COMPARACIÓN ENTRE EL DESEMPEÑO LABORAL Y LAS FORMAS DE

4.5.1. Spatial location and relative elevation.

The relative positions of the sample sites were obtained by compass bearings and pacing. This information is usually sufficient for studies of spatial variation. However, in order to derive topographic variables to relate to soil movement processes, more accurate means of location and calculation of relative elevation are necessary. Such data can be produced from either ground survey or aerial photographs. In combination with ground control points, the la tte r can very effectively produce the necessary data. However, photographs of th e study area are too large-scale to be of use for accurate photogrammetry.

A ground-based survey was undertaken over the study area for subsequent calculation of topographic variables at each sample site. Four positions were used to survey the nested grids and parts of the main sampling grid. The range of the survey a t each position was limited to approximately 400 m due to th e quality of the theodolite optics. Siting problems were introduced because of the occurrence of large and dense Guiera Senegalensis bushes. F u rth er compounding these problems was the lack of stadia hairs in the theodolite, which allow rapid tacheom etric m easurem ents to be m ade (Wilson, 1987). For this reason the tangential system of surveying was used. U nfortunately, this method is hardly tacheometric due to the necessity for the m easurem ent of two angles as far apart, on the staff, as possible. Problems w ith refraction, non-verticality of the staff and displacement between th e m easurem ent of each angle also lead to greater inaccuracies in comparison w ith other methods (Wilson, 1987). The accuracy of the survey was estim ated to be approximately ± 0.07 m.

h ad an adverse affect on the results. Several critical m easurem ents relatin g the different positions of the theodolite were erroneous. In addition, compass bearings used to relate azim uth m easurem ents to the magnetic-north co-ordinate system were lost in transit.

These reference d a ta were collected during the second field visit using a M agellan Global Positioning System (GPS) receiver. The accuracy and precision of th is system increases w ith the am ount of equipment. Provision was made to collect d a ta which had a sim ilar accuracy to the surveying already conducted. To provide survey data w ith sim ilar accuracies was impractical because a location w ith known co-ordinates could not be identified in the field. Most single terrestrial receivers have a reported accuracy of approximately ± 30 m. However, the Magellan system has the ability to average position locations in Differential Mode to obtain more accurate horizontal positions, sometimes approaching 10 m. This is especially the case when the satellite alm anac is consulted to achieve the optimal precision dilution of position (PDOP). The positions provided by the GPS in this study were obtained during a period when the lowest PDOP was available, resulting in optimal precision given th e lim itations. The GPS positions were standardised according to the map projection Universal Trans-M ercator (UTM). In addition, the GPS was set to average over 200 positions which m eant th a t the average horizontal position and standard deviation for each of the reference sites was produced (Table 4.5.1.1)

Table 4.5.1.1. Survey reference locations and error estim ates in horizontal and vertical planes. Location Northings UTM (m) Eastings UTM (m) Sd. Elevation m above datum 1 31 1499162 453513 31.5 0 2 31 1499291 453279 24.5 +1.59 3 31 1499226 453050 34.2 -8.26 4 31 1498996 453188 45.7 -7.77

These positions in combination w ith other d a ta were used to calculate the relative height difference between each theodolite location and provide the extremes in error.

In the case of location 1, the position was obtained from a lengthy averaging process of 1500 positions which also ascertained the absolute height above sea level.

4.5.2. Vegetation cover and type.

Vegetation type was observed a t each site and classified according to herbs, grasses and trees and bushes. Total vegetation cover was expressed as a percentage of a 1 qu ad rat oriented north. Estim ation of coverage was very approxim ate and difficult to gauge due to the difference in density of a particular vegetation type. The herbaceous layer was often found to be widespread but not very dense, w hilst bushes and trees often covered the quadrat very densely but were sparsely located throughout th e study area. The main problem in estim ating total vegetation cover was related to the area of bare ground w ithin the quadrat.

The tim ing of this vegetation survey is im portant as a distinct tem poral variation exists w ithin and between seasons. A retu rn visit to the study area th e following year early in the growing season revealed a significant reduction in the coverage of herbs and grasses. The coverage of bushes and trees rem ained the same. For this reason analyses were conducted on total vegetation cover to provide a statistic w ith less tem poral variation th an the other measurements.

4.5.3. Soil strength.

Soil strength m easurem ents are often used as an indicator of soil structure which m ay change w ith different land uses, becoming more compacted by vehicles and to a lesser extent by animals. In recent years the strength of soils has been used as a m easure of the apparent cohesiveness of surface layers (Mah et al., 1992).

Although m any methods for m easuring soil strength exist (Walley, 1990), few have been designed for use in the field. The shear vane provides rapid estim ates of soil stren g th away from the laboratory. The vane is pressed into the soil u n til covered or a t th e required depth and a torque is applied to the shaft. The torque is increased until, a t a m axim um value, the soil shears along a cylindrical surface enclosed by the

vane. The angular deflection of a calibrated spring is the usual way to determ ine the torque. This torque (T), m easured in kilo Pascals, is converted into undrained shear stren g th (C J when the height (h) and diam eter (d) of the vanes are known from :

C„ = T / [ Tcd^ (h/2 = d/6) ] (4.5.3.1)

The selection of the vane size is dependent on the strength of th e soil being m easured. For consistency, a single rod with a blade w idth of 1.9 cm was used for all m easurem ents. At each site and a t the same relative position, 20 cm north of the sample hole, a m easurem ent for soil strength was made using the sh ear vane. Three depth increm ents, 5 cm, 10 cm and 15 cm, were taken, m easured from the base of the rod. Problems associated w ith this device involved th e insertion of the rod into the soil w here crusts had developed or exposed B horizons were present. A t only a few sites were the m easurem ents out of range due to these surface features.

An approxim ate estim ate of m easurem ent errors is provided by replicate m easurem ents w ithin two areas of 0.16 m^ each. A total of 7 m easurem ents were conducted in each area. Unfortunately, nothing is known about th e population from which these sam ples were drawn. In this case the random variable X, from which p and a are finite can produce estim ates of E[X] = p and var[X] = / n. Although no assum ptions are made about the distribution th a t X follows reasonable approximations of the actual distribution of E[X] are still available. The stan d ard deviation of th e sampling distribution (standard error) of a statistic such as the m ean provides a m easure of the reliability of the statistic. Since two locations require comparisons and the mean varies between sites, the pooled w ithin-site variance was calculated. The standard error of the mean in a random sample of size n is o / V n. The results in Table 4.5.3.1 show th a t the m easurem ents of th e soil strength of the upperm ost 5 cm are more accurate th an the m easurem ents a t depth. This suggests th a t the upper 5 cm of the soil is less variable th an th e soil a t greater depth. This level of accuracy suggests th a t differences of approxim ately 0.9 k P a may be due to significant differences in soil strength rath er th a n to m easurem ent error.

Table 4.5.3.1. Accuracy of replicate soil property field m easurem ents.

Soil property n E[X] Var[X] SE[X] % cv

Soil stren g th (5 cm) 14 18.14 9.30 0.88 17

Soil stren g th (10 cm) 14 19.57 23.70 1.40 25

Soil stren g th (15 cm) 14 22.14 22.80 1.38 22

Fx 49 33.26 18.53 0.38 13

Bulk density 14 1.52 0.26 0.14 34

4.5.4. M ineral magnetic susceptibility (F%).

Magnetic susceptibility was m easured in the field w ith a Bartington Instrum ents survey loop connected to a portable m eter a t a scale of 0.1. The search loop produces a spherical field which integrates the effect of magnetic m inerals a t an average depth of approximately 10 cm below the surface (pers. comm. J. Dearing). M easurem ents were tak en a t th e four corners of a 1 m^ quadrat placed around the sample hole and oriented to north. Surface vegetation influences the m easurem ents and was removed in order to place the search loop directly on the soil surface, to obtain consistent results. Air readings were taken before and after the m easurem ents on the quadrat to estim ate the background level and account for wandering of th e m eter. The average of the background readings was then subtracted from the average of th e four surface m easurem ents to give the field magnetic susceptibility m easurem ent. Since a single m easurem ent was obtained from the four readings on a q u ad rat a t each site it may be assumed magnetic susceptibility is uniform over the quadrat. In this case the field m easurem ents reflect m easurem ent error a t each site over the entire study area. Although, only sites from the m ain sampling grid were included, the m ean varied considerably as shown by the pooled w ithin-site variance (Table 4.5.3.1). However, the coefficient of variation is small in comparison w ith other replicate m easurem ents suggesting th a t the m easurem ents are not affected by the m easurem ent error.

4.5.5. Soil colour.

Hue is a m easure of redness or yellowness of soil colours. Chrom aticity of soil colours reflects the relative purity of the dom inant wavelength, represented by hue. The colour value is indicative of lightness of the soil colour. The three components of soil colour are sensitive to the concentrations of particular m inerals (especially iron oxides of clay size) and organic m atter. The accumulation is more or less dependent on redox conditions which in tu rn are dependent on drainage conditions (G errard, 1981).

An attem p t was m ade to record the Munsell soil colour of samples im m ediately after extraction. However, the angle a t which the soil was viewed combined w ith th e angle of the sun m ade it difficult to obtain repeatable m easurem ents in situ.

M easurem ents were therefore conducted in the laboratory using consistent lighting. To avoid bias, m easurem ent was conducted on samples w ith locations obscured and by random selection.