HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text Free Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (10) Citing Articles via Google Scholar Google Scholar Articles by Cassiani, G. Articles by Gallotti, L. Search for Related Content PubMed Articles by Cassiani, G. Articles by Gallotti, L. GeoRef GeoRef Citation Agricola Articles by Cassiani, G. Articles by Gallotti, L. Related Collections Water Content Ground Penetrating Radar, GPR Inverse Procedures/Parameter Estimation

Vertical Radar Profiles for the Characterization of Deep Vadose Zones

^a Dipartimento di Scienze Geologiche e Geotecnologie, Università di Milano Bicocca, Italy
^b European Centre for Training and Research in Earthquake Engineering, Pavia, Italy
^c Dipartimento di Scienze dell'Ambiente e del Territorio, Università di Milano Bicocca, Italy

View larger version (58K): [in a new window]   Fig. 1. Map of the Trecate site with vertical radar profiling (VRP)-monitored boreholes. Tr 24 is the well that caused the oil spill in 1994.

View larger version (28K): [in a new window]   Fig. 2. Example of error analysis on vertical radar profiling (VRP) data at the Trecate site. The black dots are the absolute values of arrival time differences obtained by rotating the transmitter antenna at the surface by 90° in azimuth. The gray continuous lines represent standard deviation of arrival times computed over moving windows of different size along the vertical direction. A comparison between moving window standard errors and difference in arrival times at different transmitter azimuth (around 0.5 ns) indicates that the actual resolution of VRP data at this site is about 25 cm. Data were collected at 5-cm vertical spacing.

View larger version (43K): [in a new window]   Fig. 3. Computation of travel time error estimates using vertical moving windows: Borehole BB, 31 Mar. 2003. Note how the 1-m window, equal in size to the deployed antenna, overestimates the data error: within 1 m the trend in the data is clearly identifiable as a signal above the error level. On the contrary, arrival times within a window of 0.25 m are within the error range (0.5 ns).

View larger version (31K): [in a new window]   Fig. 4. Dewowing of vertical radar profiling (VRP) data from Trecate. The upper row of figures refers to data from 1.5 m b.g.l.; the lower row refers to 8.5 m b.g.l. Plots (a) and (e) are raw field data. (b) and (f) are the corresponding dewowed data using a residual median filter with length equal to 80 ns. (c) and (g) are dewowed data using a residual mean filter with length equal to 80 ns. (d) and (h) are dewowed data using a residual mean filter with length equal to 10 ns. The detrimental effect of the "wow" is more serious at depth, requiring that dewowing is applied (note that the "wow" in (e) makes first break picking impossible). The only dewowing algorithm that worked properly for all depths was the residual median filter (b) and (f). Spurious precursors were introduced by the residual mean filters.

View larger version (95K): [in a new window]   Fig. 5. Examples of vertical radar profiling (VRP) data from the Trecate site. Time zero correction and trace normalization were applied to all datasets. Data in (a) and (b) were processed with residual median dewowing, 80-ns filter length, while the corresponding data in (c) and (d) were processed with residual mean dewowing, 10-ns filter length. The latter approach was more effective at preserving reflected events. Note the distinct slope change in the first arrivals in correspondence of the water table depth and the clear up-going reflections at several depths (particularly around 2 and 6 m b.g.l.). See Fig. 7 for the geometry of reflected events.

View larger version (15K): [in a new window]   Fig. 6. Scheme considered for vertical radar profiling (VRP) data inversion, from first arrival times to interval radar velocity values. In this study the thickness hi was 5 cm.

View larger version (30K): [in a new window]   Fig. 7. Schematic representation of reflection mechanisms in vertical radar profiling (VRP) and the resulting zero-offset radargram.

View larger version (76K): [in a new window]   Fig. 8. Sequence of vertical radar profiling (VRP) processing to extract up-going reflections from data collected 31 Mar. 2003: (a) data after dewowing and static shift for instrument time-zero; (b) corrected for statics computed on first arrivals; (c) reflection enhancement via residual median filtering; and (d) reflections double corrected back with statics from first arrivals. From (d) the depth of the reflecting horizons can be read at the intersection between the corrected first arrival curve (black line) and each reflected event. Note that each reflection is represented by several peaks and troughs because of the wavelet of the radar signal. Uncertain reflectors are shown with a question mark.

View larger version (23K): [in a new window]   Fig. 9. Lithology at the Trecate site, derived from drilling logs and corrected via vertical radar profiling (VRP) reflection analysis for the two key boreholes considered in this study.

View larger version (33K): [in a new window]   Fig. 10. Moisture content profiles at different times for Borehole BB, Trecate, derived from vertical radar profiling (VRP) via Occam's inversion and conversion from dielectric properties to moisture content.

View larger version (26K): [in a new window]   Fig. 11. Comparison between vertical radar profiling (VRP)-derived moisture content profile for Borehole BB and the corresponding best result from Monte Carlo Richards' equation simulations: (a) 31 Mar. 2003; (b) 11 June 2003; (c) 26 Aug. 2003; (d) 3 Nov. 2003. The moisture content profile obtained with Occam's inversion is shown with a thick line, and relevant 95% confidence bands by thin lines. The best overall Monte Carlo simulation result is shown as a thick dashed line.

View larger version (21K): [in a new window]   Fig. 12. Comparison between the vertical radar profiling (VRP)-derived moisture content profile at Borehole BB (thick line and confidence band with thin lines), the corresponding best result from Monte Carlo Richards' equation simulations (thick dashed line), and a deterministic Richards' equation simulation based on the data available without VRP information (thick dot-dashed line). The deterministic simulation uses lithological boundaries derived from drilling logs, and the following hydraulic conductivity estimates: K1 = 10 cm d–1, K2 = 50 cm d–1, and K3 = 340 cm d–1 (value derived from in situ pumping test).

View larger version (75K): [in a new window]   Fig. 13. Dotty plots of the efficiency vs. (a) saturated hydraulic conductivity of Material 1 (topsoil); (b) saturated hydraulic conductivity of Material 2; (c) saturated hydraulic conductivity of Material 3; (d) ratio of saturated hydraulic conductivities of Materials 2 and 3. Note that while the hydraulic conductivity of the topsoil is poorly identified by the flat upper envelope of the dotty plot, the other three plots show distinct maxima.