha 5.1-11.25 ha All Sizes
Bastrop 3 1.67310.177 0.11510.054 0.02110.024 0.00210.003 1.81510.123
Borden 3 0.19710.143 0.05610.043 0.01310.013 0.00910.010 0.28710.147
Brazos 1 1 .70 0.062 0.022 0.000 1 .254
Brown 1 1.762 0.123 0.030 0.006 1.93
Burnet 4 1.3211.174 0.05610.020 0.00210.004 0.0010.00 1.38811.197
Coke 3 0.22610.087 0.01010.005 0.00210.004 0.00210.004 024010.100
Coleman 2 099610.370 008410.056 0.01410.011 001010.014 1.10710.433
Colorado 5 1.41210.882 011410060 002210.013 0.00310.006 1.55610.942
Concho 2 0.838 1 0.402 0.052 1 0.057 0.006 1.0004 0.003 1 0.005 0.900 1 0.465
Dawson 2 014410088 001510011 0.0010.00 000710010 017010062
Falls 1 0.694 0.050 0.011 0.006 0.761
Fayette 4 1.4310378 0.13310.109 001710017 000210003 1.58810.500
Fort Bend 1 0392 0.073 1 0.034 0.006 0.509
Hays 1 0.744 0.063 1 0.00 0.00 0.807
Howard 3 0.26110.118 0.01410.019 000210004 0.0010.00 027710.135
Lampasas 1 1.27 0.019 0.006 0.00 1.296
Llano 2 1.85111.244 0.04810.026 000310.005 0.001000 1.90211.274
Lynn 2 1.85111.244 004810026 0.00310.005 00010.00 052910251
Matagorda 3 0.681 10.232 0.06010.047 002310.019 0.01010.011 0.78110.306
McCulloch 2 1.32511.263 002210031 0.02210.031 0.00710.010 1.37711.336
Mills 1 2.242 0.045 0.006 0.00 2.312
Mitchell 3 037910051 0.04110.044 0.01310.013 0.0010.00 043610.048
Nolan 1 0.201 0.032 0.00 0.00 0.233
Palo Pinto 1 2.104 0.146 0.028 0.006 2.284
Runnels 3 065710.256 0.04110.038 0.00610.007 0.0010.00 070310.264
San Saba 3 2.26510.853 006310032 000710012 0.00710.004 2.34710.890
Sterling 1 0.227 0.00 0.00 0.00 0.227
Terrell 1 1.713 0.090 0.022 0.006 1.831
Tom Green 1 0.060 0.00 0.007 0.00 0.066
Travis 5 1.03410.714 006310049 000610.008 O.0041O.006 1.11010.748
Wharton 4 059510167 0.07910.050 0.02010.027 000410004 0.70710.260
Young 2 1.63610.334 0.11110.033 000910003 0.0010.00 1.76310.293

related to pond density are main-
tained on a county basis, and are
confounded by the lack of coin-
cidence between photographic
frame boundaries and county
boundaries. The county data are
suggestive, however, and can be
useful as a guide to further
research.

ln the area photographed, over
9O percent of the ponds were
smaller than 1.25 ha, and the
percentages by pond size were
nearly the same in the Brazos and
Colorado River watersheds. Mean
pond densities were similar for the
Brazos and Colorado River water-
sheds except for the smallest size
group (<1.25 ha). ln this size
group the density of pond counts
in the Brazos River watershed (1.7
ponds/per kmz) was nearly twice
that of the Colorado (0.97 ponds
per kmz). However, the area
covered by photography in the
Brazos River watershed was much
smaller, and did not include the
upper part of the basin, which
would be expected to have very
low pond densities. ln an earlier

12

unpublished study where all of the
ponds in Brazos County were
counted, the reported density was
0.927 ponds/kmz, which is similar
to the density observed in the Col-
orado River basin. The Colorado
and Brazos River watersheds dif-
fered in the distribution of pond
densities along their watersheds
from the Gulf of Mexico to
upstream reaches. Pond densities
were low near the Gulf for both
watersheds. Pond densities in-
creased for the Brazos River with
distance from the Gulf. For the
Colorado River the densities were
very erratic in the central part of
the watershed, with the highest
densities found in the study
located there, along with many
frames of very low density. This
variability coincides with the
southern Edwards Plateau region.
The upper Colorado watershed, on
the Texas High Plains, had very
low pond densities. The pond den-
sity differences are consistent with
climatic and soil conditions in the
watersheds. Total ponds 1.25 ha
and smaller are estimated to be

102,000 for the Colorado River
basin and 191,000 for the Brazos
River basin.

ln a 10 factor stepwise regres-
sion analysis, the factor of the best
six variable model which con-
tributed the most to pond density
was slope (a gross measure of
relief obtained by subtracting the
lowest elevation in the county
from the highest). Population den-
sity was the second most impor-
tant contributor, and rainfall third.

Estimates were made of the
amount of watershed required to
supply water to maintain the
observed pond surface area in the
photographic frames. Using the
simplifying assumption that the
watersheds were hydrologically in-

dependent, we calculate that in 28 ‘l

percent of the frames over half of
the available watershed would be
required to maintain the ponds.
Because the assumption of in-

dependence of watersheds fails in ‘J

some frames with a high pond
count densities, the calculated re-
quirement may exceed 100 per-
cent of the land surface area in

these frames.

lt is obvious that published
numbers for the number of ponds
in Texas are much too low. We
estimate 293,000 ponds 1.25 ha
and smaller in the Brazos and Col-
orado River basins alone, as com-
pared to SCS estimates of 299,000
ponds for the entire state. Inven-
tory studies must be done in the
other major watersheds before
satisfactory statewide totals can be
calculated, but some idea of the
order of magnitude of the number
can be estimated from our data.

lf the area of the Rio Grande
River basin is excluded, since it
contains very few ponds, and the
average pond density from this
study (1.335 ponds/kmz) is ap-
plied to the remainder of the area
of the state, it is estimated that
there are approximately 782,000
ponds 1.25 ha or smaller, and ap-
proximately 842,000 ponds of all
sizes in the state.

LITERATURE CITED

BBR. 1976. Atlas of Texas,
Agricultural Regions Map. (p.
117) Univ. of Texas at Austin,
Bureau of Business Res.

Godfrey, C. L., G. S. McKee, and
H. Oaks. 1973. General Soils
Map of Texas. Texas Agri-
cultural Experiment Station,
Texas ASM University, and Soil
Conservation Service USDA.

Pass, F., Editor. 1980. Texas
Almanac and lndustrial Guide.
A. H. Belo Corp., Dallas, TX.

Renfro, H. B., and D. E. Feray.
1973. Geological Highway Map
of Texas. American Association
of Petroleum Geologists, Tulsa,
OK.

SCS. 1970. Conservation Needs
inventory, Texas, 1967. Texas
Soil and Water Needs Commit-
tee. U.S. Conservation Service,
Temple, 

SCS. 1978. National Erosion ln-
ventory Estimates 1977. U.S.

Conservation Service. Washing-
ton, D.C.

TCLR. 1976. Texas County
Statistics, Texas Crop and
Livestock Reporting Service.

U.S. Dept. Agr. Stat. Rept. Svc.,
Agr. Statistician, Austin, TX.
TSPE. 1974. The Effects of Ponds
and Small Reservoirs on the
Water Resources of Texas.
Texas Society of Professional
Engineers, Austin, TX.

USBC. 1971. U.S. Bureau of the
Census, 1970 Census, Number

Printing Office, Washington,
D.C.

USGS. 1966. Hydrologic Investiga-
tions, Atlas HA-12, “Annual
Runoff in the Conterminous
U.S. U.S. Geological Survey,
Washington, D.C.

Wetzel, R. G. 1983. Limnology,
2nd Ed. W. B. Saunders Co.

of Inhabitants. U.S. Government Philadelphia, PA.

APPENDIX

Example calculation of watershed area required to support ponds in
the study area.

Data from observation 66 are used in the following example. The ap-
proximate size of watershed required for each acre foot of pond water
varies as a function of distance from the coast (see page 11, TSPE 1974).
An approximation of the relationship was developed using a polynomial
regression which gave

A
(1) V = 9.62 + (1.232 ><10-3)m + (1.148 ><10-“)m2
where
A = acres of watershed to support pond
M = Miles from coast
V =volume of pond in acre-feet

Observation 66 was 505 miles inland. Thus

A acres
(2) V- = 9.62 +(1.232 >< 10-3)(505) + (1.248 X 10-“)(505)2 = 39.5

converting to ha/m3, and calling the result R we have
(3) R = 0.01296 lli
m

The volumes of ponds in each size class were estimated using the follow-
ing areas and depths

According to Wetzel (1983), the mean depth of lakes tends to be about
one-half of the maximum depth. Thus, we can estimate volume of a
pond in cubic meters as

(4) V = [hectares area] >< [.5 >< max. depth m] X [10,000 m2]
For size class 1 ponds, this volume (V1) amounts to

(5) V1= [0.600] >< [0.5 >< 1.5)>< (10,000] = 4,500 m3 per pond

Observation 66 has a density (D) of size class 1 ponds of 0.01713
ponds/ha. The total volume density V1(tot) of size class 1 ponds in the
frame is

(6) V](tot) = V, >< D = 4,500 >< 0.01713 = 77.005 m3
Finally we calculate the proportion of the watershed area (PW) required
to support this volume of water,

(7) PW = V1(tot) >< R
= 77.085 >< 0.01296 = 0.9990 ha required/ha available

Since V1(tot) is based on the number of ponds per hectare, we see that
potentially, 99.9 percent of the watershed in the hectare may be devoted
to supporting size class 1 ponds.

13

acre/foot