Rice. Wheat.
Rape. Sunflower.
0%
20%
40%
60%
80%
100%
0 5 10 15 20 25 30
140 304 469 633 798 962 1126 1291 1455 1620 1880
Frequency
Frequency Cumulative %
0%
20%
40%
60%
80%
100%
0 2 4 6 8 10 12 14 16 18
194 198 203 207 212 216 221 225 229 234 239
Frequency
Frequency Cumulative %
0%
20%
40%
60%
80%
100%
0 2 4 6 8 10 12 14 16 18
9 16 24 32 40 48 56 64 71 79 87
Frequency
Frequency Cumulative %
0%
20%
40%
60%
80%
100%
0 5 10 15 20 25 30 35
44 177 309 442 575 707 840 972 110512381370
Frequency
Frequency Cumulative %
Interdependencies in GLOBIOM induce systemic risks
Location-specific yield-shock distributions are multimodal, analytically intractable. Rare-low values representing e.g. droughts in Australia, cannot be captured by normal distribution (mean-variance), what precludes the use of mean-variance approaches, requires quantile-based systemic risks and security indicators, constraints and goals.
0%
20%
40%
60%
80%
100%
0 2 4 6 8 10 12
6603 7270 7603 7936 8270 8603 8936 9270 9603 9936 Yield
Frequency
Frequency Cumulative % average 8250
5% 6663 50% 8145 95% 9095
0%
20%
40%
60%
80%
100%
0 2 4 6 8 10 12
3827 4255 4469 4683 4897 5111 5325 5539 5753 5967 Yield
Frequency
Frequency Cumulative % average 5165
5% 4386 50% 5036 95% 5795
0%
20%
40%
60%
80%
100%
0 2 4 6 8 10
1471 1734 1865 1996 2128 2259 2390 2522 2653 2784 Yield
Frequency
Frequency Cumulative % average 2155
5% 1563 50% 2073 95% 2581
0%
20%
40%
60%
80%
100%
0 2 4 6 8 10 12 14 16
1111 1727 2035 2343 2651 2959 3267 3575 3883 4191 Yield
Frequency
Frequency Cumulative % average 2509
5% 1427 50% 2331 95% 2946 0%
20%
40%
60%
80%
100%
0 2 4 6 8 10 12
2572 2757 2850 2942 3035 3128 3220 3313 3406 3498 Yield
Frequency
Frequency Cumulative % average 3078
5% 2731 50% 3030 95% 3352
Droughts
Australia
Empirical yield distribution, wheat Normal distribution with same mean- Variance: where are droughts ?
??
0%
20%
40%
60%
80%
100%
0 2 4 6 8 10 12 14
2266 2841 3129 3417 3705 3992 4280 4568 4856 5144
Frequency
Frequency Cumulative %
Tatiana Ermolieva, Yuri Ermoliev, Petr Havlík, Aline Mosnier,
Michael Obersteiner, Dadid Leclere, Nikolay Khabarov, Hugo Valin, Wolf Reuter
Global changes, increasing interdependencies, vulnerability, and systemic risks in land use systems
Land use systems (LUS) resemble a complex network connected through demand–
supply relations such that the disruption of one—perhaps due to a yield shock in one region—may catalyze systemic risks affecting LUS worldwide and thus threaten food- energy-water-environmental (FEWE) security.
Systemic risks are characterized by the structure of the interdependencies in/between LUS, as well as by the distribution of risks (natural and shaped by decisions of intelligent agents), by targets and security constraints, e.g. emissions, biofuels, food requirements, water availability and quality, land pollution, etc.
Stochastic GLOBIOM and FEWE security management
Stochastic GLOBIOM is a stochastic partial equilibrium price-endogenous model; main land uses distinguish crop land, grass land, forest (managed and non-managed) land, fast- rotation forest plantations, and natural land.
Stochasticity is represented by random yield shock scenarios. The model can include stochastic costs, other threats.
Production from LUS has to cover respective demand in all scenarios: Food security constraint ensures that the energy intake from food cannot be lower than minimum amount of kilocalories needed to satisfy dietary requirements; Feeds produced for livestock cannot be lower than the minimum livestock dietary requirements; First-generation biofuels from crops and second-generation biofuels from lignocellulosic biomass (woody crops) and agricultural residues have to fulfill biofuel production targets, etc.
Food, feed, and biofuel security targets introduce competition for limited natural resources (land and water) among different land uses.
Global interdependencies between demand, prices, international trade flows, and environmental constraints are analyzed in an endogenous manner for 30 world regions (Havlík et al. 2011), while decisions on production and land use allocation are taken at a 50 x 50 km grid cell resolution.
Robust solutions make systems and regions better-off in all scenarios.
Integrated management of land use systems under systemic risks and food-energy-water-environmental security targets:
a stochastic Global Biosphere Management Model
Robust solutions: strategic and adaptive decisions
Robust storages
Stochastic yield shocks
Stochastic GLOBIOM is formulated as a two-stage stochastic optimization model deriving robust interdependent decisions: ex ante strategic decisions (production allocation, storage capacities) and ex post adaptive (demand, trading, storage control, prices) decisions.
Strategic decisions can be viewed as decisions in the face of uncertainties (before the exact state of nature is learned). Adaptive (operational) decisions are executed when additional information on uncertainties is revealed (after learning), allowing the policies to be adjusted. The model embeds quantile-based security constraints, which are central for ensuring security, managing systemic risks, maladaptation and irreversibility.
The model recommends strategic solutions: natural ecosystems should be preserved, the conversion of natural forests into managed should slow down, grass land should be protected as an important feed source for livestock, etc. Robust strategic solutions are supplemented with adaptive scenario-specific trade and storage decisions.
The model is applied to the case of increased storage facilities, which can be viewed as catastrophe pools to buffer production shortfalls and fulfill regional and global FEWES requirements when extreme events occur.
Expected shortfalls and storage capacities have a close relation with Value-at-Risk (VaR) and Conditional Value-at-Risk (CVaR) risk measures.
Calculated Value of Stochastic Solutions of about 25% demonstrates the importance of applying the stochastic model.
Distribution of storage withdrawals, in thousand tons, at the global level.
Distribution of storage withdrawals, in thd. tons, at the global level:
Crop land Grass land
Natural forest Managed forest
9.0 9.5 10.0 10.5 11.0 11.5 12.0 12.5
2010 2020 2030 2040 2050
Robust Average yield scenario 2000 yield shock scenario
17 18 19 20 21 22 23 24
2010 2020 2030 2040 2050
Robust Average yield scenario 2000 yield shock scenario
0 5 10 15 20 25 30 35 40
2010 2020 2030 2040 2050
Robust Average yield scenario 2000 yield shock scenario
0 2 4 6 8 10 12
2010 2020 2030 2040 2050
Robust Average yield scenario 2000 yield shock scenario
Acknowledgments: The research is supported by EU FP7 projects: IMPACT2C (Nr. 282746), ECONADAPT (Nr. 603906), AGRICISTRADE (No. 612755), and the the and project on the analysis of robust solutions for long-term consistent planning of secure food, energy, water provision, conducted jointly by International Institute for Applied Systems Analysis, Laxenburg, Austria, and National Academy of Sciences of Ukraine, IIASA-NASU project.
Publications: Ermolieva T, Ermoliev Y, Havlík P, Mosnier A, Leclere D, Khabarov N, Kraxner F, Obersteiner M (2015) Systems analysis of robust strategic decisions for planning secure food, energy, water provision based on stochastic GLOBIOM. Cybernetics and Systems Analysis, 1, V. 51, 125–133. doi: 10.1007/s10559-015-9704-2.; Ermolieva T, Ermoliev Y, Havlík P, Mosnier A, Leclere D, Khabarov N, Obersteiner M (1914). Systems analysis of strategic decisions for planning secure food, energy, water provision based on stochastic GLOBIOM. In: Integrated management, Security and Robustness (Eds.: Zagorodny, A.G., Ermoliev, Y.M., Bogdanov, V.L.), Published by Committee for Systems Analysis and Presidium of National Academy of Sciences, Ukraine – National Member
Organization of the International Institute for Applied Systems Analysis (IIASA). ISBN 978-966-02-7376-4, Kyiv, 2014 (p. 336), pp. 183-198.