Master’s Thesis, 60 ECTS
Social-ecological Resilience for Sustainable Development Master’s program 2016/17, 120 ECTS
The impacts of global warming and diet change on the virtual water consumption:
Estimating the present and future invisible virtual water resource use in East Africa
Julius Fröhlich
Supervisor: Ingo Fetzer Stockholm Resilience Center Co-Supervisor: Kate Brauman Universtiy of Minnesota
Stockholm Resilience Centre
Research for Biosphere Stewardship and Innovation
water consumption:
Estimating the present and future invisible virtual water resource use in East Africa
Credits: Julius Fröhlich
Julius Fröhlich Master’s Thesis, 60 ECTS
Social-ecological Resilience for Sustainable Development Master’s Program 2016/17, 120 ECTS
i
Table of Content
ABSTRACT ... 1
1. INTRODUCTION ... 2
2. BACKGROUND ... 3
2.1. IMPORTANCE OF RAIN WATER ... 3
2.2. WATER SCARCITY IN SUB-SAHARAN AFRICA ... 3
2.3. VIRTUAL WATER TRADE CONCEPT ... 3
2.4. VIRTUAL WATER CONTENT AND WATER FOOTPRINT ... 4
2.5. MEAT CONSUMPTION AND ITS RELATION TO ECONOMIC GROWTH ... 4
3. RESEARCH QUESTION ... 6
5. HYPOTHESIS ... 7
4. CASE STUDY ... 8
4.1. STUDY REGION ... 8
4.2. TIMEFRAME OF STUDY ... 10
4.3. PAST AND FUTURE SCENARIOS ... 11
4.4. DISTINCTION OF AGRICULTURAL COMMODITIES ... 11
5. THEORETICAL FRAMEWORK ... 13
6. METHOD ... 15
6.1. CLIMWAT-MODEL ... 17
6.2. CROPWAT-MODEL ... 17
6.3. PAST-SCENARIO CALCULATIONS ... 18
6.3.1. Virtual water in crops ... 18
6.3.2. Virtual water in livestock... 20
6.3.3. Virtual water in trade ... 23
6.3.4. Calculation of the national water footprint ... 24
6.3.5. Bennett’s law calculation ... 26
6.4. FUTURE-SCENARIO CALCULATIONS ... 26
7. RESULTS ... 30
7.1. PAST SCENARIO ... 30
7.1.1. Crops ... 30
7.1.2. Livestock ... 32
7.1.3. Trade influence ... 35
ii
7.1.4. Past water footprint ... 36
7.1.5. Green and blue water for crop production ... 37
7.1.6. Influence of VW by change of dietary habits ... 38
7.2. FUTURE SCENARIOS ... 39
7.2.1. Future water scarcity ratio ... 39
7.2.2. Evapotranspiration, green and blue water ... 39
7.2.3. Potential trade impact on the national water consumption ... 41
7.2.4. Change in consumption pattern ... 41
7.2.5. Impact of temperature on the water scarcity ratio ... 42
8. DISCUSSION ... 43
8.1. LIMITATIONS... 43
8.2. PAST-SCENARIO RESULTS ANALYSIS ... 45
8.3. FUTURE-SCENARIO RESULTS ANALYSIS ... 47
9. CONCLUSION ... 49
9.1. RECOMMENDATION ... 49
10. REFERENCE LIST ... 51
I. APPENDIX: TABLES FOR METHODS ... 57
II. APPENDIX: ADDITIONAL RESULTS... 65
III. APPENDIX: RESULTS BY COUNTRY ... 110
3.1. BURUNDI ... 110
3.2. DJIBOUTI ... 111
3.3. ERITREA ... 113
3.4. ETHIOPIA ... 114
3.5. KENYA ... 116
3.6. RWANDA ... 117
3.7. SOMALIA ... 119
3.8. UGANDA ... 120
3.9. TANZANIA ... 121
i List of Figures:
Figure 1: Map of study region in East Africa. Case study countries are highlighted dark grey. .... 8 Figure 2: a) Population development between 2005 and 2030 (FAO 2016a) and b) the Median GDP per capita (US $) between 2005 and 2015 (World Bank 2017) per African region. ... 9 Figure 3: Planetary boundaries in their state. Global freshwater use is currently in the safe operating space on a global scale (Rockström et al. 2009). ... 14 Figure 4: Flow chart on the process of calculating the water scarcity ratio (WSR). Using the data from the models from CLIMWAT and CROPWAT, the virtual water contents for crops and livestock is calculated and summed to get the water footprint (WF). The water footprint is put into relation to the water availability to get the water scarcity ratio (WSR). ... 15 Figure 5: Conceptual diagram of the calculation on the national water footprint. The method is divided into crop calculations, livestock calculations and trade calculations. Descriptions in square brackets represent the units they are calculated in, whereas round brackets describe the abbreviations of the term. This diagram represents the items involved in the calculation of the water footprint for one country for one single year. ... 16 Figure 6: a) average virtual water content of all crops in the Past-scenario with mean. b) total crop water productivity of all crop types produced in each country. ... 30 Figure 7: Weight of crops produced per country between 2005 - 2014 ... 31 Figure 8: a) average of virtual water contents for one of each animal type produced in each country.
b) total virtual water used for the entire livestock production per country. ... 33 Figure 9: Total weight of livestock produced per country between 2005 – 2014. These values include the trade balance of livestock products. ... 34 Figure 10: Balance of virtual water in trade from all commodities considered for each country. A positive value means that virtual water was imported, whereas a negative value means that virtual water was exported. ... 35 Figure 11: The water scarcity ratio per country between 2005 and 2014. ... 36 Figure 12: The relation between meat consumption (kg/capita) and GDP per capita (US$), also known as Bennett’s law, between 2005 – 2014 with the national mean and smoothened confidence interval. ... 38
ii Figure 13: Water scarcity ratio per country in the Future-scenario with an included confidence interval, which is based on the fluctuations of the Past-scenario and the temperature estimations.
... 39
Figure 14: Change in meat consumption in kg per capita per country in the Future-scenario. .... 41
Figure 15: Specific water demand (SWD) for Bananas in the past scenario. ... 71
Figure 16: Total virtual water content (VWC_Tc) for Bananas in the past scenario. ... 71
Figure 17: Specific water demand (SWD) for Barley in the past scenario. ... 72
Figure 18: Total virtual water content (VWC_Tc) for Barley in the past scenario. ... 72
Figure 19: Specific water demand (SWD) for Beans Dry in the past scenario. ... 73
Figure 20: Total virtual water content (VWC_Tc) for Beans dry in the past scenario. ... 73
Figure 21: Total virtual water content (VWC_Tc) for Beans green in the past scenario. ... 74
Figure 22: Specific water demand (SWD) for Beans green in the past scenario. ... 74
Figure 23: Specific water demand (SWD) for Cabbage in the past scenario. ... 75
Figure 24: Total virtual water content (VWC_Tc) for Cabbages in the past scenario. ... 75
Figure 25: Specific water demand (SWD) for Dates in the past scenario. ... 76
Figure 26: Total virtual water content (VWC_Tc) for Dates in the past scenario. ... 76
Figure 27: Specific water demand (SWD) for Maize in the past scenario. ... 77
Figure 28: Total virtual water content (VWC_Tc) for Maize in the past scenario. ... 77
Figure 29: Total virtual water content (VWC_Tc) for Mangoes in the past scenario. ... 78
Figure 30: Specific water demand (SWD) for Mangoes in the past scenario. ... 78
Figure 31: Specific water demand (SWD) for Millet in the past scenario. ... 79
Figure 32: Total virtual water content (VWC_Tc) for Millet in the past scenario. ... 79
Figure 33: Specific water demand (SWD) for Potatoes in the past scenario. ... 80
Figure 34: Total virtual water content (VWC_Tc) for Potatoes in the past scenario. ... 80
Figure 35: Specific water demand (SWD) for Pulses in the past scenario. ... 81
Figure 36: Total virtual water content (VWC_Tc) for Pulses in the past scenario. ... 81
Figure 37: Total virtual water content (VWC_Tc) for Rice in the past scenario. ... 82
Figure 38: Specific water demand (SWD) for Rice in the past scenario. ... 82
Figure 39: Specific water demand (SWD) for Sorghum in the past scenario. ... 83
Figure 40: Total virtual water content (VWC_Tc) for Sorghum in the past scenario. ... 83
Figure 41: Specific water demand (SWD) for Soybean in the past scenario... 84
iii
Figure 42: Total virtual water content (VWC_Tc) for Soybean in the past scenario. ... 84
Figure 43: Specific water demand (SWD) for Sugar cane in the past scenario. ... 85
Figure 44: Total virtual water content (VWC_Tc) for Sugarcane in the past scenario. ... 85
Figure 45: Total virtual water content (VWC_Tc) for Tomatoes in the past scenario. ... 86
Figure 46: Specific water demand (SWD) for Tomatoes in the past scenario... 86
Figure 47: Specific water demand (SWD) for Wheat in the past scenario. ... 87
Figure 48: Total virtual water content (VWC_Tc) for Wheat in the past scenario. ... 87
Figure 49: Virtual water content per Beef cow per country in the Past-scenario. ... 88
Figure 50: total virtual water content of Beef cows per country in the Past-scenario. ... 88
Figure 51: Virtual water content per Dairy cow per country in the Past-scenario. ... 89
Figure 52: total virtual water content of Dairy cows per country in the Past-scenario. ... 89
Figure 53: total virtual water content of Pigs per country in the Past-scenario. ... 90
Figure 54: Virtual water content per Pig per country in the Past-scenario. ... 90
Figure 55: Virtual water content per Sheep per country in the Past-scenario. ... 91
Figure 56: total virtual water content of Sheep per country in the Past-scenario. ... 91
Figure 57: total virtual water content of Goats per country in the Past-scenario. ... 92
Figure 58: Virtual water content per Goat per country in the Past-scenario. ... 92
Figure 59: Virtual water content per Broiler chicken per country in the Past-scenario. ... 93
Figure 60: total virtual water content of Broiler chicken per country in the Past-scenario. ... 93
Figure 61: Virtual water content per laying Hen per country in the Past-scenario. ... 94
Figure 62: total virtual water content of laying Hens per country in the Past-scenario. ... 94
Figure 63: Virtual water trade balance of all products belonging to the crop commodity per country in the Past-scenario. ... 95
Figure 64: Virtual water trade balance of all Eggs and Milk per country in the Past-scenario. ... 95
Figure 65: Virtual water trade balance of all live animals per country in the Past-scenario. ... 96
Figure 66: Virtual water trade balance of all meat products belonging to the livestock commodity per country in the Past-scenario. ... 96
Figure 67: Burundi’s a) Crop water productivity of 10 crop types and b) water footprint between 2005 and 2014. ... 110
Figure 68: GDP per capita (US$) and meat consumption (kg/capita) relation in Burundi in the Past- Scenario... 110
iv Figure 84: Water footprint (WF) in relation to the total water availability (TRWR) of Burundi in the Future-scenario... 111 Figure 69: Djibouti's a) Crop water productivity of 2 crop types and b) the water footprint between 2005 and 2014. ... 111 Figure 70: GDP per capita (US$) and meat consumption (kg/capita) relation in Djibouti in the Past-Scenario... 112 Figure 72: Water footprint (WF) in relation to the total water availability (TRWR) of Djibouti in the Future-scenario... 112 Figure 71: Eritrea's a) Crop water productivity of 8 crop types and b) the water footprint between 2005 and 2014. ... 113 Figure 72: GDP per capita (US$) and meat consumption (kg/capita) relation in Eritrea in the Past- Scenario... 113 Figure 86: Water footprint (WF) in relation to the total water availability (TRWR) of Eritrea in the Future-scenario ... 114 Figure 73: Ethiopia's a) Crop water productivity of 16 crop types and b) the water footprint between 2005 and 2014. ... 114 Figure 74: GDP per capita (US$) and meat consumption (kg/capita) relation in Ethiopia in the Past-scenario. ... 115 Figure 87: Water footprint (WF) in relation to the total water availability (TRWR) of Ethiopia in the Future-scenario... 115 Figure 75: Kenya's a) Crop water productivity of 17 crop types and b) the water footprint between 2005 and 2014. ... 116 Figure 76: GDP per capita (US$) and meat consumption (kg/capita) relation in Kenya in the Past- scenario. ... 116 Figure 88: Water footprint (WF) in relation to the total water availability (TRWR) of Kenya in the Future-scenario. ... 117 Figure 77: Rwanda's a) Crop water productivity of 14 crop types and b) the water footprint between 2005 and 2014. ... 117 Figure 78: GDP per capita (US$) and meat consumption (kg/capita) relation in Rwanda in the Past- scenario. ... 118
v Figure 89: Water footprint (WF) in relation to the total water availability (TRWR) of Rwanda in the Future-scenario... 118 Figure 79: Somalia's a) Crop water productivity of 9 crop types and b) the water footprint between 2005 and 2014. ... 119 Figure 90: Water footprint (WF) in relation to the total water availability (TRWR) of Somalia in the Future-scenario... 119 Figure 80: Uganda's a) Crop water productivity of 11 crop types and b) the water footprint between 2005 and 2014. ... 120 Figure 81: GDP per capita (US$) and meat consumption (kg/capita) relation in Uganda in the Past- scenario. ... 120 Figure 91: Water footprint (WF) in relation to the total water availability (TRWR) of Uganda in the Future-scenario... 121 Figure 82: Tanzania's a) Crop water productivity of 16 crop types and b) the water footprint between 2005 and 2014. ... 121 Figure 83: GDP per capita (US$) and meat consumption (kg/capita) relation in Tanzania in the Past-scenario. ... 122 Figure 92: Water footprint (WF) in relation to the total water availability (TRWR) of Tanzania in the Future-scenario... 122
vi List of Tables:
Table 1: List of production values per livestock type by Chapagain and Hoekstra (2003) *1: the animal types Bovine, Dairy, Sheep and Goat have a variable annual feed weight. *2: the animal types Bovine, Dairy, Pig and Hens have individual fourth primary products. For Bovine, it is Semen, Dairy cow, it is Milk, Pig, it is Bristles and Hens, it is Eggs. ... 20 Table 2: Prediction by FAO (2003) on yield and harvest change on the main crop types in SSA by 2030. The year 1999 is used as a reference point for future predictions. These values are used for all sub-scenarios in the Future-scenario. ... 27 Table 3: Estimation of annual increase in livestock and its products for SSA in 2030 (FAO 2003).
The year 1999 is used as a reference for future predictions. These values are used by all sub- scenarios of the Future-scenario. ... 27 Table 4: Annual mean temperature increase compared to 1990 per country in the Model-sub- scenario. Adapted from B1 scenario by Hulme et al. (2001). ... 28 Table 5: Future scenario temperature increase values compared to 1990, based on the best and worst case scenarios (RCP2.6 and RCP 8.5) from the 5th IPCC report (IPCC 2015)... 29 Table 6: National average of Evapotranspiration for crop production (ETc) and the composition of water use sources by green and blue water in the Past-scenario per country. *average values for Djibouti only consists out of two crop types (Beans Dry and Maize). ... 37 Table 7: National average green and blue water fraction and the average evapotraspiration (ETc) values of produced crop types per Future-scenario. *Djibouti only consists of 2 crop types (Beans Dry and Maize). ... 40 Table 8: Relation between ETc and temperature difference between the Past-scenario and the Model sub-scenario. ... 40 Table 9: assumed maximum percentage of total virtual water trade balance (VWT_T) per country in relation to the total renewable water resource availability (TRWR) in the Future-scenario. ... 41 Table 10: Percentage of produced total average weight and average virtual water content (VWC) per commodity type in the Modeled-sub-scenario... 42 Table 11: Difference of results in water scarcity ratio (WSR) and temperature (°C) between the two Future-sub-scenarios RCP8.5 and RCP 2.6 in 2030. ... 42 Table 12: Annual rainfall in the east African capital cities in mm. (Smith and FAO 1993) ... 57
vii Table 13: Geographical details of capitals, where climate values are taken from CLIMWAT 2.0.
(*: Akaki is a weather station nearby the capital city Addis Abeba in Ethiopia) ... 57 Table 14: CROPWAT settings for all crops (Smith and FAO 1992) ... 58 Table 15: Planting dates for each crop and each country. The crop types Dates, Mango and Pulses have no available information on planting dates (FAO 2010). 1: Burundi, 2: Eritrea, 3: Ethiopia,
4:Kenya, 5: Uganda, 6: Tanzania. ... 59 Table 16: TRWR values [m³/year] per country from the year 2014 (FAO 2016b). Values are based on a long-term average... 60 Table 17: List of irrigation efficiency per country (Frenken and Gillet 2012) ... 60 Table 18: Livestock production settings adapted from Chapagain and Hoekstra (2003). *: The diet for Swine is enhanced with Maize. ... 60 Table 19: Annual feed weight [ton/animal] for animal types pig, Broiler and Hens. Adapted from Chapagain and Hoekstra (2003)... 60 Table 20: Population (in 1000) from 2005 until 2030 based on the medium fertility variant scenario starting from 2015 on (United Nations Population Division (UNPD) 2015) ... 61 Table 21: Commodities by type and UN number extracted from UNdata traded to and from countries within the case study region between 2005 and 2016 (United Nations Statistics Division (UNSD) 2016). *Dairy cows are part of the Bovine livestock and separated from the beef cows later. ... 62 Table 22: List of crops with definition from the FAO considered in this study as relevant, based on the available information and data (FAO 2016a). ... 63 Table 23: Summary of results on the crop and livestock production between 2005 and 2014. Total specific water demand (SWD) represents the total amount of specific water demand for all crops together. ... 65 Table 24: Summary of crop production results between 2016 - 2030 for all future sub-scenarios.
... 67 Table 25: virtual water trade and water footprint results for the Past-scenario per country. VWT_T:
total virtual water trade balance. TRWR: total renewable water resources. ... 97 Table 26: Results of the livestock production converted into weight of meat in tons and the fraction per animal type in the Past-scenario. Values include livestock and meat trade when available. 100
viii Table 27: average SWD values in m³/ton between the results from Hung and Hoekstra (2002a) (original) and the Past-scenario (modeled). The original results refer to the year 1999, whereas the modeled results represent an average from 2005-2014. A negative difference indicates that the modeled results are higher. A positive difference indicates that the original values are higher.
Djibouti, Eritrea, Rwanda and Somalia do not have original data available. ... 103 Table 28: average VWC_Ta values in m³/ton between the results from Hung and Hoekstra (2002a) (original) and the Past-scenario (modeled). The original results refer to the year 1999, whereas the modeled results represent an average from 2005-2014. A negative difference indicates a higher modeled value, whereas a positive difference indicates a higher original value. ... 103 Table 29: Total meat production in tons from slaughtered animals per country and animal type in the Past-scenario. ... 104
ix List of Acronyms:
- BW blue water
- CWR crop water requirements
- GW green water
- SES Social Ecological System - SSA Sub-Saharan Africa - SWD specific water demand
- TRWR total renewable water resource - VW virtual water
- VWC virtual water content - VWT virtual water trade - WF water footprint - WUE water use efficiency
1
Abstract
The production of food is the biggest water consumer and it is required to meet the growing demand of food from a growing population and changing diets. Water can be dislocated indirectly in its hidden virtual form in traded food products as the water that is needed to produce the product locally. East Africa is a generally water scarce region, often ignored in scientific literature and characterized by a currently fast-growing population and overall economy, resulting in ongoing dietary change of people with growing incomes. The aim of this thesis is to estimate the current amount of water demand as virtual water to produce food for the present and future global change scenarios facing higher temperatures and higher food demand with growing populations. By using the CLIMWAT-model and CROPWAT-database, the virtual water content of all national produced agricultural commodities is summarized as its water footprint and put into relation to the national water availability. The results of the ratio between water footprint and availability show that Somalia’s, Kenya’s and Djibouti‘s food production water demand has exceeded the water availability before 2014. Tanzania and Ethiopia are expected to join these countries between 2014 and 2030. The impact of virtual water trade and global warming balance each other out with 2.23 and 2.45% of change of the water footprint and availability ratio. Meat production consumes twice as much water than crop production with only 5% of the total food weight, making livestock the key driver of virtual water development.
2
1. Introduction
Freshwater is an irreplaceable, finite but rechargeable resource required for socio-economic development, such as food production (Zaag and Savenije 2006). Global freshwater resources are expressing local signs of overuse, which is shown in depleting groundwater levels and rivers and lakes drying out (Postel 2000, Falkenmark 2006). Agriculture is the main consuming freshwater user in which the production of food plays a big role (Jägermeyr et al. 2016). A growing population and changing diets towards more water intensive products (e.g. livestock products) cause a growing demand of food (Postel 2000). Simultaneously it is required to reduce the environmental impact during the process of food production (Postel 2000, Foley et al. 2011).
The distribution of freshwater is uneven in location and time of the year because it depends on the geographic position in the water vapor flux (Falkenmark 1989, Postel 2000). Rain water in many arid regions is the only available source of freshwater, which leads to vulnerability towards droughts and climate changes (Rockström 1999). Limited availability and high demand of freshwater can cause water scarcity, which is seen as the main limiting factor to the development of agricultural production and consequently threatens regional food security (Postel 2000). The relocation of water from water-abundant to water-scarce regions is often associated with long distances and high costs (Hoekstra 2003). An alternative way to transport water between regions is by trading it in its “virtual” form; food. This so called hidden virtual water trade (VWT) occurs, as soon as a product enters the world market (de Fraiture et al., 2004). Food products are water in its virtual form. To produce food, the use of freshwater is required. This virtual water (VW) is herein defined as the amount of freshwater (in m³) that is required to produce an agricultural commodity (Allan 2003). VW thus links water, food and trade, which makes it a concept easy to understand and globally applicable (Allan 2003).
The region of East Africa, a geographic part of Sub-Saharan Africa (SSA), is a region of importance, as it is not only heavily dependent on rain water for the food production, but also a region prone to permanent water scarcity and unpredictable droughts (Rockström et al. 2004, African Development Bank Group 2010). In the future East Africa is expected to grow in its population size, economic power and its demand for livestock products all of which will increase the water consumption by a factor of 3.1 by 2050 (Falkenmark 2006, Dolislager and Tschirley 2014).
3
2. Background
2.1. Importance of rain water
The required demand for food in the future will double and have direct impacts on regional and global water availability (Gerten et al. 2013, Jägermeyr et al. 2016). Supplying the required amount of food is crucial to the global food security (Brauman et al. 2013). Rain fed agriculture dominates the cultivated land and the supply of food products (Rockström 1999). In rain fed agriculture, green water (GW) contributes 70% to the growth and productivity for crops, but in scientific literature often neglected, when estimating the national water resources since for these mostly contents of rivers and water reservoirs are taken into account (Rockström 1999, Gerten et al. 2011). GW is the amount of water contained in the unsaturated zones of the soil accessible to plants, whereas blue water (BW) is defined as the total sum of surface runoff and ground water recharge (Rockström 1999, Hoff et al. 2010).
2.2. Water scarcity in Sub-Saharan Africa
Water scarcity is defined as a situation where water is insufficiently available to satisfy all human water needs for food, feed, drinking and other uses is available (Falkenmark 2006). It increases risks of ecological degradation, limits industrial production, reduces human health and increases potential conflicts (Postel et al., 1996). A country meeting these water demands on the other hand is described as “water self-sufficient” (Allan 2003). Most African countries do not have self- sufficient water resources, hence they encounter challenges for the current and future food production (Falkenmark 1989). Problems in improving the agricultural output in SSA are linked to droughts, crop failures, decreasing recharge of aquifers, unpredictability of rainfall and the poor African soil characteristics (Falkenmark 1989).
2.3. Virtual water trade concept
VWT is of great importance when looking at a country’s water self-sufficiency and its water use abroad. International VWT is defined as the flow of agricultural products, which are converted into the importing country’s virtual water content (VWC) of the product type (Hoekstra and Hung 2002a). VWT is a more realizable, sustainable and environmentally friendly alternative in contrast
4 to real water trade (Hoekstra 2003). Water can be saved through trade, when an imported product can be produced more water efficiently than the importing country (de Fraiture et al., 2004).
Importing food reduces the pressure on national water resources and can reduce water scarcity (Hanasaki et al., 2010). Due to limited and inaccurate knowledge on VWT flows between countries and its great potential, there is a need to explore this concept in more detail (Hoekstra and Hung 2002a, de Fraiture et al. 2004).
2.4. Virtual water content and water footprint
The value of a VWC depends on the location, seasonal period and the water use efficiency (WUE) during production of the crop type (Hoekstra 2003). WUE is defined as the amount of water evapotranspirated from the crop (ETc) required per yield of product [m³/ton] (Rockström 1999).
The WUE value depends additionally on multiple factors one of which being the daily temperature (in °C). The VWC of a product shows the overall consumption of water and provides a measure for comparison and awareness to save water (Hoekstra 2003). It is also the basis for the calculation of a country’s water footprint (WF).
The WF expresses the water needed to produce the food consumed by the people living in a country, including the external water sources (Hoekstra and Chapagain 2006). This value varies in every country and depends on the volume and pattern of the consumption, the climate and the WUE (Hoekstra and Chapagain 2006). The WF is a strong tool to show a nation’s impact on water resources (Hoekstra 2003). It usually includes also industrial and domestic water uses, which are excluded from the definition in this study.
2.5. Meat consumption and its relation to economic growth
Why the WF varies between countries can also be related to the diet of the population. The concept of diet change is closely related to Bennett’s law, a rule which describes the positive relation between the demand for livestock products per capita with the rise in economic purchasing power (Dolislager and Tschirley 2014). This economic factor in Bennett’s law is herein represented with the value of GDP per capita, which is a useful measurement when comparing countries as it is also used in Dalin et al. (2012). The concept of diet change towards more meat consumption is used
5 throughout the thesis to explain the expected increase in livestock production in relation to the predicted change in GDP per capita.
6
3. Research question
This study examines of the national total virtual water consumption of agricultural commodities produced in the East African region between 2005 and 2030. The following research questions are separated between the Past-scenario, covering the years 2005 – 2014, and the Future-scenarios, covering the years 2016 - 2030.
1. Past-scenario
- What is the current virtual water content for all crops and livestock per case study country?
- What is the virtual water trade balance of the individual nations and how much does it affect the water footprint?
- What is the water footprint in relation to the water availability for each country?
- How much is the contribution of green water to the production of crops in each country?
- In which countries can Bennett’s law be confirmed to be true?
2. Future-scenarios
- What will be the water footprint in relation to the water availability for each country in the Future-scenarios?
- How does the relation between green and blue water in the crop production change?
- Based on the Past-scenario, what is the potential influence of virtual water trade on the national water footprint?
- How does the meat consumption per capita change in the Future-scenario?
- What is the major driving force in an increase of a country’s water footprint?
7
5. Hypothesis
The VWC of the agricultural commodities produced in the case study region are expected to have a higher value of required water in the Future-scenario compared to the Past-scenario, as the average annual temperature values are higher and might be less favorable for the crop growth. The result could be a higher water demand, which eventually has an adverse effect on the WF of especially those countries with a low water availability and a high temperature increase.
It is expected that the GW percentage plays a dominant role in the water use for crop production, as it is the main water source for most SSA-countries.
Provision of food from the East African countries to the world food market does not play a major role. The accessibility of the world food market is, however, expected to be of major significance to the food production in these countries. It is expected that countries, which are warmer and drier than others, like Eritrea, Djibouti and Somalia, import more food but experience a simultaneous increase in water use for crops production. On the other hand, countries like Ethiopia, Kenya or Tanzania are assumed to export more of their agricultural commodities, which would have negative influence on their WF. The effect of the two forces, trade and temperature, on the WF may balance each other out. The tendency to exceed the water availability, however, is more likely than a trend towards a lower WF.
8
4. Case study
4.1. Study Region
The majority of scientific studies on VW focuses on the Middle-East and North African region with the reason that their water resources are far more stressed and scarce, their economic stability, however, equalizes the deficit by food imports (Hoekstra and Hung 2002b). This study draws the attention to the region of East Africa, which is defined and encompassed as the nine countries Eritrea, Djibouti, Ethiopia, Somalia, Kenya, Uganda, Tanzania, Rwanda and Burundi (Figure 1).
Figure 1: Map of study region in East Africa. Case study countries are highlighted dark grey.
9
Figure 2: a) Population development between 2005 and 2030 (FAO 2016a) and b) the Median GDP per capita (US $) between 2005 and 2015 (World Bank 2017) per African region.
10 The important aspect of East Africa compared to other African regions is the highest population growth and size, but with a low GDP (Figure 2) (FAO 2016a, World Bank 2017). East Africa is a developing region, which is characterized by low annual rainfall (see Appendix: Table 12), poor water infrastructure, high importance of agriculture to the regional GDP and high nomadic pastoralism (African Development Bank Group 2010). As for its current livestock production, the region suffers from limited pasture availability and quality (African Development Bank Group 2010).
The vulnerability towards water scarcity in this region is high. In a study by Yang et al. (2003) the threshold of renewable freshwater available for human use per capita was estimated to be at 1500 m³/year. Three of the case study countries (Burundi, Kenya, Rwanda) are expected to drop below this threshold by 2030; five others (Eritrea, Ethiopia, Somalia, Tanzania and Uganda) are considered to have crossed this line already in 2000 (Yang et al. 2003). Like this threshold, water estimations on water required per capita usually exclude GW, which is a significant amount of freshwater required to produce food. Including the amount of water used in the production of food consumed increases the threshold. This additional value can be as much as 70 times the amount a person uses for domestic purposes (Falkenmark 2006).
4.2. Timeframe of study
This case study analyses the national VW consumption in the period between 2005 and 2030 and is divided into two parts. The first part covers the years 2005 until 2014 and provides a historical reference and a source for the confidence interval for the following years. It gives information about the prior development in the country’s WF trend. The second part dates from 2016 until 2030 and attempts to predict the WF of each country for these 15 years. Year 2015 is skipped in the timeframe, as the available data on agricultural commodities are incomplete and the information used for the estimations starts in 2016. Due to restricted availability and inaccuracy of relevant data, longer term predictions were not considered feasible.
11 4.3. Past and future scenarios
The climate and rainfall values for this scenario are based on the measured climate information from the data base CLIMWAT 2.0 by the FAO (http://www.fao.org/land-water/databases-and- software/climwat-for-cropwat/en/) (Smith and FAO 1993). Crop type details, such as rooting depth or water uptake per crop growth stage is given by the software program CROPWAT 8.0 of the FAO (http://www.fao.org/land-water/databases-and-software/cropwat/en/) (Smith and FAO 1992). Crop and livestock data is taken from the FAO database (FAO 2016b). The trade data for all countries, except Somalia and Eritrea, is obtained from the United Nations database (United Nations Statistics Division (UNSD) 2016). Data on population and economic values come from the World Bank (World Bank 2017). This scenario is named “Past” and acts as a reference and is used to create the confidence interval for the second part. It covers the time frame of 2005 – 2014.
The second part of the study continues with the WF analysis for the years 2016 -2030. The initial crop and livestock data is built upon the last available data from the Past-scenario and the change in production based on estimations taken from FAO (2003). This so called “Future-scenario”
consist of three sub-scenarios with varying temperature values, which correlate to the global warming scenarios. Those sub-scenarios with extreme temperature values are used to outline the possible development of WF related to estimated future global warming situations. These temperature data are named according to the IPCC scenarios, from where they were obtained from (IPCC 2015). “RCP 2.6” represents a radiative forcing of 3 W/m² before 2100 and then decline. It is also commonly known as the best-case scenario and will serve as the lower boundary conditions in this study. The worst-case scenario, or “RCP 8.5”, stands for a rising of 8.5 W/m² by 2100 and acts as the upper boundary conditions. The third sub-scenario, “Model”-scenario, contains temperature information from Hulme et al. (2001). The Model-sub-scenario presents a temperature condition which lies between the two extremes and represents a mean value.
4.4. Distinction of agricultural commodities
In this study, agricultural commodities are separated into livestock and crops. Crop types, Livestock and their primary products are defined in the Appendix: Table 22. The seven livestock types used in this study are grouped into two sub-categories. “Grazers” represent the animal types
12 beef and dairy cows, goat and sheep. “Non-grazers” consist of the remaining animal types, pig, broiler chicken and laying hens.
Crops are herein defined as all plant based product used for human or animal consumption (see Appendix: Table 22). A crop type is included in the study when data on the area harvested, the yield and the production between 2005 and 2014 is given and details are available CROPWAT.
13
5. Theoretical framework
Natural resources, such as water, are in many regions of the globe exploited beyond sustainable use (Postel 2000, Foley et al. 2011). This indicates a lack of understanding of the relation between the role of ecosystem services, such as food and the economic systems (Allison and Hobbs 2004).
The interaction between these interdependent systems is defined as a social ecological system (SES) (Folke et al., 2010). The ability of a SES to absorb disturbance, such as a drought, regain its original structure, function and feedbacks is called the system’s resilience and helps to understand the capacity of an ecosystem to persist perturbations (Folke et al. 2010). The moment when a SES’s resilience is exceeded and the system changes into a new and often unwanted state, is defined as the resilience threshold (Allison and Hobbs 2004). The definition of an accurate resilience threshold in a SES is the most important aim of a resilience analysis, as it helps to define boundaries and prevent human actions to make the system drift away from its current state where it provides essential ecosystem services.
Closely related to this is a framework which focuses on a global scale to address the safe human operating space within global biophysical boundaries (Rockström et al. 2009). The concept of planetary boundaries describes how strong essential global process can be stressed before the overall global systems’ resilience is weakened and threatens to change into an unsafe state for humanity on a worldwide scale (Gerten et al., 2013; Häyhä et al., 2016). In the planetary boundary framework, which acknowledges freshwater usage by humans as an essential boundary, the threshold is difficult to determine because of the complexity in the water flow allocation, the different estimation methods and the neglected but crucial difficulty to estimate the actual quantity of GW (Gerten et al., 2013).
As stated earlier, available water capacity for human and environmental use depends on the geographic location. The planetary boundary category for freshwater use is estimated to be located within the “safe”-zone (Figure 3). Comparing this estimation with the other categories from the planetary boundaries, the perception perceived from the current state of global freshwater use hides the disproportionate distribution of freshwater among regions, where water is scarce or inaccessible.
14
Figure 3: Planetary boundaries in their state. Global freshwater use is currently in the safe operating space on a global scale (Rockström et al. 2009).
The calculation of virtual water of crops can be done using a variety of biophysical models, which calculate the processes and evapotranspiration values. A thorough analysis of such models was made by Hoff et al. (2010), comparing seven global hydrological models (H08, GEPIC, LPJmL, GCWM, WB; plus, WaterGap2, IMPACT) in its vital features. The major difference in these models was the calculation method for the potential evapotranspiration values. The Penman- Monteith calculation method was chosen in this study to be the preferred way to calculate the potential evapotranspiration, as it is a reliable, for a broad range of climatic regions applicable method and often used as a reference for other calculation methods (Nadrah et al. 2012). Th Penman-Monteith method is not only used by the GCWM model, but also by the CROPWAT model by the FAO (Smith and FAO 1992).
Literature often mentions the food gap, which is the missing approx. 7.5*1015 kcal per year required to feed the undernourished population on a global scale in 2050, of which most will be living in SSA (Jägermeyr et al. 2016). These missing nutrients are not a matter of too little food production, but rather the inaccessibility to the required food sources (FAO 2003).
15
6. Method
This study builds on the methods of the calculation of VWC for livestock and crop production by Chapagain and Hoekstra (2003) and Hoekstra and Hung (2005) and applies a few adaptations to the case study area and the calculation methods.
Other than by Hoekstra and Hung (2005), where 37 crop types and 5 years were used, this study considers 18 out of 51 crop types and the time frame contains 25 years. In the livestock calculations by Chapagain and Hoekstra (2003), the water content for forage is set at the global average of 445 m³/ton. This value is replaced with specific water content values for each individual country. The feed composition for grazing animal types are replaced with a diet composition closer related to one used in developing countries. The water used for mixing the feed for grazing animals, as it is used in the original literature, is removed. In Chapagain and Hoekstra (2003), the earliest growth stage for all animals do not receive numerical values on their feeding content. The virtual water content of feed per animal is thus calculated by dividing the virtual water content in feed in the adult stage by the amount of growth stages to get the average VWC for all growth stages. In this study, the virtual water content in feed is calculated for all growth stages.
Figure 4: Flow chart on the process of calculating the water scarcity ratio (WSR). Using the data from the models from CLIMWAT and CROPWAT, the virtual water contents for crops and livestock is calculated and summed to get the water footprint (WF). The water footprint is put into relation to the water availability to get the water scarcity ratio (WSR).
16 The calculation of a nation’s WF is done by using the local climate information from CLIMWAT- database and the soil and crop characteristics from CROPWAT-model to calculate the VWC for each crop type. These values are then used to calculate the VWC of livestock and its primary products. Additionally, all traded agricultural products are converted into their VWC and used to calculate the net impact of trade. This VW trade balance is then added to the sum VWC of all crops and livestock, which gives the WF. This value is the put into relation to the water availability to get the water scarcity ratio (WSR) (Figure 4).
The complete presentation of factors involved in the calculation for the virtual water content of crops, livestock, trade and the WF are displayed in more detail in Figure 5.
The method of calculating the total VW consumption of one country, differs slightly between the Past- and the Future-scenarios. The Past-scenario is calculated first. The three sub-scenarios between 2016 and 2030 diverge in temperature information from each other and the Past-scenario, due to the estimations and assumptions made for this time frame.
Figure 5: Conceptual diagram of the calculation on the national water footprint. The method is divided into crop calculations, livestock calculations and trade calculations. Descriptions in square brackets represent the units they are calculated in, whereas round brackets describe the abbreviations of the term. This diagram represents the items involved in the calculation of the water footprint for one country for one single year.
17 6.1.CLIMWAT-model
The first step is to extract climate data for the required time frame, which sets the basic production conditions for all crops. The FAO software CLIMWAT extracts monthly surface temperature (minimal and maximal), humidity [%], wind speed [km/day], sunlight [hours], radiation [MJ/m²/day], rainfall [mm/month], effective rainfall [mm/year] and the reference evapotranspiration (ET0) [mm/day] (Smith and FAO 1993). These values are observations and computations from meteorological stations in the database of the Agromet Group of the FAO. Like in Hoekstra and Hung (2002), the climate data of the capital city is taken as the reference for each country during the entire study (see in Appendix Table 13). These climate values are not representable values for the entire country. In order to compare the results with the values in Hoekstra and Hung, Hoekstra and Chapagain (2002, 2003), this method is used as well.
6.2.CROPWAT-model
The data from CLIMWAT are inserted into the crop calculation model CROPWAT 8.0 (Smith and FAO 1992). The crop calculations in CROPWAT are done according to Savva and Frenken (2002) (see Appendix: Table 14). Combining climate, rainfall, crop and soil values, the program calculates the reference Evapotranspiration (ET0), Evapotranspiration under standard conditions (ETc) and the irrigation requirements by using the FAO version of the Penman-Monteith equation (Allen et al., 2006) [1].
[1]
𝐸𝑇0 =
0.408∆(𝑅𝑛− 𝐺) + 𝛾 ( 900
(𝑇 + 273)) 𝑢2(𝑒𝑠− 𝑒𝑎)
∆ + 𝛾(1 + 0.34𝑢2) where
ET0 reference evapotranspiration [mm/day], Rn net radiation at the crop surface [MJ/m2/day], G soil heat flux density [MJ/m2/day],
T mean daily air temperature at 2 m height [°C],
18 u2 wind speed at 2 m height [m/s],
es saturation vapor pressure [kPa], ea actual vapor pressure [kPa],
es-ea saturation vapor pressure deficit [kPa],
∆ slope vapor pressure curve [kPa/°C], γ psychrometric constant [kPa/°C].
6.3.Past-scenario calculations
6.3.1. Virtual water in crops
The first step is to calculate all crops, of which some serve as feed for the livestock production. 18 crop types were used in this study, including the five important ones: Wheat, Sorghum Sugarcane and Maize. The crop data contained values of yield, production and area harvested. The crop calculations method is adapted from Hoekstra & Hung (2002).
Crops are produced under optimal conditions, which implies that all planted crops are also harvested. Each crop plant type has individual planting and harvesting dates which were taken from the FAO’s website on planting schedule (2010) (Table 15). To calculate the VWC_c, the ETc
value from CROPWAT is converted into the crop water requirement (CWR) value and divided by the crop yield. Multiplied by the total production [ton] it gives the virtual water content of that crop type produced in that country and year [2].
[2]
𝑉𝑊𝐶_𝑐 = (𝐸𝑇𝑐 ∗ 𝑐 𝐶𝑌 ) ∗ 𝑃 Where
VWC_c virtual water content of crop type [m³/year]
ETc evapotranspiration of crop type [mm/year]
CY crop yield [ton/ha]
19
P production [ton]
c conversion coefficient [10 m³/mm*ha]
Next to the ETc value, CROPWAT also gives information about the amount of rainfall and irrigation water used for the production. These values depend on the planting schedule and the rainfall for that geographic location. In this study, it is assumed that irrigation water is available in all countries because the conditions for production are assumed to be optimal. This means that the results are overestimates in terms of the water use, since these optimal conditions do not represent reality (Hoekstra and Hung 2002b). On the other hand, this approximation might be equalized as this study does not encounter post-harvest loss. The loss of food after it has been harvested can make up more than 40% in developing countries, due to processing, transport and storage conditions (Foley et al. 2011).
The water used from rainfall is called effective rain (mm/month) and represents the water from rain, which becomes available for crop growth. The sum of water used from rainfall and irrigation rarely matches the ETc value. This is because the effective rain available in one month exceeds the ETc value, whereas irrigation water is only used to replace the missing rain water amount.
Values of the individual virtual water content (VWC_c) all crop types in one year are summed up to calculate the total virtual water content of all crops of one country in one year (VWC_Tc).
A special role plays forage or pasture, which is included in the crop commodity category. Pasture production plays a special role in the calculation for the virtual water content in livestock.
Information on the countries’ individual pasture yield and production is unknown. The specific water demand (SWD) of a crop type, gives the amount of water required to produce one ton of the crop type in a specific year and location. The SWD for pasture is calculated for each country by calculating the average crop yield using the “Grassland-Index” (Andersson et al. 1990). With the crop yield, the VWC_c of pasture for each country is calculated as done in calculation formula [2].
According to the extensive grazing pasture from Savva and Frenken (2002) the Kc values were changed to 0.3,0.75 and 0.75 and a crop height of 0.1 meter in the CROPWAT calculation, as this is considered closer related to the average forage growth.
20 6.3.2. Virtual water in livestock
The data taken from FAO on livestock production consists of information on total animal number [head], animals slaughtered [head], the yield per carcass weight [hg/animal or Mg/animal] and production [tons] (FAO 2016a). This study considers the major seven animal types and their primary products (Table 1).
Table 1: List of production values per livestock type by Chapagain and Hoekstra (2003)
*1: the animal types Bovine, Dairy, Sheep and Goat have a variable annual feed weight.
*2: the animal types Bovine, Dairy, Pig and Hens have individual fourth primary products. For Bovine, it is Semen, Dairy cow, it is Milk, Pig, it is Bristles and Hens, it is Eggs.
Animal type
Lifetime (years)
Service + drinking water (m³)
Weight of feed (ton/year) *1
Product fraction carcass
Product fraction offal
Product fraction skin
Product fraction other *2
Bovine 3 17.52 0.52 0.07 0.06 ~0.00
Dairy cow 7 71.54 0.52 0.07 0.06 0.35
Pig 1 6.35 0.72 0.76 0.04 0.02 0.01
Sheep 2 4.12 0.53 0.08 0.08
Goat 2.5 4.02 0.50 0.08
Broiler 0.3 0.02 0.03 0.78
Hens 1.52 0.37 0.03 0.78 0.22
Two animal types need to be distinguished from their general description to calculate their VWC more accurately. The number of broiler chicken is not given in the database. Therefore, the national number of poultry animals per country is reduced by the known number of laying hens. The outcome is the assumed number of broiler chicken per country per year. Similar approach is done by determining the exact number of bovine cows per country, not used for milk production.
All livestock are born, raised and slaughtered in the same country. The livestock in the case study area is produced under the grazing livestock system, which is the dominates the industrial production system (Chapagain and Hoekstra 2003a, Hoekstra 2003, African Development Bank Group 2010). The demand for drinking and service water for each livestock type is set according to the animal’s global average (Chapagain and Hoekstra 2003a). All animals are assumed to be raised under optimal conditions (no disease, food shortage, etc.) and all the required feed and
21 pasture consumed by the animals is assumed to be produced locally. All animals are expected to consume 100% of the required feed. Further production settings for each animal type are taken from Chapagain and Hoekstra (2003b) (see Appendix: Table 18).
Non-grazing livestock, which includes pigs, broiler chicken and laying hens, are considered not to be fed exclusively from or use pasture but from a specific mixture of specific crops crop products (Chapagain and Hoekstra 2003a). Additionally, half of the feed’s weight of the feed is added, which represents the water used to mix the feed [3]. The total virtual water requirement for livestock is calculated as following:
[3]
𝑉𝑊𝐶_𝑓1 = [(∑ 𝐹𝑄𝑖∗ 𝑆𝑊𝐷𝑖
𝑛
𝑖=1
= (𝐹𝑄1∗ 𝑆𝑊𝐷1) + (𝐹𝑄2∗ 𝑆𝑊𝐷2) … (𝐹𝑄𝑛∗ 𝑆𝑊𝐷𝑛))
∗ 𝑡𝑜𝑡𝑎𝑙 𝑎𝑔𝑒 (𝑦𝑒𝑎𝑟𝑠)] + [(∑ 𝑊𝑓 = 𝑊𝑓1+ 𝑊𝑓2… 𝑊𝑓𝑛
𝑛
𝑖=1
) ∗ 𝑐𝑓]
Where
VWC_f1 total virtual water content feed [m³/year]
FQ feed requirement [ton/year]
SWD specific water demand of individual feed [m³/ton]
WF weight of individual feed type [ton]
cf conversion coefficient [0.5 m³/ton *year]
All animals belonging to the grazers are fed exclusively by pasture, which differs from the feeding method used for non-grazers. Herein they are collectively called as pasture-animals. The VWC of pasture-fed animals is calculated by using the estimation by Ayuko et al. (1976), whereby a grazing animal’s daily pasture consumption in dry matter equals to 2.5% of its entire bodyweight. The formula for calculating the VWC for grazers is the following [4]:
[4]
22 𝑉𝑊𝐶_𝑓2 = ((𝑊𝑎 ∗ 0.025) ∗ 365) ∗ 𝑆𝑊𝐷
Where
VWC_f2 virtual water content of feed [m³/year]
Wa average animal weight [ton]
SWD specific water demand of pasture [m³/ton]
The feed itself, however does not make up the entire VWC of an animal. The water it drinks [5]
and needs for service [6] is also required to be included in the VWC of the animal.
[5]
𝐷𝑊 = 𝐶𝑑∗ 𝑎 ∗ 𝑐𝑠 DW water used for drinking [m³/year]
Cd daily consumption [L/day]
cs conversion coefficient [1000 m³/L]
a age [days]
[6]
𝑆𝑊 = 𝐶𝑑∗ 𝑎 ∗ 𝑐𝑠 SW water used for service [m³/year]
Cd daily consumption [L/day]
cs conversion coefficient [1000 m³/L]
a age [days]
23 The total use of water for one animal (VWC_a) is calculated by adding all water requirements together [6].
[7]
𝑉𝑊𝐶_𝑎 = 𝑉𝑊𝐶_𝑓 + 𝐷𝑊 + 𝑆𝑊
The sum of all animals produced in one country in one year is described as the total animal virtual water content (VWC_Ta).
6.3.3. Virtual water in trade
The SWD of traded crop types into or out of a country is sometimes not given. In such a case, the SWD of a neighboring country, where the value of that specific crop type is available, is used instead. This adjustment assumes that due to the geographic proximity the production conditions are similar, as it is similarly done in Hoekstra and Hung (2002a). All other crops, where no SWD can be calculated, are excluded from VWT. Rice in trade is separated into four kinds, which are merged in this study and considered to have the same SWD value.
The VWC_a is applied for every livestock animal traded into or out of the related country. Looking at trade commodities, it is however not always the entire animal being traded between two markets, but rather their primary products. Primary products are defined as products obtained from an animal in its unprocessed form. These are for example an animal’s meat, intestines, milk or eggs.
Each primary product has an individual VWC value (VWC_p), which will help to give a better evaluation of traded products. Every VWC_p is calculated per country, which imports/exports the primary product. For this calculation, the VWC_a is distributed among the primary products per product fraction (pfp) and divided by the animal’s average weight [7].
[7]
𝑉𝑊𝐶_𝑝 = 𝑝𝑓𝑝∗ 𝑉𝑊𝐶_𝑎 Where
VWC_p virtual water content of the primary product [m³/animal]
pfp product fraction of primary product
24 VWC_a total virtual water content of animal [m³/animal]
Milk in trade has many different types but only the unconcentrated and unsweetened values are considered in this study (<1% and 1-6%). Both types are considered to have the same VWC_p. Re- imported/exported goods are excluded from the study.
All traded commodities are converted into the VWC value of the country being imported/exported to. For each year and country, the exported virtual water content (VWT_e) [9] is subtracted from the imported content (VWT_i) [8] to gain the total trade balance (VWT_T) [10].
[8]
𝑉𝑊𝑇_𝑖 = ∫ 𝑉𝑊𝑇_𝑖
𝑛
𝑖=1
= 𝑉𝑊𝑇1+ 𝑉𝑊𝑇2… 𝑉𝑊𝑇𝑛
[9]
𝑉𝑊𝑇_𝑒 = ∫ 𝑉𝑊𝑇_𝑒 =
𝑛
𝑖=1
𝑉𝑊𝑇1+ 𝑉𝑊𝑇2… 𝑉𝑊𝑇𝑛
[10]
𝑉𝑊𝑇_𝑇 = 𝑉𝑊𝑇_𝑖 − 𝑉𝑊𝑇_𝑒 Where
VWT_i imported virtual water [m³/year]
VWT_e exported virtual water [m³/year]
VWT_T total virtual water trade balance [m³/year]
6.3.4. Calculation of the national water footprint
Before the crop VWC_Tc, the livestock VWC_Ta and the VWT_T are summarized to gain the nation’s WF, another calculation needs to be made to exclude the feed consumed by pigs and poultry from being counted twice. For this purpose, the VWC_f1 by both animal types is summarized into the virtual water double counting (VWC_dc) value [11].
25 [11]
𝑉𝑊𝐶_𝑑𝑐 = ∫ 𝑉𝑊𝐶_𝑓1𝑖 = 𝑉𝑊𝐶_𝑓11+ 𝑉𝑊𝐶_𝑓12… 𝑉𝑊𝐶_𝑓1𝑛
𝑛 𝑖=1
Where
VWC_dc virtual water in double counting [m³/year]
The final formula calculates the total value of water, that is needed to produce the agricultural commodities required by the country in one year.
[12]
𝑊𝐹 = (𝑉𝑊𝐶_𝑇𝑎 + 𝑉𝑊𝐶_𝑇𝑐 + 𝑉𝑊𝑇_𝑇 − 𝑉𝑊_𝑑𝑐) Where
WF water footprint [m³/year]
To get a better understanding of the significance of water use in relation to each countries water availability, the WF is brought into relation to the country’s total renewable water resource (TRWR), which is taken from Aquastat (2016b). The TRWR value includes internal water sources from precipitation (km³/year), surface- and groundwater, as well as external water sources from neighboring countries (see Appendix: Table 16). This relation is described as water scarcity ratio (WSR) and describes the relation between water use and water availability (Hoekstra and Hung 2002a) [13].
[13]
𝑊𝑆𝑅 = ( 𝑊𝐹 𝑇𝑅𝑊𝑅) Where
WSR water scarcity ratio
TRWR total renewable water resource [km³/year]
26 When the WF does not exceed the water availability, the country has enough water to satisfy its water demand and the WSR is below 1. When the WSR is greater than 1, the country uses more water for its food production than it can provide with its national renewable water resources.
6.3.5. Bennett’s law calculation
The positive relation between meat consumption and the economic purchasing power is described as Bennett’s law. This relation is measured by linking the meat consumption per capita with the GDP in US$ per capita over the same period. The meat consumption patterns on a national scale are unknown. Instead, we use the modeled national meat production in tons to represent the national consumption. From the meat production exported and imported meat products are subtracted/added. The remaining meat production value is divided by the national population to give the average meat consumption per capita (see Appendix: Table 20). The meat consumption per capita is put into relation to the GDP per capita, which represents the economic purchasing power [14]. This ratio is then described as the diet and economic relation (DER).
[14]
𝐷𝐸𝑅 = 𝑀𝐶 𝐸𝑃𝑃 Where
DER diet and economic relation
MC meat consumption (kg/capita)
EPP economic purchasing power (US$/capita)
6.4. Future-scenario calculations
The Future-scenario is a combination of three sub-scenarios, based on predictions and assumptions from literature. The sub-scenarios are called RCP2.6, Model and RCP8.5, which differ only in the initial annual temperature data for CROPWAT. The calculation method for crops and livestock are the same as in the Past-scenario. The VWT values from the Past-scenario are used to represent the potential impact of trade in the Future-scenario. The confidence interval is created by
27 combining the WSR of the sub-scenarios RCP2.6 and RCP8.5 and the fluctuations in WSR from the Past-scenario.
All three sub-scenarios have the same livestock and crop conditions (Table 2, Table 3). These assumptions are reasoned with population growth, an increase in consumption levels per capita and arable land expansion compared to 1999 (FAO 2003).
Table 2: Prediction by FAO (2003) on yield and harvest change on the main crop types in SSA by 2030. The year 1999 is used as a reference point for future predictions. These values are used for all sub-scenarios in the Future-scenario.
Harvest (ton) Millet Pulses Sorghum Soybean Harvest increase (%) Maize Rice Wheat
1999 36 60 40 41 1999 49 14 17
2030 38 57 45 72 2015-2030 24 7 8
Change 2 -3 5 31 Annual 0.0158 0.005 0.00548
Annual 0.065 -0.097 0.161 1
Yield (ton/ha) Millet Pulses Sorghum Soybean Yield increase (%) Maize Rice Wheat
1999 0.73 0.67 1.11 1.84 1999-2030 51 88 83
2030 1.12 1.09 1.66 2.63 2015-2030 25 43 40
Change 0.39 0.42 0.55 0.79 Annual 0.0165 0.028 0.02677
Annual 0.013 0.014 0.018 0.025
Table 3: Estimation of annual increase in livestock and its products for SSA in 2030 (FAO 2003). The year 1999 is used as a reference for future predictions. These values are used by all sub-scenarios of the Future-scenario.
Item Annual change (%)
animals meat Carcass weight
Bovine 1.1 3 0.7
Ovine 1.2 3 0.7
Pig 1.3 2.4 0.7
Poultry 2.2 5.1 0.7
Milk 2.8
Eggs 4.1
