The gravimetric method of soil moisture determination Part III An examination of factors influencing soil moisture variability

Journal of Hydrology 11 (1970) 288-300; © North-Holland Publishing Co., Amsterdam Not to be reproduced by photoprint or microfilm without written permission from the publisher

THE GRAVIMETR1C METHOD

OF SOIL MOISTURE

DETERMINATION PART HI AN EXAMINATION

OF FACTORS INFLUENCING

SOIL MOISTURE

VARIABILITY

S. G. REYNOLDS*

South Pacific Regional College of Tropical Agriculture, Alafua, Western Samoa Abstract: The influence of various environmental factors on soil moisture variability in the 0-8 cm horizon is examined. The investigation of the influence of vegetation types revealed at least three tentative variability groups (in order of increasing variability) corresponding to areas of recent cultivation, areas of permanent or ley grass, and predominantly uncultivated moor or heathland areas. The dynamic nature of seasonal variability changes are illustrated and correlations established with the amounts of rainfall and insolation in the week preceding sample collection, and the degree of moisture saturation at the time of sampling. Required sample size was directly related to mean moisture content, or the degree of moisture saturation, and the amount of insolation. It is suggested therefore that a decrease in soil moisture content is associated with a decrease in the variance of sample means, and an increase in moisture content is associated with an increase in variance.

Introduction T h er e have been n u m e r o u s investigations o f soil m o i s t u r e variability, but ap ar t f r o m the w o r k o f Lull and R e i n h a r t l ) , and M o l t h a n 2 ) , noted below, and Ball an d Williams a), m o s t o f these have been c o n c e r n e d with generating sample size estimates and few a t t e m p t s have been m a d e to relate the degree o f variability to v a r i o u s e n v i r o n m e n t a l factors. I f relationships can be f o u n d then it should be possible to explain differences in the m a g n i t u d e o f variability between areas with slight or m a j o r e n v i r o n m e n t a l differences. A m o r e precise u n d e r s t a n d i n g o f the m a n y c o m p l e x factors influencing variability wo u l d enable m u c h m o r e accurate p r e d ic ti o n s to be m a d e a b o u t required sample size. F a c t o r s e x a m i n e d include vegetation cover, season, and rainfall and insolation received. * Formerly a postgraduate student in the Department of Geography, University of Bristol, England. 288

GRAVIMETRIC METHOD OF SOIL MOISTURE DETERMINATION. III

289

The main causes of soil moisture variability The degree o f soil moisture variability at any point will depend partly on the degree o f variability of the soil factors most directly influencing the moisture holding capacity o f the soil, and also on factors outside the soil system proper, such as vegetation, aspect, slope, weather conditions etc., which influence and control the a m o u n t o f water made available to the soil. According to McGuinness and Urban "~) "... variability in soil moisture is expected from point to point ... because o f variations in slope, aspect, soil type, vegetation cover, clay content of the soil and distribution o f rainfall ..." It is suggested that the main causes of soil moisture variability can be divided into two b r o a d groups. These are the static or slowly changing factors, and the dynamic or more rapidly changing factors. These groupings are not necessarily mutually exclusive, and some factors may change groups where certain environmental conditions prevail. As m a n y soil properties are dynamic their levels can vary with time of sampling, a factor often ignored in soil moisture studies. Gallagher and Herlihy s) reviewed much o f the relevant literature and also carried out various experimental investigations. They demonstrated that pH, K and P are influenced by season, climate and time elapsed after liming. Similar variations are to be expected in soil moisture content. Therefore any estimate o f the required sample size for an area must consider the dynamic causes of variability and take the seasonal changes in variability into account. The sample size mean values in Table I were c o m p u t e d from ten or more distinctly separate sampling occasions spread over a twelve m o n t h period. However, a n u m b e r o f papers have given sample size estimates based on a single set o f samples collected at one point in time. These may give a representative picture o f moisture variability t h r o u g h o u t the year, but probably reflect variability under only one particular set of environmental conditions. TABLE 1

A summary of required sample sizes for Windmill Hill and Exmoor

Location

Windmill Hill Exmoor

Number of plots

Sample sizes required to estimate True mean at 95 ~ probability level ~ 10 (St. error 5) ~ 5 (St. error 2.5) + 2 (St. error 1) range mean range mean range mean

200

1-7

4

2-19

7

3-116

30

120 170

2 14 2-18

6 7

3-55 5 73

14 21

4 346 14-459

79 125

290

s.G. REYNOLDS

Exceptions include papers by Lull and Reinhartl), Molthan, z) Fitzgerald

et al.6), and H a m m o n d and PopenoeT). Static factors include the degree of homogeneity of the soil body itself, incorporating texture, organic matter content, soil horizons, structure, and the numbers of channels formed by roots and by worms and other organisms. The amount and uniformity of soil constituents like organic matter and the number of channels, can to some extent be regarded as dynamic factors if for example the soil worm population or the vegetation type changes rapidly because of the influence of man. Especially important is the influence of soil variability on factors such as infiltration rate. Other static factors include the degree of variability of topographic factors such as slope, aspect, elevation. and microtopography (the last has been emphasised by Krumbach s)). Dynamic factors include the amount and variability of precipitation and insolation that an area receives, the length of time since the last heavy fall of precipitation (i.e. the degree of moisture saturation of the soil when the measurements were made: Lull and Reinhartl)), the depth to the water table (if one is present), and the amount of variability in vegetation type, degree of cover, density of cover, and extent of the litter layer. Lull and Reinhart 1) studied eight sites sampling them weekly for three months in the wet season and three months in the dry season. According to Molthan2), they concluded that "the variation in moisture measurements within a given site was primarily caused by the type and amounts of vegetative cover, depth to water table and the degree of moisture saturation of the soil ... the standard deviations for the wet season measurements were greater than those for the dry season at four of the five sites ..." Fitzgerald et al. 6) showed how the precision of estimate of soil moisture content increased as soil moisture content decreased from 34 per cent to less than 10 per cent. Similarly, H a m m o n d and Popenoe 7) and Verhoeven 9) found that the variance of sample means decreased with a decrease in moisture content. The type of plant is very important because it is the plant that intercepts precipitation and passes it to the soil, and whose roots take moisture from the soil. For example a grass cover has a very different influence from bracken (Pteridium aquilinum), the latter probably concentrating the supply of water to specific areas. "... water collects on certain parts of the plant shoot or crown and drips to the ground or runs down the stem and is deposited at the base. In this way some parts of the ground received many times the rainfall. As a result wetting of the soil in some vegetations is very variable ... the completeness of the vegetative cover and its composition determine to some extent how much rainfall wets the ground, how much solar energy is available for evaporation and how much air movement there is ... probably there

G R A V I M E T R I C M E T H O D OF SOIL MOISTURE D E T E R M I N A T I O N . III

291

are considerable differences between vegetative covers in their abilities to dry out soils during a period when demand exceeds rainfall ... (Reynolds1°)). Although the likely influence of vegetation on soil moisture variability depends on several factors it is probable that "the effect ... is least when the area is fully occupied with a cover of uniform composition and greatest when the soil is only partially covered - as with young row-crops or scattered bunch grasses or shrubs" (Anonn)). Thus Lull and Reinhart 1) reporting the variability of soil moisture from eight study sites, found that one of the major factors influencing the variation of soil moisture measurements was the type and amount of vegetation cover. The site with the greatest variance had only a 5-10% cover of short bunch grass and the remainder of the site was bare. Other dynamic factors include the size of the area to be sampled, and man induced variability in the form of heavy vehicle tracks criss-crossing an area or the addition of farmyard manure to the soil. This paper investigates a number of the dynamic factors mentioned above.

Site description Two areas in Somerset were selected for the investigations. The soils over the Windmill Hill site were fine textured, almost stoneless Brown Earths, developed on Keuper Marl, while the Exmoor soils were much stonier, coarse textured acid Brown Earths, developed on Devonian shales. Table 1 (in Part [[) contains various site details.

Sampling details SAMPLE DESIGN

Details of the linear site were outlined in Part H of this paper, however, in the investigation of vegetation type on moisture variability the linear site was discarded in favour of a subjective choice of suitable sub-areas in one small area of mixed vegetation types. Two 3.1 × 1.9 m plots were randomly located in each sub-area, each set of ten individuals being selected by random coordinates. SAMPLING METHODS

Details were outlined in Part 1 of this paper. Moisture data are reported on the gravimetric weight basis and relate to the surface 0-8 cm of soil.

Results and discussion The probable influence of vegetation on soil moisture was first noticed when comparing sample size estimates from Windmill Hill and Exmoor. The

292

s.G. REYNOLDS

plots located in the Exmoor area were much more heterogeneous than those at Windmill Hill, despite a certain amount of variability within the two areas. Sample size estimates for the two areas were contrasted in Part H: The differences are probably caused by the different vegetation covers of the two areas; Windmill Hill is covered by short ley grass while the Exmoor sites are mainly bracken (Pteridium aquilinum) covered, with a pronounced litter layer on the soil surface. In order to deduce the influence of various vegetation types on soil moisture variability, a preliminary investigation was carried out in the Exmoor area. Ten sites were chosen with vegetation types ranging from cultivated fields, ley and permanent grass to bracken and 'flush' areas. It is interesting to note, in Table 2 that the variability rankings l to 10 correspond to a dry to wet progression in vegetation and soil types, the significance of which is discussed later. One problem with the ten sites was that not all could be located on areas with the same slope angle. A Spearman's Rank correlation test suggests that the relationship between vegetation type and degree of variability might in part be a relationship between the slope angle of the plots and the degree of variability. Although there is this possibility, that the wetness of site could be related to slope angle, the mass of data supports the tentative vegetationvariability correlations mentioned above. it is therefore suggested that at least three vegetation variability groups can be recognised, but that many exceptions occur; for example, if an area of permanent grass is only patchy, it may fall into a higher variability group. Also if some particular soil factor, like degree of stoniness becomes dominant, then even a soil with a continuous grass cover may fall into a high variability group. Details of the groups are outlined in Table 3, with suggested average sample sizes required to estimate the mean with a standard error of 2.5 at the 95~o probability level. These tentative recommendations are intended as a relative guide only. A comparison of values for the same group of plots over time (Table 4 and Hills and Reynolds12)) illustrates the very rapid changes in soil moisture variability that can occur. Sample size estimates made for a group of plots at any one point in time are the result of a particular set of environmental factors influencing soil moisture distribution. Even within the group of plots at any one sampling occasion there is considerable diversity. The spatial variability changes which result partly from soil heterogeneity have been discussed in Parts [ and [[ and by Hills and Reynolds a2) but there are other factors involved, which are also important in terms of the variability changes over time. These are the dynamic factors mentioned above, in

293

GRAVIMETRIC METHOD OF SOIL MOISTURE DETERMINATION. Ill

TABLE 2

The influence of vegetation type on soil moisture variability

Site no.

1 2

3

4

Vegetation cover

Ley Bare soil and Swedes cultivated 3 m o n t h s earlier Bare soil and young Swedes cultivated 1 m o n t h earlier U n i f o r m tree cover (Oak Quercus robur and Hazel Corylus avellana) Grass and Bramble clumps

Variability ranking on basis of

Best estimate

Standard error

of standard deviation

2.5 sample size required

13½°

8.15

11

6

7°

3.89

5

2

8½°

2.69

4

1

25 °

5.27

6

3

19 ° 19½~)

7.09 6.40

9 8

5 4

12°

9.74

15

7

22 '~

10.46

17

8

20 °

11.32

20

10

23 °

10.79

19

9

Slope angle

sample size 1

( Roseieceae fruticosus) 5 6 7

Permanent G r a s s Permanent Grass S p h a g n u m moss, thick litter layer and bracken cover (Pteridium

aquilinum) 8

9

Bilberry ( Vaceinium myrtillus) Heath (Calluna vulgaris and Erica tetralix) s p h a g n u m moss, Grass (Molinia eaerulea etc). Flush vegetation s p h a g n u m moss, (Callu-

na vulgaris, Alolinia eaeruea, Eriophorum angustifolium, Juncus effhsus, Pteridium aquilinum. 10

Bracken, litter layer and scattered Molinia

eaerulea 1 Ranking on basis of sample size required to estimate the mean with a standard error of 2.5 at 9 5 ~ probability level 1 smallest sample size ...... 10 -- largest sample size.

294

S. G. REYNOLDS TABLE 3 Suggested vegetation variability groups 1

Variability group

Low Intermediate

High

Vegetation cover

Recently cultivated soil, cultivated areas with young crops Areas of permanent or ley grass. Areas of grass with a uniform tree cover (Assuming sample not taken directly adjacent to tree trunks) Areas of 'rough' and Moorland vegetation e.g. Heath, bracken, bilberry, coarse clump grass, sphagnum moss, etc.

Average sample sizes St. error 2.5, 95% probability level 2--5 individuals

5-t0 individuals

15-20

individuals

1 Based on samples from 3.1 × 1.9 m plots.

particular precipitation, insolation and vegetation. The degree o f influence will depend on the location o f the area and the season of the year at which the investigations are made. A l t h o u g h some areas of the world have marked wet and dry seasons and associated vegetation assemblages, the climate of south west England is m u c h more varied with no distinct wet and dry season changes. However, the area does experience wet and dry periods t h r o u g h o u t the year, and distinct seasonal vegetation changes. It is therefore possible that the temporal variability changes mentioned above result from differing amounts of precipitation, insolation, and to a lesser degree, vegetation changes, t h r o u g h o u t the year. Few studies exist which throw any clear light on the problem. However, some suggestions can be made: i. It is probable that the m o v e m e n t o f water into the soil is extremely variable because of the natural heterogeneity o f soils. ii. Variability will probably be at a minimum after a very dry period, when the effects o f soil heterogeneity on infiltration and the soil moisture holding capacity will be at a minimum. iii. Variability will be considerable after a fall of rain. iv. After a prolonged period of very heavy rain there are two possibilities. Variability may be very low because o f the considerable a m o u n t of water added to the soil which could mask soil differences. On the other hand, variability may be greatly increased because the effects of soil heterogeneity on infiltration and moisture holding capacity may be at a maximum. The work o f Lull and Reinhartl), Fitzgerald et al.6), and H a m m o n d and Popehoe7), suggest that the latter is the case.

8.6 13

tr

required* sample size

22

Required 2 sample size

1

ll.7

o-1

Sampling occasions

1

Plot No.

9

6.7

2

13

8.8

2

14

9.2

3

12

8.7

3

* Calculated from m e a n tr values because of large r o u n d i n g error. 1 Best estimate of s t a n d a r d deviation. Standard error 2.5, 95 ~ probability level.

M e a n value o f the 8 plots for each of the 10 sampling occasions

M e a n plot values for each of the 8 plots over I0 sampling occasions

TABLE 4

12

8.7

4

8

6.6

4

6 9.8 16

8.0 11

10

7.4

6

5

11

8.2

5

C h a n g e s in soil m o i s t u r e variability with time - m e a n values

12

8.5

7

12

8,7

7

17

10.3

8

16

9.9

8

9

6.9

9

14

9.2

10

296

s.G. REYNOLDS

v. Some factors which might be of great importance are: the length of time since the last heavy fall of precipitation, the amount of precipitation and insolation in the period directly preceding sampling and the degree of moisture saturation of the soil when the measurements are made. vi. If there is a marked seasonal change in the character of the vegetation of an area then a change in soil moisture variability may also take place. Data from eight plots on a south facing Exmoor slope sampled on ten separate sampling occasions, over a period of approximately eight months, were used to investigate these suggestions. A mean sample size figure was evaluated from the eight plots on each of the ten sampling occasions. It was decided to correlate the degree of variability with three climatic factors and to attempt some assessment of the influence of vegetation change on variability. The climatic factors were the amount of rainfall in the period directly preceding sampling, the amount of insolation in the same period, and the actual moisture content or "degree of moisture saturation" at the time of sampling, a factor which will obviously be closely related to the other two. The ground surface over the Exmoor site area is covered with a thick litter layer of decomposing bracken, but from late May to October the annual growth of new bracken interposes a very different medium between the elements and the soil, and the bracken often reaches a height of six feet. The growth in May and June is gradual and similarly, in the autumn the end of growth is a gradual process; but the bracken often stands long after it has become dry and brittle. Thus over this whole period it will be difficult to see any abrupt changes in moisture variability resulting from the bracken cover. The time when some distinct change may be observed is when strong winds flatten the tall bracken in October and November. [t was decided to correlate the variability figures (for sample size, with a standard error of 2.5 at the 95~o probability level) with the amount of rainfall and insolation in the period of seven days preceding sampling. (In fact, use of the best estimate of standard deviation figures instead of sample size, gives similar results). This period of time was chosen quite arbitrarily, and it is probable that various other periods of time, and also climatic factors expressed in a completely different manner, are better related to the changes of variability. This is one attempt to explain the changes in variability. It was thought that one or two days was probably too short a time for climatic factors to have much influence on soil moisture because of the time lag in effect. When the Vicksburg Waterways Experimental Station (Anon ll)) determined how closely solar radiation and other weather factors correlated with soil moisture losses, it was found that solar radiation data showed less variation and better correlation with other factors if summations of data for a period of days rather than daily values were used.

GRAVIMETRIC

METHOD

OF

SOIL

MOISTURE

DETERMINATION.

Ill

297

It must be emphasised that these are only rather crude, simple correlations, where a multiple correlation approach of the type used by Zingg13), would probably give much better results. The relationships between the amount of rainfall and insolation in the week preceding sampling, the moisture content and vegetation at the time of sampling, and the required sample size are shown in Fig. 1. Decreases in required sample size are associated with dry periods (i.e. high insolation and Soil moisture

%

6° I

/ / ",

50

~

40

30

/

x

t

/

l

~ ~

i

v Litter

~..

/.\

x

/

/ "

/

/

" Frr~sh Bro cken

~TIIIIIIIllLitter///////'/7"D~

Litter

Sorn~ size at st. error 2.5(+5) 950/o prob. I¢v¢1

/ /

\

I 2O

~I00

Insolation

(hours) 30 0o

in inches

i

4O

4

23"5

20

4-6

21-6 21-7 28-7

4-9

26-9 26:10 2%11

18~1

0

Date of sampling

Fig. 1. Relationship between soil moisture variability and rainfall and insolation in week preceding sampling, vegetation, and soil moisture content. (Rainfall data from Exford Rectory station SS(21)857384 Insolation - hours of sunshine - data from Hawkridge station SS(21)877327).

low rainfall values) and increases in required sample size with wet periods (i.e. low insolation and high rainfall values). Similarly the wet and dry periods result in respectively high and low mean moisture content values. Therefore, the results show agreement with the conclusions of Lull and Reinhartl), Fitzgerald et al.6), H a m m o n d and PopenoeV), and Verhoeven 9) in that the

298

s.G. REYNOLDS s a m p t e size 17 st. error 2.5 95% prob. I6 LeveL

x

14

P.

13

x ~~"C:LQ

12

x~'.pj

II

\

×( ve91

o

6

2~

~o 4b

sb ~

Hours Fig.

7b 8b

4o ~

,b

I~o

Sunshine (insoLation)

Relationship between magnitude of variability and insolation.

2a.

r - - 0 . 8 3 ; r 2 - 0 . 6 9 ; L e v e l o f s i g n i f i c a n c e : O.Ol.

Sample st. 95*

error

size t7 2,5

prob.

x

16

Level 15 14

12: I IC

=
o

.2's ~

.7;

,60 ,.~s ,-~o ,hs 2.6o 2.'2s 2~o R a i n t a t t in

Fig.

2b.

inches

Relationship between magnitude of variability and rainfall. r = 0 . 7 4 ; r 2 = 0 , 5 5 ; Level of significance: 0 . 0 1 .

variance of sample means (and therefore required sample size) decreases with a decrease in moisture content. The one variability value which fails to follow the general trend is that for 29th November; it is very low despite a very wet period in the week preceding sampling. The explanation for this probably lies with the seasonal changes of bracken cover. Although a full twelve month period was not studied it is possible to observe a general upward trend in the required sample size in the

GRAVIMETRIC

METHOD

OF SOIL MOISTURE

DETERMINATION.

Sample size 17 st. error 2.5 95°/o pro. 16 [eve[

III

299

x

I

J

f2,

o L ~

---

ab

- 4% . . . .

5-~---

5'5

66

Mean % Moisture Content

Fig. 2c. Relationship between magnitude of variability and mean moisture content. r = 0.81 ; r 2 ~ 0.66; Level of significance: 0.01. period of the fresh bracken cover. The sampling date 29th November followed directly upon the flattening of the bracken cover by strong late autumn winds, which considerably altered the surface vegetation factor. This probably resulted in a sharp decrease in the variable influence of vegetation on moisture penetration and accumulation, and gave lower variance values for soil moisture means and smaller required sample sizes. Although an attempt was made to analyse the data with statistical correlation procedures, the main aim of Figs 2a, b and c was to suggest general trends. Relatively low r 2 values are probably a result of the arbitrary selection of some of the values, the rounding errors associated with sample size calculation, the distance of the meteorological stations from the plots, and the multifactorial complexity of the problem which necessitates multiple correlation techniques. [n spite of the large amount of variability, trends are apparent. Required sample size (and thus variance) shows a direct relationship with mean moisture content and rainfall, and an inverse relationship with insolation. In summary, it is suggested that there is considerable evidence to suggest that required sample size is directly proportional to the degree of moisture saturation of the soil: a decrease in soil moisture content is associated with a decrease in the variance of sample means, and an increase in moisture content is associated with an increase in variance. Conclusion

The main result of this paper has been to illustrate the relationships

300

S . G . REYNOLDS

between soil moisture variability and a n u m b e r of e n v i r o n m e n t a l factors. Knowledge of these relationships should enable future studies to be p l a n n e d with a more precise idea of the degree of variability to be expected. W h e n collecting soil variability i n f o r m a t i o n involving d y n a m i c soil properties, it is i m p o r t a n t that seasonal a n d other changes in variability are taken into account. Where sample size estimates are based o n a single set of samples collected at one p o i n t in time, they may give a representative picture of moisture variability t h r o u g h o u t the year, b u t p r o b a b l y reflect only one particular set of e n v i r o n m e n t a l conditions. O f particular i m p o r t a n c e is the considerable evidence suggesting that required sample size is directly proportional to the degree of moisture s a t u r a t i o n of the soil.

Acknowledgement [ a m indebted to Mrs M. R e y n o l d s for assistance with c o m p u t i n g and to D r P. Radford, Principal, South Pacific Regional College of Tropical Agriculture, for helpful criticism.

References 1) H. W. Lull and K. G. Reinhart, Soil moisture measurement. Occ. Paper 140. S. For Exp. Sta. New Orleans. La. (1955) 2) 14. D. Molthan, Influence of soil variability on soil moisture and strength predictions. U.S.A.E. Waterways Exp. Sta. (Working Draft) (1966) 3) D. F. Ball and W. M. Williams, Variability of soil chemical properties in two uncultivated brown earths. J. Soil Sci. 19(2) (1968) 379 399 4) J. L. McGuinness and J. B. Urban, Soil moisture sampling plan for watersheds. U.S. Dept. Agric., Agric. Res. Serv. 41-87, 12p. (1964) 5) P. A. Gallagher and M. Herlihy, An evaluation of errors associated with soil testing. Irish J. Agric. Res. 2 (1963) 149 167 6) P. D. Fitzgerald, D. S. Richard and N. S. Mountier, Sampling errors associated with gravimetric soil moisture determinations. N. Z. J. Agr. Res. 6 (1963) 307-309 7) L. C. Hammond and H. Popenoe, Soil moisture measurement for timing irrigation. Soil Sci. Soc. Fla. Proc. 15 (1956) 154 164 8) A. W. Krumbach, Effects of microrelief on distribution of soil moisture and bulk density. J. Geophys. Res. 64 (10) (1959) 9) B. Verhoeven, Salt and moisture conditions in soils flooded with seawater (Staatsdrukkerij, 's-Gravenhage, 1953) 10) E. R. C. Reynolds, The hydrological cycle as affected by vegetation differences. J. Inst. Water Eng. 21 (1967) 11) Anon, Forecasting trafficability of soils. Developing and testing of some average relations for predicting soil moisture. U.S.A.E. Waterways Expt. Sta. Corp. Eng. Vicksburg. Miss. Tech. Mem. No. 3-331 Rept. 5 (1959) 12) R. C. Hills and S. G. Reynolds, Illustrations of soil moisture variability in selected areas and plots of different sizes. J. Hydrol. 8 (1969) 27-47 13) A. W. Zingg, Soil moisture studies. Estimating the moisture content of the 0 to 6inch soil horizon from climatic data. Soil Sci. Soc. Amer. Proc. 8 (1943) 10%111