Bio-physical impact analysis of climate change with EPIC Presented by Christine Heumesser at the AGRODEP Workshop on Analytical Tools for Climate Change Analysis June 6-7, 2011 • Dakar, Senegal For more information on the workshop or to see the latest version of this presentation visit: http://www.agrodep.org/first-annual-workshop

1. Bio-physical
impact analysis
of
climate change with EPIC
Presented by:
Christine Heumesser
University of Natural
www.agrodep.org
Resources and Life Sciences, Vienna
AGRODEP Workshop on Analytical Tools for Climate
Change Analysis
June 6-7, 2011 • Dakar, Senegal
Please check the latest version of this presentation on:
http://agrodep.cgxchange.org/first-annual-workshop

2. Bio-physical impact analysis of climate change with EPIC
Ch. Heumesser1, E. Schmid1, F., Strauss1, R., Skalský2, J., Balkovič2, et al.
1)University of Natural Resources and Life Sciences, Vienna
Institute for Sustainable Economic Development
2)Soil Science and Conservation Research Institute (SSCRI), Bratislava, Slovakia
AGRODEP members’ meeting and workshop-
June 6-8 2011- Dakar, Senegal

3. 1. EPIC - Environmental Policy Integrated Climate
1.1. Global EPIC database

4. EPIC - Environmental Policy Integrated
Climate
Model Objectives:
 Assess the effect of soil erosion on productivity.
 Predict effects of management decisions on soil, water,
nutrient and pesticide movements and their combined impact
on soil loss, water quality and crop yields.

5. EPIC - Environmental Policy Integrated
Climate
 Developed in the 1980s as “The Erosion Productivity Impact Calculator” to
assess the status of U.S. soil and water resources (Williams et al., 1984;
Williams, 1990; Jones et al., 1991).
 EPIC compounds various components from CREAMS (Knisel, 1980),
SWRRB (Williams et al., 1985), GLEAMS (Leonard et al., 1987), and has
been continuously expanded and refined to allow simulation of many
processes important in agricultural land management (Sharpley and
Williams, 1990; Williams, 1995, 2000) => Environmental Policy
Integrated Climate (Williams, 1995).
 A major carbon cycling routine was performed by Izaurralde et al. (2006).
Current research efforts are focusing on model algorithm that address green
house gases emissions (e.g. N2O, CH4).

6. EPIC is part of a model family
Watershed Scale Field Scale:
APEX EPIC
Agricultural Policy
Environmental
Environmental
Policy Impact
eXtender
Calculator
SWAT
Soil Water
Assessment Tool

7. Major EPIC components
 weather
 hydrology (runoff, evapotranspiration, percolation)
 erosion-sedimentation (wind and water) EPIC Model (Williams, 1995)
Solar radiation
 nutrients (N, P, K) and carbon cycling (C)
Wind
 salinity Plant
Precipitation
 plant growth and competition growth
 soil temperature and moisture Erosion
 tillage & management & grazing Operations
Runoff
 cost accounting Soil
layers
C, N, & P cycling Pesticide fate
 EPIC operates on a daily time step, and is capable of simulating
hundreds of years if necessary.
 EPIC is programmed in FORTRAN

8. EPIC – Study Outline
EPIC study outline:
 An EPIC study may involve several sites e.g. field, farm, and
each site must have assigned soil, topography, weather station.
 Multiple runs may be defined for each site, with alternative
weather, or field operations schedule data sets specified.

9. EPIC: Major input datasets
1. Site Description/Topography
slope, elevation, size of field, latitude, etc.
2. Weather
monthly weather statistics (generator), daily weather records (obs. or
simulated)
3. Soil
soil texture, pH, bulk density, etc.
4. Land Use Management
crops/crop rotations,
planting & harvesting
tillage operation (time & type),
fertilization (time, type & amount), irrigation (time & amount),
grazing, trees, etc.
5. Constant Data
Number of years and beginning year of simulation, weather gen. options,
also includes information on production cost.
 Data file editor UTIL- Universal Text Integrated Language

10. EPIC - Homogenuous Response Units (HRU)
 Spatially explicit site-specific qualities
Concept of HRUs allows to consistently integrate/aggregate bio-physical
effects in economic land use models

11. EPIC - Homogenuous Response Units (HRU)
Altitude, slope and soil class
value assigned to 5’ spatial
resolution pixel represents
spatially most frequent class
value (not average)
HRU is spatially delineated as a
zone of global grid having same
class of altitude, slope and soil

13. EPIC: major simulation output (txt files)
Dry matter crop yield [t/ha] Nitrogen (& P) Balances:
Total biomass yield [t/ha] Fertilization (FTN) [kg/ha],
Water stress days [d/crop] Deposition (NPCP) [kg/ha],
Nitrogen stress days [d/crop] Fixation (NFIX) [kg/ha]
Carbon Balance: Nitrogen in Crop Yield (YLN) [kg/ha],
Carbon in Crop Yield (YLC) [kg/ha], Air Volatilization (AVOL) [kg/ha],
Carbon Respiration (RSPC) [kg/ha], Denitrification (DN) [kg/ha],
Carbon in Sediment (YOC) [kg/ha], Organic Nitrogen in Sediment (YON) [kg/ha],
Carbon in Percolation (CLCH) [kg/ha], Soluble Nitrogen in Runoff (QNO3) [kg/ha],
Carbon in Runoff (CQV) [kg/ha], Soluble Nitrogen in Subsurface Flow (SSFN) [kg/ha],
Topsoil Organic Carbon (OCPD) [t/ha] Soluble Nitrogen in Percolation (PRKN) [kg/ha],
Water Balance: Nitrogen losses through Burnning (BURN) [kg/ha]
Rainfall (PRCP) [mm], Others
Irrigation (IRGA) [mm], Sediment losses (MUST, USLE, RUSL) [t/ha]
Potential EvapoTranspiration (PET) [mm], Gross Nitrogen Mineralization (GMN) [kg/ha]
Actual EvapoTranspiration (ET) [mm], Net Nitrogen Mineralization (NMN) [kg/ha]
Runoff (Q) [mm], Nitrification (NITR) [kg/ha]
Subsurface flow (SSF) [mm], Denitrification (DNIT) [kg/ha]
Percolation (PRK) [mm] etc.

14. EPIC OUTPUT DATA

15. Integrative Climate Change Impact
Modelling
Economic-,
Thematical Bio-physical process Economic Land Use Energy &
Databases modelling Models Ecosystems models
Integration
Climate Field, Farm, FAMOS[space] CGE
Environment (BOKU) (Wegener Zentrum)
Soil and
Topography
Region & PASMA PROMETEUS
Austria
EPIC (BOKU/WIFO) (WIFO)
APEX Pixel & BeWhere MultiREG
Land use & (TAES/JGCRI/BOKU)
Countries (IIASA/BOKU) (Joanneum)
management
EU-Regions
& ROW EUFASOM ERSEM
(Uni. Hamburg, (Uni. Hamburg)
IIASA, BOKU, u.a.)
Economics &
Administration
& Statistics
Global GLOBIOM POLES
(IIASA, Uni. Hamburg, (JRC-IPTS)
BOKU, u.a.)
Scale

16. 1.1. Global EPIC database

17. Global EPIC – Land Cover
Global Land Cover (GLC2000; IFPRI, 2007)
crop lands: 0.9 bil. ha forest lands: 4.0 bil. ha
other agri. lands: 1.5 bil. ha wet lands: 0.2 bil. ha
grass lands: 1.1 bil. ha nat. veg. lands: 2.5 bil. ha

18. Global EPIC – CROPS
20 crops simulated on all GLC
=> crop production possibility set in GLOBIOM
 BARL barley
 CASS cassava  POTA potatoes
 CHKP chick peas  RAPE rape seeds
 CORN corn  RICE rice
 COTS cotton  RYE rye
 COWP cow peas  SOYB soybeans
 DRYB dry beans  SPOT sweet potatoes
 GRSG grain sorghum  SUGC sugar cane
 OATS oats  SUNF sunflower
 PMIL millet  WWHT wheat
 PNUT peanuts/groundnuts

19. Global EPIC – crop management
 3 Crop Input Systems (rule-based) simulated on all GLC
=> crop management possibility set in GLOBIOM:
o AN: automatic nitrogen fertilization – N-fertilization rates based on N-
stress levels (e.g. N-stress free days in 90% of the vegetation period). The
upper limit of N application is 200 kg/ha/a.
o AI: automatic nitrogen fertilization and irrigation – N and irrigation rates
are based on stress levels (e.g. N and water stress free days in 90% of the
vegetation period). N and irrigation upper limits of 200 kg/ha/a and 300 mm/a.
o SS: subsistence farming – no N fertilizations and irrigation.
 Global Climate Data
o using Tyndall Climate Change Data, A1fi Scenario

20. Mean Temperature change on cropland
in 2050 in °C (Base 2000)
Tyndall Climate Change Data, A1fi Scenario

21. Mean Temperature change on cropland in
2100 in °C (Base 2000)
Tyndall Climate Change Data, A1fi Scenario

22. Annual Precipitation Change on cropland
in 2050 in mm (Base 2000)
Tyndall Climate Change Data, A1fi Scenario

23. Annual Precipitation Change on cropland
in 2100 in mm (Base 2000)
Tyndall Climate Change Data, A1fi Scenario

24. Corn Yields in t/ha (DM) on cropland,
automatic fertilization and irrigation
(AI management), (Base 2000)

25. Changes in Corn Yields on cropland in
2050 in t/ha (DM), AI management system
(Base 2000)

26. Changes in Corn Yields on cropland
in 2100 in t/ha (DM), AI management system
(Base 2000)

27. Changes in irrigation water on cropland
in 2050 in mm, AI management system
(Base 2000)

28. Changes in irrigation water on cropland
in 2100 in mm, AI management system
(Base 2000)

29. 2. CLIMATE CHANGE AND IRRIGATED
AGRICULTRE IN SEMI-ARID CENTRAL EUROPE

30. Irrigated Agriculture in
Europe
 In parts of Southern Europe: Agriculture accounts for up to 80 % of total
water use (mostly crop irrigation)
 In Northern Europe, agriculture's contribution to total water use varies
from almost zero to over 30 % (mostly livestock farming) (EEA 2009).
 Proportion of area equipped for irrigation in selected countries in Central
Europe.
Country Country area Arable land Area equipped for irrigation: Proportion of area actually irrigated
(1000 ha) area total (1000ha) from area equipped for irrigation
(1000 ha) [proportion of arable area]
Austria 8387 1382 119 [~9%] 34%
CzechRepublic 7887 3032 47 [~2%] 37%
Hungary 9303 4592 153 [~3%] 49%
Slovakia 4903 1377 180 [~13%] 25%
Slovenia 2027 177 4 [~ 2%] 51%
cp. Trnka et al. 2010; AQUASTAT and FAOSTAT statistical databases (FAO 2009).
 Assumed fraction of irrigation methods in Europe: Basin and Furrow:
~34%, Drip ~18%,Sprinkler ~48% (Sauer et al. 2010).

31. Climate Change in Europe
 A warming trend (+0.90 C for 1901-2005) throughout Europe has
been observed, which has been accelerating in the last 30 years
(Alcamo et al. 2007)
 Regarding observed changes in precipitation (1961–2006):
Observed changes in annual
precipitation 1961–2006
Source: The data come from two
projects: ENSEMBLES
(http://www.ensembles-eu.org)
and ECA&D (http://eca.knmi.nl)
EEA, 2009

32. Climate Change in Europe
 Regional climate models project a larger warming in winter than
in summer in Northern Europe and the reverse in central and
Southern Europe (cp. Christensen and Christensen 2007).
 Trends in precipitation and changes in seasonal precipitation are
more variable spatially and temporally (IPCC 2007)
 For all scenarios mean annual precipitation increases in northern
Europe and decreases further south (IPCC 2007).

33. Climate Change in Europe
 Mediterranean regions, Central Europe and Eastern Europe :
o Precipitation trends are projected to be negative.
o Precipitation sums will decline in the early growing season
(April-June) (Trnka et al. 2010)
o Major and unprecedented drought events are more likely to
occur in the near future than at any time in the past 130 years
(Brazdil et al. 2009a,b; Trnka et al. 2010)
o A reduced groundwater recharge rate is predicted for Central and
Eastern Europe (Eitzinger et al. 2003; in IPCC 2007)
o For Central and Southern Europe, areas under water stress can
increase from 19% in 2007 to 35% in 2070 (IPCC, 2007).

34. Adaptation options in the Austrian Marchfeld
3.1. The region Marchfeld
3.2. Statistical climate data for Austria (EPIC input)
3.3. Investment in Irrigation Systems

35. Case Study – the region Marchfeld
 Marchfeld is part of the Marchfeld
Vienna Basin and
influenced by a semi-arid
climate
 Arable area: 65,000 ha.
 Area supporting irrigation: 60,000 ha
of which 30% are regularly irrigated (sprinkler irrigation).
 Cereals, root crops and vegetables comprise the main agricultural products
of the region.
 312 soil types can be differentiated in Marchfeld (Anonymous 1972).
 1975-2007: the average annual precipitation sum was 531 mm
 Vegetation period from April – September the average monthly
precipitation sum was only 331 mm

36. Statistical Climate Model for Austria
Strauss, F., Formayer, H., Asamer, V., Schmid, E., 2010. Climate change data for Austria and the period 2008-2040
with one day and km² resolution (No. Discussion Paper DP-48-2010). Institute for Sustainable Economic
Development, University of Natural Resources and Life Sciences, Vienna, Austria.
 Statistical climate model for Austria based on in situ weather observations from 1975-
2007 (Central Institute for Meteorology and Geodynamics)
 Based on linear regression and bootstrapping methods.
Precipitation Class Temperature [ C] Class
[mm]
100 bis <500 500 <0 0
>500 bis <600 600 >0 bis <2.5 1
>600 bis <700 700 >2.5 bis <4.5 3
>700 bis <800 800 >4.5 bis <5.5 5
>800 bis <900 900 >5.5 bis <6.5 6
>900 bis <1000 1000 >6.5 bis <7.5 7
>1000 bis <1250 1250 >7.5 bis <8.5 8
>1250 bis <1500 1500 >8.5 bis <9.5 9
>1500 2000 >9.5 bis <10.5 10
Statistical Climate Model Databse for Austria (average over 1961-1990
(Strauss et al. 2010; Auer et al., 2000) : StartClim (Schöner et al., 2003)

37. Statistical Climate Model for Austria
o Observed increase in temperature until 2007 -> Increase in annual
average temperature until year 2040.
o No trend in precipitation in period 1975-2007-> Assumption:
distribution of precipitation similar to past 30 years.
 To generate a climate spectrum: Stochastic weather scenarios for the
period 2008-2040 by bootstrapping of temperature residuals, observed
data for solar radiation, precipitation, relative humidity, wind: drawn
from observations of historical period
 Various precipitation scenarios for sensitivity analysis
o increasing/decreasing annual precipitation sums;
o unchanged annual precipitation sums with seasonal redistribution

38. Case studies:
For each year
For each year Realizations of
daily values: Crop yields for all
•Temperature management
•Precipitation options
Statistical •Radiation
Investment
Climate Model •Relative EPIC model
humidity
•Wind Profits
•Site specific
information

39. Investment in Irrigation Systems under weather
uncertainty
Together with S. Fuss (IIASA), J. Szolgayova (IIASA), Franziska Strauss
(BOKU), Erwin Schmid (BOKU)
Leading questions:
 Aim to model an farner’s decision to invest in a sprinkler or drip
irrigation system under precipitation uncertainty.
 How is the decision to invest affected by:
o Various soil types?
o Policy instruments
o Water prices
o Subsidies

40. Investment in Irrigation Systems –
Data and Methods
 Climate scenario with step-wise decreasing precipitation
(-20% in 2040 compared to base period 1975-2007)
 EPIC
o Carrots, Sugar Beet, Potato, Corn, Winter Wheat,
o Conventional tillage
o Drip and Sprinkler irrigation
o Automatically determined nitrogen fertilizer and irrigation amount.
o 2 soil types
 Profit calculation: Producer’s Information, Statistic Austria, Average
commodity prices 2005-2008

41. Investment in Irrigation Systems –
Data and Methods
 Characteristics of the model:
o Weather/climate uncertainty for period 2009-2040 (i.e. 300
precipitation scenarios)
o Agents choose the optimal investment strategy and time to maximize
expected sum of profits in planning period 2009-2040.
o Model is performed for each crop and soil type separately.
 Dynamic programming approach under weather uncertainty.
o Stochastic optimal control problem on a finite horizon with a discrete
stochastic component.
o The optimal actions are derived recursively by dynamic programming
using the Bellman equation

42. No Irrigation Sprinkler Drip
Soil 1 Soil 2 Soil 1 Soil 2 Soil1 Soil 2
Mean Mean Mean Mean Mean Mean
CORN Yields t/ha/a 6.2 4.6 7.9 7.5 7.9 7.5
CARROTS 5.4 3.5 5.5 5.2 5.5 5.3
POTATO 7.0 5.3 7.1 6.6 7.1 6.7
S.BEETS 7.8 6.2 10.1 9.8 10.3 10.0
W.WHEAT1 4.7 3.0 4.8 4.7 4.8 4.7
W.WHEAT2 4.9 3.0 5.1 4.9 5.1 5.0
CORN Irrig.mm/ha/a 127 254 113 229
CARROTS 39 148 34 132
POTATO 53 163 47 144
S.BEETS 162 280 143 262
W.WHEAT1 35 141 32 126
W.WHEAT2 36 142 32 126
CORN Profit €/ha/a 130 -82 10 -115 -249 -319
CARROTS 8323 4715.4 8352 7592 7910 7358
POTATO 2349 1234 2113 1750 1816 1532
S.BEETS 48 -215 60 -57 -166 -229
W.WHEAT1 460 107 204 120 -100 -132
W.WHEAT2 516 127 280 186 -22 -65

43. Investment in Irrigation Systems –
Results: Optimal Timing of Investment
Soil 1
(higher quality)
Soil 2

44. Investment in Irrigation Systems –
Policy Scenario 1 : Water Prices
SPRINKLER Sugar Winter Winter
IRRIGATION Corn Carrot Potatoes beets wheat 1 wheat 2
No policy - 2024 - 2024 - -
Soil 1
0.20 € - 2024 - - - -
(higher quality)
0.50 € - 2024 - - - -
1€ - 2024 - - - -
2€ - 2028 - - - -
SPRINKLER Sugar Winter Winter
Soil 2 IRRIGATION Corn Carrot Potatoes beets wheat 1 wheat 2
(lower quality) No policy - 2009 2009 2009 2023 2015
0.20 € - 2009 2009 2009 - -
0.50 € - 2009 2009 2023 - -
1€ - 2009 2009 - - -
2€ - 2009 2009 - - -

45. Investment in Irrigation Systems –
Policy Scenario 2 : Subsidies
Soil 1 (higher quality)
Corn Carrots Potatoes Sugar beets Winter w. 1 Winter w. 2
Drip Spri. Drip Spri. Drip Spri. Drip Spri. Drip Spri. Drip Spri.
No
policy - - - 2024 - - - 2024 - - - -
10% - - - 2024 - - - 2024 - - - -
30% - - - 2024 - - - 2024 - - - -
50% - - - 2024 - - - 2024 - - - -
60% - - - 2024 - - - 2024 - - - -
70% - - 2020 - - - - 2024 - - - -
80% - - 2011 - - - 2024 - - - - -
90% - - 2009 - - - 2009 - - - - -

46. Investment in Irrigation Systems –
Policy Scenario 2 : Subsidies
Soil 2(lowerquality)
Corn Carrots Potatoes Sugar beets Winter w. 1 Winter w. 2
Drip Spri. Drip Spri. Drip Spri. Drip Spri. Drip Spri. Drip Spri.
No
policy - - - 2009 - 2009 - 2009 - 2023 - 2015
10% - - - 2009 - 2009 - 2009 - 2023 - 2015
30% - - - 2009 - 2009 - 2009 - 2023 - 2015
50% - - - 2009 - 2009 - 2009 - 2023 - 2015
60% - - 2009 - - 2009 - 2009 - 2023 - 2015
70% - - 2009 - - 2009 2009 - - 2023 - 2015
80% - - 2009 - - 2009 2009 - 2022 - - 2015
201
90% 9 - 2009 - 2009 - 2009 - 2015 - 2009 -

47. Investment in Irrigation Systems –
Conclusion
 Though more water efficient, drip irrigation seems too expensive for adoption,
regardless whether crops are cultivated on soil 1 or 2.
 Water prices do not enforce adoption of drip irrigation but rather drive out
all irrigation systems.
 Subsidies on drip irrigation systems seem effective to support the adoption of
drip irrigation. However, to ensure full adoption of drip, subsidies of ~ 70-90%
of capital costs are needed, regardless of soil type.

48. CONCLUDING REMARKS

49. CONCLUDING REMARKS
 Climate change impact analyses require data and models (i.e.
biophysical and economic models) with sufficient reliability,
detail and resolution.
 Adaptation options need to be locally/regionally as well as
empirically assessed/evaluated => stakeholder participation
 Empirical model analysis yield powerful complementary
information about adaptation options, impacts and externalities
over space and time.

50. Thank you for your attention!
christine.heumesser@boku.ac.at
erwin.schmid@boku.ac.at