Passive capillary pan samplers : an efficient system to monitor in-situ percolation fluxes in soils

(1)

Passive capillary pan samplers :

an efficient system to monitor in-situ

percolation fluxes in soils.

System presentation and estimation of measurement

uncertainties

J.-G. Lacas1,2_{, M. Voltz}2_{, P. Cattan}3_{, X. Louchart}2

1. CEMAGREF, UR Qualité des Eaux et Prévention des Pollutions, Lyon, France. 2. INRA, UMR L.I.S.A.H., Montpellier, France

3. CIRAD FLHOR, Capesterre Belle Eau, France

Session 1 : Alternative and innovative approaches for characterising the pollution level and pollution hasard of water ressources

(2)

THE SCIENTIFIC CONTEXT :

Measurement of water and solutes percolation fluxes in

the unsaturated zone of soils

Many interests in hydrology :

•groundwater recharge

•mass balance for a specific layer

•leaching hazard for chemical contaminants

groundwater Unsaturated zone of soil rootzone Chemicals applied (pesticides) Soil contamination Groundwater contamination

2

(3)

THE TECHNICAL CONTEXT :

Few field technics :

•Tensiometry

•Suction cups lysimeter

•Weighing lysimeter

•Zero tension pan sampler

Local (spatial variability), C_FLUXneeded, K-h needed (source of uncertainty)

Local (spatial variability), C_RESIDENT, flow distortion, concentration effect.

Soil destruction (stabilization period), flow pattern distortion, cost)

h=0 ⇒undersampling (e≈50%), flow distortion (divergent flows)

Poorly representative measures :

(4)

PRESENTATION OF THE PASSIVE CAPILLARY SAMPLER

Sample extraction and pressurisation tubes Unraveled fibers contacting soil Installation cavity Wick support Wick tubing Sample collection bottle Hanging water-column [50cm-90cm] Collection depth [50cm-2m]

Typical design of a passive capillary sampler (PCAPS)

Photography of a sampler during installation under a banana tree.

4

(5)

PRESENTATION OF THE PASSIVE CAPILLARY SAMPLER

Photography of the sampling bottles

and extraction tubes – lysimeter

installed under a grassed zone.

Lacas - CEMAGREF

(6)

PRESENTATION OF THE PASSIVE CAPILLARY SAMPLER

Photography of a wick lysimeter with two cells before installation – unraveled fibers

(7)

PRESENTATION OF THE PASSIVE CAPILLARY SAMPLER

Photography of the lysimeter in contact with the overlying soil – the collector is pressed into the soil by wood blocks

(8)

HOW DO PCAPS COLLECT SOIL FLUXES ?

Typical pressure head profiles within a fiberglass wick under different soil flux intensities.

Wick

length

(cm)

Pressure Head (cm H2O)

Succion exerted on the overlying soil by a 100cm long wick

0 20 40 60 80 100 -100 -80 -60 -40 -20 0 q_s=0,001cm/h q_s=0,01cm/h q_s=0,1cm/h Soil flux : Wick conductivity : Ksat = 460cm/h Collection area : Wick area :227 cm 2

5

0,7 cm2

(9)

CARACTERISTICS OF THE PCAPS MEASURE

•Unsaturated and saturated fluxes (macropores and matrix) •Non-disturbed soil volume

•Significant area (typic. 900cm2_).

•No divergent flows •Continuous sampling •Non sorbing material

•No additional delay nor dispersion in solute breakthrough

Very representative measurement of leaching (e≈100%)

•Well suited for field studies (no external vacuum generator needed, cheap materials)

(10)

0 20 40 60 80

Flux from 17-Sep-2001 10:30:00 to 14-Dec-2001 08:20:00

R a in fa ll ( m m/d a y ) 0 50 100 150 200 250 W a te r fl u x ( m m/d a y ) 16/09 26/09 06/10 16/10 26/10 05/11 15/11 25/11 05/12 15/12 10-2 100 102 104 C a d u sa fo s co n c. (µ g /L )

EXAMPLE OF MEASUREMENTS (Guadeloupe)

6

Wick lysimeter

Zero tension lysimeter

Wick lysimeter

(11)

PCAPS MUST BE DIMENSIONNED

In order to achieve a minimal disturbance of the native flow regime,

the wick material, wick length, wick number and the collection area have to be dimensioned

so that the “soil pressure vs. flow” conditions of the wick match those of the soil.

Analytical equations have been proposed in the literature :

area Wick K area Plate K wicks of number wick s soil s × × = • Holder et al. (1991) -100 -80 -60 -40 -20 0 -1 -0,8 -0,6 -0,4 -0,2 0 Soil material

Wick system n°1 : well matched Wick system n°2 : too much resistive Wick system n°3 : too much conductive

Downward soil water flux (cm/h)

Pressure Head (

c

m H2O)

• Knutson and Selker (1994)

⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⎟⎟ ⎠ ⎞ ⎜⎜ ⎝ ⎛ − ⎟⎟ ⎠ ⎞ ⎜⎜ ⎝ ⎛ + = wsat w s s wsat w s s w w w wick K A A q K A A q L a a h 1 ln exp( ) 1 ⎥ ⎦ ⎤ ⎢ ⎣ ⎡− = sat s s s K q a h 1 ln K(h) = Ks eah

7

(12)

QUESTION OF MEASUREMENT UNCERTAINTY :

PCAPS are widely used to determine percolation fluxes… but there isn’t any data on the measurement uncertainty.

two main sources of errors :

•theoretical hypothesis underlying the analytical equations proposed by Knutson and Selker (1994) to dimension PCAPS.

(unit gradient in soil, permanent flow, 1D flow, Gardner model)

•experimental errors associated to parameters used for dimensioning or associated to the PCAPS installation procedure in field.

(soil Ks, wick Ks, wick length, effective collection area…)

(13)

ERROR ANALYSIS : A NUMERICAL METHODOLOGY

Boundary conditions used for « soil with a wick sampler » and « reference soil ». We used the Hydrus 2D code (Simunek et al., 1999), solving the

Richards equation without hypothesis for simulating two-dimensional unsaturated flow,

to analyze the deviations of a soil-instrumented system in comparison to an undisturbed system, in realistic conditions.

0 4 8 12 16 0 100 200 300 Time (h) R a inf a ll ( c m/ h )

Surface condition applied

(14)

ERROR ANALYSIS : RESULTS 1

• no significant bias on the wick-extracted volumes with Ida Silt Loam (2%) •The small bias observed with Rehovot Sand (19%) is due to the fact that soil and wick properties couldn’t be perfectly matched.

10

Comparison of cumulated fluxes in both wick and

reference systems during the 11days simulation y = 1,0629x R2 = 0,9986 y = 1,1973x R2 = 0,9914 0 2000 4000 6000 0 2000 4000 6000 Reference flux (cm3) Wi c k fl u x ( c m 3 )

Rehovot Sand + Pepperell 1/2 Ida Silt Loam + Amatex 3/8

(15)

ERROR ANALYSIS : RESULTS 2

•However, the bias can be significant for short time scale measurements

Temporal variations of bias in the (Ida Silt Loam + Amatex 3/8in. Wick) system

0 80 160 240 0 50 100 150 200 250 300 350 Time (h) F lu x ( c m 3 /h ) -0,5 0 0,5 1 1,5 Wick flow Referecne flow Bias B ias ( W ic k f low /R ef er enc e f low )

11

(16)

ERROR ANALYSIS : RESULTS 3

•The typical uncertainties about the hydraulic properties of both soil and wick materials can induce large deviations of wick-extracted volumes

Error due to measurement uncertainty on soil saturated conductivity

0 3000 6000 9000 0 2 4 6 8 10 Ks / Ks reference C u m u la ted f lux (c m 3

) Ida silt loam + AM 3/8 MD

Rehovot sand + PEP 1/2

Error due to measurement uncertainty on wick saturated conductivity

0 3000 6000 9000 0,0 0,5 1,0 1,5 2,0 2,5 Ks / Ks reference C u mu la te d fl u x (c m3

) Ida silt loam + AM 3/8 MD

Rehovot sand + PEP 1/2

(17)

ERROR ANALYSIS : RESULTS 4

•An error made on the wick length can have a large impact on extracted volumes

Wick length impact on total extracted volumes

2000 4000 6000 8000 40 60 80 100 120 140 Wick length (cm) C u mu la te d v o lu me s ( c m3 )

Ida Silt Loam + AM3/8MD Rehovot Sand + PEP1/2

(18)

CONCLUSIONS

•

PCAPS are attractive systems that permit a representative

measurement of in-situ leaching fluxes (C_FLUXand water flux) in the unsaturated zone.

•The hydraulic conductivities of both soil and wick must be measured accurately to have a reliable estimation of flux. Attention should also be taken at the wick length.

• Analytical methods proposed in the litterature to dimension the lysimeter are relevant for Gardner type soils and large time-scale measurement

• A 2D numerical simulation approach should be used in most cases to accurately evaluate the PCAPS efficiency and the measurement bias. It allows to post-correct measured percolation fluxes

(19)