overland flow and erosion in agricultural lands -...
TRANSCRIPT
1
1
Overland flow and erosion in agricultural lands
Process, model, assessmentPaul van Dijk
A.Véronique Auzet
FORM-OSEPost-Graduate Training School“Living with hydro-geomorphological risks :From theory to practice”
2
Structure
a. Introduction, processes and factors
b. Impacts
c. Models and applications
3
Structure
a. Introduction, processes and factors
b. Impacts
c. Models and applications
4
Soil erosion is an ”old problem”on steep slopes
in semi-arid regions
Recognised more recently In loess regions in the European temperate climate zone with moderate relief
Focus of today’s course
5 6
The french example, inventory based on reports
Source :Auzet A.V., 1989.In Le grand atlas de la France rurale.(Eds INRA). J.P. de Monza, Paris.
2
7
Background of erosion problems in temperate climate
Change from forest/pasture towards arable land use
<< org.matter on the surface
>> Temp (exposure of the soil surface)
>> Aeration (tillage)
Exposure of the soil surface to raindrop
impact during part of the year>> activity of soil micro-biology
>> decomposition of mat. org. by mineralisation et oxidation
<< supply of org. matter in the soil
<< organic bounds in the soil
<< aggregate stability
>> crust formation
<< infiltration
>> overland flow >> splash erosion
>> erosion8
Why in areas with loess-derived soils (1) ?
Source : Kihlbom 1970
9
F0
F1
F2
Scodro (2001)
Low structural stability
Why in areas with loess-derived soils (2) ?
10
Soil surface conditions, crusting
11
Timing aspects : climat and agriculture(ex: Bretagne © C. Gascuel, 2001)
SowingWeeding harvesting
Rainfall characteristics
Soil conditions: roughness and infiltrability
Vegetation cover
Successive agricultural operationsSoil and hydrologic conditions
Winter:hydrologic conditions
Spring: a quick soil desagregation
Two key-periods 12
The processes
• Splash erosion
• Overland flow generation Hortonien (rainfall intensity > infiltration cap)
Rainfall intensity, soil surface state
Saturation (soil pore space is filled with water, additional rainfallcannot infiltrate)
Rainfall quantity, soil moisture storage capacity, impeding layer
Exfiltration
Snowmelt (on frozen soil)
• Soil detachment by overland flow
• Transport
• Deposition
3
13
Erosion
Deposition
Transport
Sediment delivery
Spatial aspects, scales
14
Upstream overland flow formationand diffuse erosion
15
Overland flow concentration and rill erosion
Photos: A.V. Auzet 16
Transport phase
Photos: A.V. Auzet
17
Sediment deposition
Photos: P.M. van Dijk
18
Sediment delivery to streams
4
19
Structure
a. Introduction, processes and factors
b. Impacts
c. Models and applications
20
Impacts
• On siteSoil loss, reduced soil depth
Loss of nutrients
Modified physical and chemical properties, AWC
Destruction of crop
In case of gullies : problematique access with agr. machines
• Off siteSediment and pollutants transfers
Surface water quality and ecosystem degradation
Damage to infrastructure and built-up areas (muddy flows)
• Direct and indirect economical costs for communities and individuals
21
Damage depends on the land occupation
22
Damage in fields
Blotzheim, juin 2003© R. Armand
23Blotzheim, juin 2003 (©R. Armand) 24
Damage to infrastructure © P. van Dijk
5
25
Damage in built-up areas Landser, 25 mai 2001
© Lemmel M., Scodro E.26Blotzheim, juin 2003 © R. Armand
27 28
Water quality impacts and effects on aquatic ecosystems
La Dijle (Belgique)Mai 1986
29
Structure
a. Introduction, processes and factors
b. Impacts
c. Models and applications
30
Models, in order to :
• Identify problem areas (target areas for measures)
• Increase our knowledge
• Evaluate and predict (effects of measures, catchment management, climate/land use change)
At different spatial scales (field, catchment, region…)
At different time scales (minutes to years)
6
31
Model types
• empiricalUSLE, RUSLE, MUSLE
• deterministicCREAMS, AGNPS, SWAT, SWIM
• physically-basedWEPP, KINEROS, EUROSEM, LISEM, SHE
• stochasticRillGrow…
32
Models, objectifs, scales
Rillgrow, 3 m2LISEL, 4 km2
33
Two models, two spatio-temporal scales
• LISEM: the Limburg Soil Erosion ModelEvent (time step : seconds)
Physically-based
Distributed
Catchment-scale (raster cell size 5 to 50 m)
• RECODES: climate and land use change impacts on sediment delivery to streams in the Rhine basin
Time step : 1 month
Distributed, (raster cell size : 1 km2)
To reply to water management questions for the Netherlands in theperiod untill 2100
Model 1 : LISEM
35
Model 1LISEM : GIS embedded catchment model
DEM
Land use
Soil
Rainfall
Erosion
Overland flow
Hydrogramme
Deposition
Input Output
36
Flowchart of LISEM
7
37
LDD (Local Drainage Direction)
The LDD defines flow direction in all cells and thus spatial structures
38
Basic data
Rain of ruturn period 5 years
0
1
2
3
4
5
6
7
8
618 30 42 54
duration (min)
pluv
iom
etry
(mm
)
rain of return
period 5
years
DEM Soil
Land use
Prec.
39
Model parameters derived from soil type
1 2 3 4 5 6D50 (mu) 22,30 23,65 22,42 26,30 23,12 22,30KSAT F(0) (mm/h) 21,10 22,50 20,70 28,60 21,50 22,00KSAT F(1) (mm/h) 10,00 10,00 10,00 10,00 10,00 10,00KSAT F(2) (mm/h) 2,00 2,00 2,00 2,00 2,00 2,00KSATCRST (mm/h) / / / / / /KSATWT (mm/h) 2,11 2,25 2,07 2,86 2,15 2,20PSI (cm) 100 100 100 100 100 100SOILDEP (mm) 200 250 350 300 300 300THETAI (-) 0,30 0,30 0,30 0,30 0,30 0,30THETAS (-) 0,48 0,50 0,51 0,47 0,49 0,50
LISE
M
SOIL TYPE
40
Model parameters derived from land use
AGGRSTAB (-) 19 18,5 14 -1 -1 -1 -1 -1
CH (m) 0 0 0,35 0 0,05 10 15 0
COH F(0) F(1) (kPa) 0,69 0,47 0,91 0,51 3,32 1,93 1,93 9999
COH F(2) (kPa) 0,75 0,52 0,95 0,57 3,32 1,93 1,93 9999
COHADD (kPa) 0 0 0,8 0 3,32 1,45 1,45 9999
LAI (m2/m2) 0 0 1,1 0 1,86 11 11 0
n (-) 80 115 123 120 227 300 300 0,001
PER (-) 0 0 0,4 0 0,95 0,95 0,95 0
STONEFRAC (-) 0,05 0,05 0,1 0,05 0,05 0,05 0,05 0 et 1
RR F(0) (cm) 0,89 1 0,48 0,97 0,7 0,73 1,36 0
RR F(1) (cm) 0,29 0,35 0,48 0,97 0,7 0,73 1,36 0
RR F(2) (cm) 0,12 0,25 0,48 0,97 0,7 0,73 1,36 0
Betterave
Les chiffres en italique sont ceux proposés par la table de LISEM pour la deuxième quinzaine du mois de mai. Leschiffres en gras correspondent à des valeurs que nous avons mesurées, observées ou bien choisies en fonction decritères prédéfinis.
Blé JachèrePlan d'eau et habitat
LISE
M
Bande enherbée/Prairie
Verger ForêtMais
OCCUPATION DU SOL
41
Application of LISEM, no. 1
Effects of soil surface conditions
(Mickaël Lemmel, 2001)
42
F0
F2
Runoff
8
43
Erosion
G1AF0
G1AF1
G1AF2
G1AF0
G1AF1
G1AF2
-Initial soil surface state
- Moderate degradation
-Strong degradation
44
Application of LISEM, no. 2
LISEM + TCRP (tillage controlled runoff pattern)
Accounting for tillage structures effects on runoff patterns and erosion
45
Dead furrow
Headlands 46
47 48
Effects of tillage direction
LDD topographique TCRP
Structures de drainage de la partie amont du BV GUTZ
9
49
Runoff distribution over the catchment
LDD modifié avec TCRPLDD topographique
Lame d’eau ruisselée (mm)
0 500
N
mètres
•Different spatial structure
•Overland flow concentration often occurs at field boundaries 50
topography tillage
Takken et al., 1999
51
Application of LISEM, no. 3
Scenarion studies at the catchment scale
(Marc Pichaud, 2001)
52
Secteur de l’Outre Forêt
53 54
10
55
reduction of eroded surface (<0.5tonne/ha)
05
101520253035
band
e enhe
rbée
tcs à
20 %
15 di
gues
be+tcs
20%
be+di1
5
tcs1+
di15
tcs2+
di15
be+tcs
2+di1
5
scenarii
redu
ctin
g fa
ctor
(%)
2 years5 years10 years
Scenarios with different combinations of : - grass strips- reduced tillage- retention structures (dikes)
56
reduction of peak discharge
05
1015202530354045
grass
strips
(gr)
mulch (2
0%)
15 da
ms
gr + m
ul.
gr + da
ms
mul.(10%)+d
ams
mul.(20%)+d
ams
gr+mul.+
dams
scenarii
redu
ctio
n fa
ctor
(%)
2 years5 years10 years
57
Model 2: RECODES (GAMES + USLE)
Climate and land use change : impacts on sediment delivery in the Rhine basin
(Van Dijk, 2001)
58
0 100 km
THENETHERLANDS
Andernach
Kaub
SWITZERLAND
FRANCE
Rheinfelden
GERMANY
LUXEMBURG
AUSTRIA
Cochem
Worms RockenauKleinheubach
Viereth
SchweinfurtKalkofen
Diepoldsau
Approach
• fully distributed model with a time step of one month
• a sediment production module to quantify mobilisation of sediment by soil erosion at the hillslopescale
• a module accounting for sediment storage through a delivery ratio expression
• sediment supply: between or within cells
• written in PCRaster dynamic modelling language
60
The basic modules
• Sediment production: USLE, RUSLE, ABAG
• Delivery ratio: GAMES, the fraction of mobilised sediment that actually reaches the stream depends on:
proximity of source to stream
the probability of overland flow occurrence
the character of the terrain along the route towards the channel(roughness, slope angle)
11
61
R-factor
62
K-factor
63
Snowmelt effects are taken into account
64
Hydrological coefficient described the availability
of overland land to transport sediment towards streams
65
Sediment source to stream distances are estimated
using a detailed drainage network map
66
Local annual erosion
Sediment delivery to streams
Source: Van Dijk (2001)
12
67
0
5
10
15
20
25
30
35
N D J F M A M J J A S O
mois
éros
ion
(t/km
2)
Neckar
Aare (UL)
Mosel
Temporal variations
Source: Van Dijk (2001) 68
-30
-20
-10
0
10
20
30
40
CPC+UKHI(high)
CPC+UKHI(central)
CPC+UKHI(low)
UKHI (high) UKHI (central) UKHI (low) CP
scenario
Effe
t sur
l'ap
port
de
sédi
men
t (%
)
climate and land use change only climate change only land use change
best estimate
Results of scenario calculations
Source: Van Dijk (2001)
69
Attention! Errors and uncertainty!
• Errors in input data
• Model errors
Never consider model output as the truth
Uncertainty analyses ! (such as Monte Carlo analyses)
70
0
500
1000
1500
2000
2500
sim
ulat
ed
0 500 1000 1500 2000 2500
measured
cal
val
Total Discharge (m3)
Example of Govers and Jetten
71
0
100
200
300
400
500
600
700
0 100 200 300 400 500 600 700
sim
ulat
ed
measured
cal
val
Peak Discharge (l/s)
Example of Govers et Jetten72
0
10
20
30
40
0 10 20 30 40
sim
ulat
ed
measured
cal
val
Total Soil loss (ton)
Example of Govers et Jetten
13
73
Why do erosion models perform poorly ?
Source: Govers et Jetten74
Deterministic, spatio-temporal models require many input parameters which all contain some
degree of errors. Model output can be very sensitive to some of them (non-linearity)
Source: Govers et Jetten
75
In case rainfall or infiltration has an error of 5,10,15 et 20%...
Error propagation
Source: Govers et Jetten76
error propagation h->V->Q
0.0
100.0
200.0
300.0
400.0
0 5 10 15 20 25
relative error input (%)
rela
tive
erro
r out
put (
%)
Q V h
…the error in modelled catchment discharge can be > 300% !
Source: Govers et Jetten
77
Example: sensitivity to Ksat
Source: De Roo et al (1995)78
Incertainty / model complexity
Source: Grünwald (1997)
14
79
Thanks !