SSSAJ Grow Your Career with SSSA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (15)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Martínez-Fernández, J.
Right arrow Articles by Ceballos, A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Martínez-Fernández, J.
Right arrow Articles by Ceballos, A.
Agricola
Right arrow Articles by Martínez-Fernández, J.
Right arrow Articles by Ceballos, A.
Related Collections
Right arrow Water Content
Right arrow Time Domain Reflectometry, TDR
Right arrow Soil Hydrology
Published in Soil Sci. Soc. Am. J. 67:1647-1656 (2003).
© 2003 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA

DIVISION S-1—SOIL PHYSICS

Temporal Stability of Soil Moisture in a Large-Field Experiment in Spain

José Martínez-Fernández* and Antonio Ceballos

Dep. of Geography, Univ. of Salamanca, Cervantes, 3. 37002-Salamanca Spain

* Corresponding author (jmf{at}usal.es).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Here we analyze the temporal stability of the soil moisture content in an area of 1285 km2 located in the central sector of the Duero basin (Spain). The data from a network consisting of 23 soil moisture-measuring stations (REMEDHUS) over a period of 36 mo (from June 1999 to May 2002) were used. At each station, soil moisture was measured fortnightly, using time domain reflectometry (TDR), at different depths to the bottom of the profile. From the hydrological point of view, the stations are independent from one another and were distributed in space using physiographic and pedological criteria. The results clearly show which stations are representative of wet and dry conditions and the persistence of the representative stations under even the most extreme situations. The temporal evolution of the stability patterns points to a considerable degree of persistence during the periods analyzed. The stations representative of dry conditions are seen to be much more stable at all the depths analyzed (5, 25, 50, and 100 cm) than those identified within the wet sector. Overall, the soil moisture content showed considerable temporal stability throughout of the investigation. However, a clear correlation was observed between mean soil moisture and variance (r = 0.85) across the range of measurements used. Accordingly, temporal stability is always higher when the soils are dry than when the water content is high. It was also observed that the periods with the lowest temporal stability coincided with situations involving the transition from dry to wet. The critical periods in terms of temporal stability are those in which the recharge of water in the profile occurs.

Abbreviations: PET, potential evapotranspiration • TDR, time domain reflectometry


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
SOIL MOISTURE is a key status variable for understanding many hydrological processes that, in turn, are involved in a large variety of natural processes (geomorphological, climatic, ecological, etc.) that act at different spatio-temporal scales (Entin et al., 2000). Soil moisture is involved in the partitioning of net radiation into sensible and latent heat; it determines the amount of water available for evapotranspiration, and it controls subsurface flow and the migration of chemicals toward aquifers. Antecedent soil moisture is a key factor in hydrological and erosional modeling. Knowledge of soil moisture processes and their spatial distribution provides essential information for climatic and hydrologic models.

From an applied point of view, knowledge of soil moisture dynamics is crucial for the prediction of flood episodes, the estimation of aquifer recharge, and the risk of the migration of pollution. From an agricultural point of view, it is also an extremely important variable. Indeed, in non-irrigated territory the soil water coming from the infiltration of rainfall is the sole source of soil moisture for crops. In irrigated areas, good management practices in terms of technical efficiency, the optimization of water resources, and agricultural production will all largely depend on a good knowledge of the soil moisture status of the soils at any given moment.

Much has been published concerning the spatial variability of soil properties but less is known about their temporal variability. Despite this, in recent years there has been increasing interest in analyzing the temporal dynamics of soil moisture contents, especially since the publication of the well-known article by Vachaud et al. (1985). In that work, the authors introduced the concept of time stability as the time-invariant association between spatial location and the classical statistical parametric values of a given soil property. The authors showed that certain sampling locations expressed the mean of the whole area studied, whereas others were characteristic of extreme values. Thus, the temporal stability of soil moisture is a reflection of the temporal persistence of the spatial structure (Kachanoski and de Jong, 1988).

Over the past decade several researchers have analyzed the temporal dynamics of different soil properties (Goovaerts and Chiang, 1993), among which hydrological properties are outstanding owing to the attention given to them. Jaynes and Hunsaker (1989) studied the temporal stability of infiltration and its application to irrigation. Van Pelt and Wierenga (2001) analyzed the temporal stability of the soil matric potential with a view to optimizing sampling strategy. The spatio-temporal variability of the soil moisture content has been studied by several authors (Kachanoski and de Jong, 1988; Van Wesenbeeck and Kachanoski, 1988; Jaynes and Hunsaker, 1989; Comegna and Basile, 1994; Famiglietti et al., 1998; Grayson and Western, 1998; Gómez-Plaza et al., 2000; Hupet and Vanclooster, 2002).

In all the above works, the sampling schemes and the scales of analysis were very different. A common procedure has been to use transects (Kachanoski and de Jong, 1988; Van Wesenbeeck and Kachanoski, 1988; Famiglietti et al., 1998; Jacques et al., 2001; Gómez-Plaza et al., 2000). Occasionally, systematic or grid samplings have been performed in crop fields or basins (Goovaerts and Chiang, 1993; Grayson and Western, 1998; Van Pelt and Wierenga, 2001; Hupet and Vanclooster, 2002). Other authors have used combined schemes of transects and sampling grids (Jaynes and Hunsaker, 1989; Comegna and Basile, 1994). The size of the study areas addressed also varies considerably, ranging from a few meters (Jacques et al., 2001) to a few hectometers (Kachanoski and de Jong, 1988; Jaynes and Hunsaker, 1989; Famiglietti et al., 1998; Gómez-Plaza et al., 2000) in the case of transects, and <1 ha (Vachaud et al., 1985; Goovaerts and Chiang, 1993; Comegna and Basile, 1994; Van Pelt and Wierenga, 2001; Hupet and Vanclooster, 2002) or a few hectares (Famiglietti et al., 1998; Grayson and Western, 1998) in the case of sampling grids. However, there are very few works addressing surfaces larger than 1 km2 (Grayson and Western, 1998).

Regarding the depth of the soil analyzed in studies on temporal stability, the usual case is to limit analysis to the surface-most layer of the soil: 0 to 20 cm, (Van Wesenbeeck and Kachanoski, 1988; Goovaerts and Chiang, 1993; Famiglietti et al., 1998; Gómez-Plaza et al., 2000) and there are very few works that refer to the whole soil profile (Kachanoski and de Jong, 1988; Comegna and Basile, 1994; Jacques et al., 2001; Hupet and Vanclooster, 2002). There are even fewer studies that have explored temporal stability as a function of depth (Hupet and Vanclooster, 2002). This strongly hinders application of the results with respect to the availability of water throughout the soil profile.

Regarding soil moisture, the time periods of the investigations tend to be short since measurements are highly technical and time-consuming. In works addressing temporal stability, discrete time periods of between 3 and 6 mo tend to be used, based on specific field experiments with intensive sampling (Jacques et al., 2001; Hupet and Vanclooster, 2002) or linked to the growing season (Van Wesenbeeck and Kachanoski, 1988; Comegna and Basile, 1994). Very few works, such as those of Grayson and Western (1998) or Gómez-Plaza et al. (2000), analyze time periods longer than 1 yr.

In the light of the foregoing, it is clear that it is also necessary to know the temporal dynamics of soil moisture over larger territories and longer periods. One of the goals set up by Vachaud et al. (1985) on proposing the analysis of temporal stability was to offer a method that would reduce the number of measuring sites necessary to analyze the behavior of a given soil. The measurement of soil moisture contents requires sophisticated techniques and is therefore time-consuming and costly. The alternatives offered to direct measurement in the field are estimations by remote sensing or the use of simulation models (Albertson and Kiely, 2001). Both methods require in situ measurements for the calibration and validation steps, together with information about the temporal dynamics of the soil moisture variable. The use of remote sensing, from measurements of either passive (Famiglietti et al., 1999, Mohanty and Skaggs, 2001) or active (Wagner et al., 1999) microwaves, is a very promising alternative for obtaining the water content of a soil in a given area. However, the temporal variability of soil moisture may introduce systematic uncertainty into remotely sensed soil moisture data (Mohanty and Skaggs, 2001). A broadening of the range of spatial-temporal scales in the analysis of temporal stability of soil moisture contents may generate additional information that would be useful for reducing that uncertainty. As pointed out by Kachanoski and de Jong (1988), hydrological processes operate at different scales and hence the temporal stability of spatial patterns will also be a function of scale.

Here we analyze the temporal stability of soil moisture using data from a network of stations installed over a surface of 1285 km2 during a 3-yr measurement period. The specific aims of this work were (i) to gain insight into the temporal pattern of each of the stations and of the whole dataset; (ii) to analyze how that pattern varies as a function of the measurement depth; (iii) to identify the stations representative of both the mean values and of the different soil moisture statuses, and (iv) to check the relationship between soil moisture status and temporal stability.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Study Area and Experimental Layout
The experimental layout covers an area of 1285 km2 and is located in the central sector of the Duero basin (Spain), called La Guareña (Fig. 1) . Detailed information about the area can be found in Ceballos et al. (2002). The climate is semi-arid continentalized Mediterranean, with a mean annual rainfall of 385 mm, a mean temperature of 12°C, and an annual potential evapotranspiration (PET) of 908 mm. Maximum monthly precipitation is 47.2 mm in May and the monthly minimum is 10.8 mm in August. The maximum monthly PET is 155.8 mm in July and the monthly minimum is 14.4 mm in December. Rainfall is very uniform throughout the study area. The five weather stations in the area show a high correlation (r = 0.9) with the amount of precipitation. The mean soil physical characteristics of the profiles studied are shown in Table 1. The most abundant soils are Luvisols and Cambisols (FAO classification), with a predominantly sandy texture (mean sand content, 71%), above all at the surface horizons and, occasionally, there are clayey horizons at the bottom of the profiles. The organic matter content is very low (mean, 0.9%). Throughout almost all the territory the soil is used for agricultural purposes, rain-fed crops being the norm (80% cereals).



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 1. Location map of the study area and Soil Moisture Measuring Station Network (REMEDHUS).

 

View this table:
[in this window]
[in a new window]
  		Table 1. Average soil physical characteristics from each profile.

 
In the spring of 1999 a network of 23 soil moisture-measuring stations (REMEDHUS) was set up in the area. The distribution of the stations is irregular and was based on the distribution of the main physiographic and pedological units of the area. With the exception of three stations located at the bottoms of valleys used for grazing, all the rest were located in areas used for non-irrigated crops (cereals and vineyards). The slope angle was low—<10%—and the altitude of the stations ranged between 700 and 900 m above sea level. Each station comprised a soil profile equipped with two-wire TDR probes measuring 265 mm in length, installed horizontally at depths of 5, 25, 50, and 100 cm. At two stations, L7 and M9, reduced soil depth made it necessary to install the deepest probe at 40 and 60 cm, respectively. For probe installation, a pit was dug in the soil of sufficient width to be able to insert the probes. The probes were inserted into the soil through the side of the pit that was unaltered and were placed horizontally in the direction of maximum slope of the terrain. Once the probes had been inserted, the pit was carefully refilled, avoiding perturbations as far as possible. The stations were set up at the beginning of March 1999 and measurements were not begun until 4 mo later with a view to allowing the soil to settle. Measurements were taken fortnightly at each of the stations. All stations were considered to be hydrologically independent even though they were all included within the same climatic context.

For TDR soil moisture measurements, a Tektronix 1502C (Tektronix, Beaverton, OR) was used as a waveform generator. The waveforms were analyzed visually in the field, following the method described by Cassel et al. (1994), and soil moisture was obtained using the formula proposed by Topp et al. (1980). Before this, the method was calibrated at the laboratory (Martínez-Fernández and Ceballos, 2001) by measurements in monoliths of the main soil types present in the study area. The gravimetric and TDR methods were used to test the applicability of the formula, as suggested by Zegelin et al. (1992).

For this work, we used the REMEDHUS dataset from June 1999 to May 2002. Accordingly, we had 23 stations, with four probes measured on 74 occasions, providing a total of 6800 observations. To analyze temporal stability, we employed the complete dataset of 36 mo to determine the mean evolution of the study area. We also subdivided the period into intervals of 12 mo—that is, three complete annual cycles—with a view to detecting any particular trend among discrete temporal periods. The three individualized periods were June 1999 to May 2000, June 2000 to May 2001, and June 2001 to May 2002, with precipitations of 438.6, 425.1 and 269.4 mm, respectively. The soil moisture data measured at depths of 5, 25, 50, and 100 cm were analyzed separately to include the depth variable within the analysis of temporal stability.

Data Analysis
Temporal stability was defined by Vachaud et al. (1985) as the time-invariant association between spatial location and classical statistical parametric values of a given soil property. To analyze the temporal stability of the spatial patterns of soil moisture in the present work, four statistical techniques were employed:

  1. Analysis of the cumulative probability function of data corresponding to different observations. With this, the aim is to see whether a given location maintains its rank in the cumulative probability function at different sampling times.
  2. The parametric test of relative differencing, proposed by Vachaud et al. (1985). This allows one to plot the data with a view to highlighting the differences, in terms of constancy in temporal stability, between sampling locations. The relative difference, {delta}ij, is calculated from:

    [1]
    where

    [2]
    and

    [3]
    Sij being the soil moisture content at location i on Day j, and N the sampling locations. Thus, the mean relative difference for each location is defined as:

    [4]
    where m is the number of sampling days.

  3. The standard deviation of the mean relative difference at each location, {sigma}({delta}i), was calculated as an estimator of temporal stability:

    [5]
    From this point of view, time-stable locations are defined as those with a low value of {sigma}({delta}i).

  4. The Spearman non-parametric test was used to analyze the temporal persistence of the ranks of several measuring locations. With Rij as the rank of the variable Sij at location i on Day j, the rank of the same variable at the same location on Day j' is defined as:

    [6]
    where n is the number of observations. A value of rs = 1 corresponds to the identity of the rank for a given location; that is, perfect temporal stability between Days j and j'. The closer rs approaches 1, the more stable the process analyzed will be. The SPSS software was used for statistical treatment of the data.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Temporal Stability of Soil Water Content
For a better understanding of the results (Fig. 2 and 3) , the data on relative differences in each case were ordered from smaller to greater, indicating the standard deviation, {sigma}({delta}i), by error bars above and below the points. According to Vachaud et al. (1985), with this type of methodology it is possible to identify the points that systematically over or underestimate the mean soil moisture value.



View larger version (30K):
[in this window]
[in a new window]
  		Fig. 2. Plots of relative differences for an average soil moisture profile, considering the whole time period of measurements and discrete annual time periods. Vertical bars correspond to associated time standard deviation.

 


View larger version (31K):
[in this window]
[in a new window]
  		Fig. 3. Plots of relative differences for several measurement depths at each station. Vertical bars correspond to associated time standard deviation.

 
Figure 2a shows the results for the mean data on soil moisture from the 23 stations and the complete period. For better visualization and uniformity in the plot, Station M13 is not shown because its mean relative difference value is >100%. This station is located at the bottom of a valley that sometimes becomes flooded in winter, which is why it systematically shows soil moisture values above those of the other stations, together with marked variability. Stress should be placed on the lack of symmetry with respect to the null value of relative difference among the stations. Fourteen of the stations are seen to be systematically below the mean value (i.e., the dry sampling locations) while nine are above it. The latter are also characterized by a much lower temporal stability in terms of standard deviation. At the wettest stations, this ranges between 7.1 and 28.8%, while at the dry stations, with the exception of L7 and K10, it is lower than 10%. Following the proposal of Grayson and Western (1998), the average soil moisture monitoring sites (ASMM) were identified; that is, the points that are located close to the zero of mean relative difference and that also have the minimum value of standard deviation. For the whole study area, Station J14 would be representative of the mean soil moisture value. Likewise, K4 and H7 would be representative of dry conditions, and Q8 and N9 of wet conditions, although the latter two with some degree of uncertainty because they are much less stable. The range of variation in the mean relative differences (between ±50%) is high with respect to results published elsewhere (Vachaud et al., 1985; Comegna and Basile, 1994; Grayson and Western, 1998; Van Pelt and Wierenga, 2001), although it should be recalled that the study area is more extensive and hence more diverse from the point of view of soil types and landscape positions. However, the values are similar to or lower than those reported by Gómez-Plaza et al. (2000) for Mediterranean bioclimatic conditions. The lowest limit (-50%) is always maintained (Fig. 2a–2d), indicating the presence of a physical threshold with respect to water storage and pointing to the high temporal stability of the stations defining this sector. The upper limit varies, sometimes surpassing 100% (Station M13), due to the strong seasonality of soils that in winter may become saturated, while they lose most of their water during prolonged periods of drought (Ceballos et al., 2002). This phenomenon is common in regions with a Mediterranean climate.

Figure 2b through 2d show the results for each of the discrete annual time periods analyzed. In general, the stations maintain their position, regardless of the periods, even though from the pluviometric point of view these periods are different. The groupings of stations characterizing the dry and wet locations are always identical. Only Station L7 fluctuates between positive or negative values, depending on the year. The soil at this location is only 40 cm deep and hence only has a limited water storage capacity. The stations showing the most marked temporal instability maintained this characteristic, regardless of the period considered.

When the whole profile is analyzed, the spatial-temporal scheme seems to be well defined, whereas on studying this at different depths the variability between stations is greater. Figure 3 shows the results of the whole time period for 5 (3a), 25 (3b), 50 (3c), and 100 (3d) cm. In all cases, the above-mentioned trends can be seen. The representative stations [a lower {sigma}({delta}i)] of the dry sector of the area are much more time-stable at all depths and their temporal stability ranges around 6 to 9% at the four measurement depths considered. In this set of stations, the predominance of the sandy fraction (Table 1) is reflected in the inability of the soils to retain water, which explains why the soil moisture values are never high and why temporal stability is relatively high. However, at the different depths considered it is difficult to identify stations representative of the wet sector because in all cases stability is low and {sigma}({delta}i) is always >10%. Not even at the bottom of the profile, which is where, a priori, one would expect greater stability due to the reduced dependence on the climatic, biological, and hydrological factors that determine the soil moisture dynamics, is it possible to find a single station approaching the 5% value that in some cases (Van Pelt and Wierenga, 2001) has been proposed for the selection of representative locations. Regarding the mean value at each depth, the representative stations would be J14 (5 and 25 cm), as occurred in the analysis at profile level, and I3 (50 and 100 cm). This means that joint use of both these stations would probably be more appropriate when searching for a valid estimator of the mean soil moisture of the whole area than if only J14 were used, as seems to be indicated by the global analysis (Fig. 2a).

The cumulative probability function for the dates on which the lowest (5 Sept. 2000, 0.123 cm3 cm-3 ± 0.047 cm3 cm-3) and the highest (6 Mar. 2001, 0.258 cm3 cm-3 ± 0.098 cm3 cm-3) mean soil moisture values were recorded is shown in Fig. 4 . Strikingly, the positioning of the points hardly undergoes any variation, despite the fact that they are the two most extreme situations. This indicates that both the wettest stations and the driest ones will always follow such a trend. There does exist the possibility of some stations changing position appreciably from the dry to the wet condition, but this is due to highly specific circumstances. For example, K10 and E10 are subject to drainage constraints after very rainy periods due to the presence of clay layers at the profile bottom. In general, however, the spatial scheme is maintained over time, despite there being dissimilar soil moisture contents.



View larger version (18K):
[in this window]
[in a new window]
  		Fig. 4. Cumulative probability function of mean soil moisture content at each station for the two extreme soil moisture contents during the study period.

 
Temporal Patterns of Soil Moisture Content
Owing to the huge amount of space that it would occupy, the matrix of rank correlation coefficients corresponding to all 74 observation dates made in the study period has not been included here. In that matrix, the correlation coefficient ranges between 0.57 and 1, significance in all cases being better than 0.01. It should be stressed that the lowest values are related to periods of transition from a dry soil to a wet one. In particular, the coefficient of 0.57 refers to 13 Nov. 2000, a date that in most cases has a coefficient of <0.8 with the other dates. That day lies in the middle of the longest soil water recharge period of the whole study period between September 2000 and January 2001.

Table 2 shows the matrix of rank correlation coefficients corresponding to the measurements made in January and August, representative of the periods of most extreme soil moisture contents. In all cases, the coefficient is very high, pointing to high temporal stability between the most extreme situations.


View this table:
[in this window]
[in a new window]
  		Table 2. Matrix of rank correlation coefficients corresponding to soil moisture data for January and August.

 
Figure 5 shows the value of rs of the Spearman test and its evolution against the mean soil moisture content for the whole area. This is the rs of a given day with respect to the value of the previous day. The level of temporal stability is very high at all times, as expected. The correlation coefficient ranges between 0.89 and 1, and the fluctuations are not very marked. The interest in using the test in this way lies in the analysis of situations in which rs is unexpectedly low. Again, it is interesting that the declines in rs always occur at times of transition; that is, they remain more or less stable in the periods with either high or low soil moisture, and decrease when a period occurs in which the water of the profile is recharged. This would reflect the different trend followed by each soil profile when rainy periods occur and an advance in the soil moisture front occurs. Under these circumstances, the different response of each soil becomes more patent as a function of its specific physical characteristics and the result is therefore greater variability. Certain profiles (E10, J12, L7, N9, O7, Q8, and M13) have underlying clayey horizons, whose hydrological behavior is markedly different from those of the sandy surface horizons. The response of these soils during the period of soil moisture redistribution is very different from that of soils exhibiting greater textural uniformity in the profile. This type of pattern is not seen either when the soils are dry or when they are very wet, when the response is more homogeneous. During the drying phases, the coefficients are always high. This could be more due to the greater importance of atmospheric factors participating in evaporation (which are very homogeneous throughout the territory) than to the physical characteristics of the soils. It would appear that the greater instability seen in the periods of water recharge in the soil is due to textural differences among the soils and such differences are less evident with other situations. The importance of transition periods in soil moisture dynamics has been highlighted in issues as sensitive as modeling (Albertson and Kiely, 2001).



View larger version (19K):
[in this window]
[in a new window]
  		Fig. 5. Evolution of the Spearman test rs against mean soil moisture content in the whole area during the study period.

 
There has been considerable debate as to whether temporal stability is greater in dry periods or in wet ones. Certain authors feel that there is greater stability during dry periods (Robinson and Dean, 1993; Famiglietti et al., 1998), while others consider that instability is greater when the soil contains less water (Van Wesenbeeck and Kachanoski, 1988; Gómez-Plaza et al., 2000; Qiu et al., 2001; Hupet and Vanclooster, 2002). In the present work we found a very clear and direct relationship between temporal instability and soil moisture contents. Figure 6 shows the mean soil moisture content and the variance associated with each observation date. A noteworthy parallelism can be seen between the mean and variance, the latter being between four- and five-fold lower in the summer period, when the soil moisture content reaches a minimum, than in the wettest periods. The correlation between the mean and variance (r = 0.85, sig. <0.001) is very high for the whole range of measurements (Fig. 7) , unlike what has been reported elsewhere (Famiglietti et al., 1998), in which it becomes more scattered as soil moisture increases.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 6. Evolution of variance against mean soil moisture content during the whole study period and discrete annual periods in the whole area.

 


View larger version (13K):
[in this window]
[in a new window]
  		Fig. 7. Correlation between mean soil moisture content and variance using the complete dataset.

 
Table 3 shows the results of the correlation analysis between the mean of the relative differences and the standard deviation for each observation date and each depth at which soil moisture was measured; for the whole profile; for the whole time period of measurements, and for each discrete annual time period analyzed. In all cases, a positive correlation is seen between the relative difference and standard deviation. This means that temporal stability increases as relative differences decrease. The sampling locations that are representative of the dry conditions of the study area are always more stable, and the wetter locations are less stable. No type of specific pattern of stability is seen with respect to depth. A reduction in the correlation is observed for the base of the profile. This is probably due to the fact that at that depth variations in soil moisture are relatively gradual; no strong fluctuations occur and the soil exhibits a certain inertia in its hydrological dynamics. Even though the three annual periods were different as regards the amount of rainfall, this factor was not seen to modify the patterns of temporal stability. In all cases the correlation is of the same sign, with the same level of significance. Only at a depth of 5 cm can a reduction be seen in the relationship between relative difference and standard deviation in the third year; this does not occur for the rest of the soil profile.


View this table:
[in this window]
[in a new window]
  		Table 3. Correlation analysis between the mean of the relative differences and standard deviation for each depth and different discrete periods analyzed.

 
From the results of the temporal patterns analysis it seems clear that temporal stability is much greater when the soil is dry than when the water content is high. Additionally, it was observed that in transition periods—more specifically, in water recharge periods—instability is enhanced. This kind of temporal pattern has consequences for sampling strategies. It is likely that in periods with high soil moisture contents it would be necessary to increase sampling density and that periods in which the advance of the soil moisture front is occurring would be the least appropriate for such activity to be performed. As has been seen, it is in these latter periods when the different internal response of the soils can be appreciated, even in cases in which they do not show marked physical differences. From a practical point of view, these observations have clear implications in tasks such as the estimation of soil moisture contents by remote sensors, where optimum sampling design is required for the collection of information for the calibration and validation steps.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
In this work we show the appropriateness and usefulness of the methodology used, which is based on the pioneer research work of Vachaud et al. (1985). In extensive areas, such as that studied here, it is of great interest to be able to identify which locations are representative of both the mean soil moisture conditions and of the wet and dry conditions. Setting up networks of stations for soil moisture measurements on a continuous basis is costly, tedious, and time-consuming, and hence the identification of representative stations can be highly advantageous.

Here was observed that the patterns of temporal stability persist across the whole time period analyzed. The stations preserve their characteristics regardless of the time, even under extreme conditions of soil moisture content. This persistence can help to reduce the degree of uncertainty in sampling.

The results obtained here also show that during dry periods temporal stability is more pronounced. There is a clear correlation between mean soil moisture contents and variance for the whole measurement range considered. It was also observed that the locations representative of dry conditions are always more stable, and locations representative of wet conditions are less stable. This phenomenon was confirmed for all the measuring depths and for the different periods into which the dataset was divided.

One very striking issue is the observed confirmation that temporal stability is always lower during the transition periods between dry and wet soil moisture status. The periods in which the profile is being recharged highlight the different responses of the different soils much more than the other time periods considered. This temporal pattern is of great interest for the design of sampling schemes. It is clear that when the soil is dry, homogeneity is greater and that it is in the transition periods when uncertainty is greatest. With a view to estimating soil moisture for large areas, it would be appropriate for sampling to be performed under circumstances that will guarantee the least variability.

When indirect methods are used to estimate soil moisture in broad areas (for example, remote sensing) it is necessary to perform a careful selection of the periods in which they are to be used for the calibration and validation steps. Thus, it would be necessary to bear in mind the different patterns of temporal stability as a function of the soil moisture content and the different responses of the soils during periods of transition from dry to wet conditions. If the temporal stability of soil moisture were included in the sampling design, the level of uncertainty in the estimations would probably be reduced.


   ACKNOWLEDGMENTS
 
This study was fully supported by the Ministerio de Ciencia y Tecnología (REN2000-1157-HID Project) and the Junta de Castilla y León (SA55/00A Project). We thank Miguel Ángel Luengo, Carlos Yuste, Segismundo Casado y Carlos Morán for assistance in the fieldwork, and two anonymous referees for their suggestions.

Received for publication November 8, 2002.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Soil Sci.Home page
A. Douaik, M. Van Meirvenne, and T. Toth
Statistical Methods for Evaluating Soil Salinity Spatial and Temporal Variability
Soil Sci. Soc. Am. J., August 27, 2007; 71(5): 1629 - 1635.
[Abstract] [Full Text] [PDF]

Home page
 Vadose Zone JHome page
H. Lin
Temporal Stability of Soil Moisture Spatial Pattern and Subsurface Preferential Flow Pathways in the Shale Hills Catchment
Vadose Zone J., January 1, 2006; 5(1): 317 - 340.
[Abstract] [Full Text] [PDF]

Home page
 Soil Sci.Home page
Ya. A. Pachepsky, A. K. Guber, and D. Jacques
TEMPORAL PERSISTENCE IN VERTICAL DISTRIBUTIONS OF SOIL MOISTURE CONTENTS
Soil Sci. Soc. Am. J., March 1, 2005; 69(2): 347 - 352.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (15)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Martínez-Fernández, J.
Right arrow Articles by Ceballos, A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Martínez-Fernández, J.
Right arrow Articles by Ceballos, A.
Agricola
Right arrow Articles by Martínez-Fernández, J.
Right arrow Articles by Ceballos, A.
Related Collections
Right arrow Water Content
Right arrow Time Domain Reflectometry, TDR
Right arrow Soil Hydrology

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Vadose Zone Journal Journal of Plant Registrations
Journal of Natural Resources
and Life Sciences Education		 Journal of
Environmental Quality