HORN OF AFRICA: NATURAL HAZARD PROBABILITY AND RISK INTRODUCTION: METHODOLOGY: The Horn of Africa (HOA), composed of Djibouti, Eritrea, Ethiopia, Kenya, and Somalia, suffers frequent natural The probability of one or more hazard events occurring within a grid cell is given by disasters that commonly result in losses of life, destruction of infrastructure, and reduction of agricultural production. P(1 or more events) = 1 - exp(-E[N(HI)]), Formulating effective contingencies to respond to such emergencies is constrained by a limited understanding of the where E[N(HI)] = λ(HI)*T. E is the expected number of events within time T given historical rate of occurrence λ. likelihood of a natural hazard occurring within a particular region and risks associated with that hazard. Robust early The summation is over all events within a given grid cell. This calculation assumes a Poissonian, or time- warning systems exist for national response and "hot spot" maps of risk have been produced at a global level; this level of independent, distribution of probability. This means that the probability of a future event is independent of the timing resolution, however, is often not sufficient for sub-national resource distribution. This study presents the probabilities of and magnitude of the most recent events in the region. The measure of risk R(HI) is obtained by combining the natural hazards and the risk to populations or agricultural systems within the HOA, calculated on 1° by 1° grid cells. probability P(HI) of a hazard occurring with the societal exposure X(HI), or exposure, to that hazard: Such an analysis can provide a regional understanding of the probability of natural hazards as well as a more specific R(HI) = Probability of hazard occurring [P(HI)] * Societal exposure to that hazard [X(HI)]. local characterization of the associated risks.

The Horn of Africa is known for In the period from 1988-2006, the Analytical grid Analytical grid cell boundary frequent and severe droughts. On cell boundary Shabeelle (20 floods), Juba (12), International Annual Probability average, a severe drought may be International Annual Probability Awash (11) and Tana Rivers (9) were boundary anticipated once every couple of boundary the most flood-prone rivers within the Administrative Red Sea Administrative Red Sea Horn of Africa. On average, the boundary of Drought years during the months that boundary of Flooding Provisional ERITREA comprise the rainy season(s). Most Provisional ERITREA Shabeelle floods once a year and the administrative Asmara administrative Asmara ! recently, a two-year drought occurred ! Juba and Tana flood once every two line line 15° in an area that covered much of 15° years. Worst effects of flooding eastern Ethiopia, northeastern Kenya, include human deaths (typically 10s 0 km 200 400 Mek'ele 0 km 200 400 Mek'ele ! and southern Somalia. These ! to 100s), displacements (typically 1k Gonder Gonder to >100k), destroyed homes, ! droughts affect hundreds of ! DJIBOUTI Gulf of Oman thousands of residents and are DJIBOUTI Gulf of Oman submerged agriculture, and Boosaaso Boosaaso waterborne diseases (e.g. malaria, Dese Dikhil Djibouti ! exacerbated by the extreme poverty Dikhil Djibouti ! ! ! ! ! ! Ali Sabieh Ceerigaabo and persistent conflict. Awash Ali Sabieh Ceerigaabo diarrhea, and Rift Valley fever). Debre Markos ! Debre Markos ! ! ! 10° Hargeysa Burao 10° Hargeysa Burao Flooding by Country Harar ! ! Harar ! ! 20 Nek'emte Addis Ababa ! Rainfall Regimes Nek'emte Addis Ababa ! ! ! ! ! Somalia Laascaanood Garoowe Laascaanood Garoowe Kenya Gore ! ! Gore ! ! 15 Ethiopia ! ETHIOPIA ! ETHIOPIA Eritrea Jima 15° Jima ! ! Djibouti 10 Awasa Goba Awasa Goba ! ! Gaalkacyo ! ! Gaalkacyo ! Shabeelle ! 5

Arba Minch Arba Minch Number of Floods ! Dhuusa Mareeb ! Dhuusa Mareeb ! ! 10° 0 Beledweyne Beledweyne 5°N ! 5°N ! 1990 1994 1998 2002 2006 Xuddur Xuddur ! ! Years Garbahaarey Garbahaarey ! Baydhabo 5°N ! Baydhabo ! Jawhar ! Jawhar Locations of Floods (1988-2006) ! ! SOMALIA SOMALIA ! ! Wajir Wajir Jub ! > 45% ! KENYA Mogadishu KENYA a Mogadishu Bu'aale Number of Rainy Djibouti ! 0° 41 - 45% Seasons Per Year 70 - 80% 5% Eritrea 2% ! 1 ! Kisumu 36 - 40% Kisumu 60 - 70% ! Nakuru 2 ! Nakuru Somalia 0° ! Nyeri Kismaayo 0° ! Nyeri Kismaayo ! Embu ! ! Embu ! ! 31 - 35% 3+ ! T 50 - 60% 31% a n Ethiopia Nairobi 26 - 30% Nairobi a 40 - 50% ! 5°S ! 36% 21 - 25% 35°E 40° 45° 50° 30 - 40% 16 - 20% 20 - 30% Kenya 11 - 15% Land Area and Drought Probability 10 - 20% 26% Annual Probability Indian Indian Annual Probability Mombasa 6 - 10% Mombasa 0 - 10% ! 13% 19% ! Ocean 0 - 5% 4% 26% 2% Ocean a e

r 28% 5°S 61% 5°S Generalized HOA Biannual Rainfall Calendar A 65% 35°E 40° 45° 50° 35°E 40° 45° 50° Dry season Dry season 54% 71%

t

n J F M A M J J A S O N D e

c 59% Exposure to Exposure to Drought Risk r Flooding Risk e 37% Major rains Minor rains Drought P Land 31% Flooding 20% ERITREA ERITREA 10% ERITREA ERITREA DROUGHT ! ! ! ! Data from the Dartmouth Flood Asmara Asmara Asmara 15° 15° 15° 15° Asmara FLOODING DJIBOUTI ERITREA ETHIOPIA KENYA SOMALIA Observatory is the basis for calculating Drought probability: , , < 25% 25-50% > 50% flood probability. A grid cell is DJIBOUTI DJIBOUTI DJIBOUTI DJIBOUTI Djibouti Djibouti This project uses the Standardized Djibouti Djibouti considered to have flooded anytime a

ETHIOPIA ETHIOPIA ETHIOPIA ETHIOPIA 10° 10° Precipitation Index (SPI) to 10° 10° river intersecting that cell has flooded.

Addis Ababa Addis Ababa determine the occurence of drought Addis Ababa Addis Ababa Flooding in the HOA often leads to within the historical record. The SPI drowning, disease (cholera, malaria, method provides a statistical Rift Valley fever), disrupted 5°N 5°N characterization of anomalous 5°N 5°N transportation and food supply, and rainfall in the historical data (CSU, infrastructure damage. Heavy rains, SOMALIA Mogadishu SOMALIA Mogadishu SOMALIA Mogadishu SOMALIA Mogadishu 2005; Husak, 2005). Only droughts 4.5 - >5.0 however, are often followed by bumper KENYA > 45% KENYA 0.9 - 1.0 KENYA KENYA 0.9 - 1.0 4.0 - 4.5 41 - 45% 0.8 - 0.9 0.8 - 0.9 that occur during the agriculturally 3.5 - 4.0 crop yields. The largest recent floods 36 - 40% 0.7 - 0.8 0.7 - 0.8 0° 0° 0° 3.0 - 3.5 0° 31 - 35% 0.6 - 0.7 0.6 - 0.7 important rainy seasons are counted. 2.5 - 3.0 in the region occurred in 2006, 2001, 26 - 30% 0.5 - 0.6 0.5 - 0.6 2.0 - 2.5 Nairobi 21 - 25% Nairobi 0.4 - 0.5 Rainy seasons were uniquely Nairobi Nairobi 0.4 - 0.5 1998/1997, 1996, and 1988. 16 - 20% 0.3 - 0.4 1.5 - 2.0 0.3 - 0.4 1.0 - 1.5 0.2 - 0.3 Land Coverage 11 - 15% 0.2 - 0.3 Normalized Risk Normalized Risk defined for each grid cell based upon 0.5 - 1.0 0.1 - 0.2 Population density, derived from the 6 - 10% 0.1 - 0.2 Population (in millions) Percent of Agricultural Percent of 0 - 5% 0.0 - 0.1 0 - 0.5 0.0 - 0.1 LandScan dataset, is the measure for 5°S 5°S monthly rainfall frequencies over the 5°S 5°S 35°E 40° 45° 50° 35°E 40° 45° 50° period of analysis (1986-2006). 35°E 40° 45° 50° 35°E 40° 45° 50° societal exposure to floods.

Locust infestations occur periodically The Horn of Africa has significant Analytical grid and are endemic to a broad band Analytical grid seismic hazard associated with the East cell boundary cell boundary Shaking level with a 10% International Annual Probability of covering northern Africa that extends International African rift system. A number of boundary east to the margins of the Red Sea. A boundary destructive earthquakes, some deadly, Administrative Red Sea single swarm can cover 1200 square Administrative chance of exceedence have been reported this century. boundary Locust Infestation boundary kilometers and can contain between ERITREA Notable events include the 1921 Provisional ERITREA Provisional Red S administrative Asmara 40 and 80 million locusts per square administrative Asmara in 50 years earthquake that destroyed the port city of ! ! line 15° line 15° kilometer. With each insect capable of ea Massawa in Eritrea. In Ethiopia, the eating its own body weight in 1961 M=6.1 Kara Kore earthquake 0 km 200 400 Mek'ele 0 km 200 400 Mek'ele ! vegetation each day, a swarm that ! destroyed the town of Majete, the 1969

Gonder size could consume 192,000 metric Gonder M=6.3 Serdo event resulted in four ! ! tons of vegetation each day. lf of Oman deaths, and the 1989 M=6.5 Dobi DJIBOUTI Gulf of Oman DJIBOUTI Gu Boosaaso Boosaaso graben earthquake destroyed several Dese Dikhil Djibouti ! Dese Dikhil Djibouti ! ! ! ! Saudi ! ! ! Ali Sabieh Ceerigaabo Ali Sabieh Ceerigaabo bridges between Assab and Addis Debre Markos ! Arabia Debre Markos ! ! ! Ababa. 10° Hargeysa Burao 10° Hargeysa Burao Harar ! ! Sudan Harar ! ! Nek'emte Addis Ababa ! Nek'emte Addis Ababa ! ! ! ! breeding areas Red ! Laascaanood Garoowe Sea Laascaanood Garoowe Earthquake Locations Gore ! ! Gore ! ! ! ETHIOPIA ! ETHIOPIA (M>4.4, AD 627-1994) Jima Jima ! ! Awasa Goba Awasa Goba ! ! Gaalkacyo ! ! Gaalkacyo 15° ! Yemen ! Arba Minch Arba Minch ! Dhuusa Mareeb ! Dhuusa Mareeb ! Eritrea ! 5°N Beledweyne 5°N Beledweyne ! ! 10° Xuddur Xuddur ! March 21-31, 2007 Vegetation Anomaly (% NDVI) Image from NASA ! Garbahaarey -100 0 100 Garbahaarey ! Baydhabo ! Baydhabo ! Jawhar ! Jawhar ! Along the Red Sea coast, the locusts' ! SOMALIA 5°N SOMALIA winter breeding area, swarms develop ! Wajir ! Wajir ! ! when rain falls on the sandy soil to Mogadishu 1.8 - 2.0 KENYA Bu'aale Mogadishu KENYA Bu'aale ! ! ) 1.6 - 1.8 initiate the hatching of locust eggs. If 2 ! ! 1.4 - 1.6 Earthquake Kisumu plenty of water and vegetation for food Kisumu ! ! Nakuru 0° Magnitude 0° Nakuru Kismaayo 0° ! Nyeri Kismaayo ! Nyeri Embu ! ! Embu ! ! ! 50 - 60% exist in the locust breeding areas, a ! 1.2 - 1.4 6.5 - 7.5 Nairobi Nairobi 1.0 - 1.2 5.5 - 6.5 ! 40 - 50% large number of the insects hatch and ! 5.0 - 5.5 30 - 40% form swarms. This SPOT satellite 0.8 - 1.0 4.5 - 5.0 0.6 - 0.8 4.4 - 4.5 20 - 30% image shows in green where 5°S 10 - 20% vegetation was more thick and lush in 0.4 - 0.6 35°E 40° 45° 50° Acceleration (m/s

Indian Peak Ground Surface 0.2 - 0.4 Mombasa Indian 0 - 10% March 2007 than the average March in Mombasa ! Annaul Probability ! 0.0 - 0.2 Earthquake Locations by Country Ocean 1999 - 2006. Strips of dark green line Ocean 80 5°S the shores of the Red Sea in the same 5°S Somalia Kenya 35°E 40° 45° 50° areas where locust swarms were 35°E 40° 45° 50° Ethiopia 60 Eritrea spotted. Djibouti LOCUSTS Locust Infestation Locations of Locust Infestations Exposure to Exposure to Ground Shaking 40 Locust Infestation Risk Ground Shaking Risk ERITREA ERITREA ERITREA ERITREA Somalia ! ! ! ! 15° Asmara 15° Asmara 17% 15° Asmara 15° Asmara 20 Number of Earthquakes

Ethiopia DJIBOUTI DJIBOUTI Eritrea DJIBOUTI DJIBOUTI 0 26% Djibouti Djibouti 57% Djibouti Djibouti 4.5 5.0 5.5 6.5 Magnitude ETHIOPIA ETHIOPIA ETHIOPIA ETHIOPIA EARTHQUAKES 10° 10° 10° 10° ! Earthquake location, magnitude, and Addis Ababa Addis Ababa Addis Ababa Addis Ababa probability data are taken from the This locust risk analysis incorporates Global Seismic Hazard Assessment 5°N 5°N data from 1992-2006 on gregarious 5°N 5°N Program (GSHAP), which evaluated populations of hoppers and adults earthquakes occurring in the period A.D. Mogadishu SOMALIA Mogadishu SOMALIA Mogadishu SOMALIA SOMALIA Mogadishu ! 627-1994. Shaking levels are shown as 4.5 - >5.0 KENYA > 45% KENYA 0.9 - 1.0 compiled by the FAO Locust Control KENYA KENYA 0.9 - 1.0 41 - 45% 0.8 - 0.9 4.0 - 4.5 0.8 - 0.9 peak ground surface acceleration. 36 - 40% 0.7 - 0.8 Office. Infestation frequency and 3.5 - 4.0 0.7 - 0.8 0° 0° 0° 0° 2 31 - 35% 0.6 - 0.7 3.0 - 3.5 0.6 - 0.7 Accelerations of 1 to 2 m/s are probability are aggregated to the 2.5 - 3.0 0.5 - 0.6 ! 26 - 30% 0.5 - 0.6 2.0 - 2.5 Nairobi 21 - 25% Nairobi 0.4 - 0.5 Nairobi Nairobi 0.4 - 0.5 considered strong to very strong shaking 16 - 20% 0.3 - 0.4 corresponding one-degree grid cells. 1.5 - 2.0 0.3 - 0.4 1.0 - 1.5 0.2 - 0.3 0.2 - 0.3 Normalized Risk Land Coverage 11 - 15% Normalized Risk 0.5 - 1.0 and will produce light to moderate 6 - 10% 0.1 - 0.2 Societal exposure is based upon the Population (in millions) 0.1 - 0.2 Percent of Agricultural Percent of 0 - 0.5 0 - 5% 0.0 - 0.1 0.0 - 0.1 damage. Note that a M > 7 event could intensity of crop and livestock 5°S 5°S 5°S 5°S 2 35°E 40° 45° 50° 35°E 40° 45° 50° activities in the region. 35°E 40° 45° 50° 35°E 40° 45° 50° produce shaking that far exceeds 2 m/s . KEY FINDINGS:

Analytical grid This joint probability map shows the likelihood that a Analytical grid cell boundary Annual Probability of given region will sustain any natural disaster (i.e., cell boundary Most Probable Drought: Eastern and western Ethiopia and western Kenya are the areas most frequently affected by International International boundary drought, flood, or locust infestation) within any given year. boundary severe drought, with a greater than 40% annual probability of moderate to severe drought during Administrative Drought, Flood, or Earthquakes are not included in the joint probability Administrative Hazard Type boundary boundary ERITREA R calculation because large damaging events occur on a ERITREA R Provisional e Provisional e d S Locust Infestation the rainy seasons. administrative Asmara far less frequent time scale than the other three hazard administrative Asmara d S ! ! line e line e 15° a types. The map suggests that most regions of the Horn 15° a Flooding: On the long-term average, the Shabeelle River floods once a year and the Juba, Awash, and could expect to receive a natural disaster on an annual to 0 km 200 400 Mek'ele 0 km 200 400 Mek'ele ! semi-annual basis. Regions typically affected by ! an Tana Rivers flood once every two years. The densely populated areas along these rivers in Gonder Om Shabeelle River flooding suffer most frequently, while the Gonder man ! f of ! f O Gul Bari and Sanag regions of Somalia, and the Gulf o DJIBOUTI DJIBOUTI Ethiopia, Kenya and Somalia face the highest risk associated with flooding within the HOA. Boosaaso northeasternmost Ethiopia may expect disasters least Boosaaso Djibouti ! Dikhil Djibouti ! Dikhil! ! ! ! Awash Ali Sabieh Ceerigaabo frequently. Awash Ali Sabieh Ceerigaabo Debre Markos ! Debre Markos ! Locust Infestation: Historical infestations are clustered in northern Eritrea and near the conjunction of ! ! 10° Hargeysa Burao 10° Hargeysa Burao Harar ! ! Harar ! ! Nek'emte Addis Ababa ! Nek'emte Addis Ababa ! ! ! Somalia, Ethiopia, and Djibouti, and the annual probability in these regions can exceed 30-40%. ! ! Laascaanood Garoowe Laascaanood Garoowe Gore ! ! Area Coverage of Most Probable Hazards Gore ! ! ! ETHIOPIA ! ETHIOPIA Jima 100% Jima Over the past 15 years there have been 106 documented infestations, with 57% occurring in ! ! Awasa Goba Awasa Goba ! ! Gaalkacyo ! ! Gaalkacyo ! 80% ! Shabeell Shabeell Indian Eritrea, 26% in Ethiopia, and 17% in Somalia. Arba Minch Arba Minch ! e Dhuusa Mareeb 60% ! e Dhuusa Mareeb Ocean ! ! Seismic Activity: The largest earthquakes are most likely to occur along the rift regions of Djibouti, Beledweyne Beledweyne 5°N ! 40% 5°N ! Xuddur Xuddur ! ! 20%

National Land Area southern Eritrea and northeast Ethiopia; however, due to the relative infrequency of large Garbahaarey Garbahaarey ! Baydhabo ! Baydhabo ! Jawhar ! Jawhar ! 0% ! Djibouti Eritrea Ethiopia Kenya S omalia SOMALIA SOMALIA damaging earthquakes in the HOA, the risk from seismic hazard is not an annual concern. ! J ! Wajir J Wajir ! ub ! ub This map showing the most probable type of natural KENYA Mogadishu KENYA a Mogadishu a Bu'aale Bu'aale ! Annual disaster in each region can aid in the allocation of ! Combined Hazard: Calculations of joint probability of drought, flooding, and locust infestation ! ! Kisumu Probability mitigation resources. Each distaster type dominates in a Kisumu 0° ! Nakuru Kismaayo 0° ! Nakuru Kismaayo ! Nyeri Embu ! ! Nyeri Embu ! suggest that most regions of the Horn could expect to receive a natural disaster on an annual to ! ! Tan 80 - 90% paricular physiographic environment - locusts ! ! Tan 70 - 80% Nairobi a a predominate along the Red Sea and Gulf of Aden, flloods Nairobi ! 60 - 70% ! semi-annual basis with drought being the most probable over the most land area, followed by 50 - 60% are limited to the main drainages, and droughts dominate Hazard Type 40 - 50% all other regions. The bar chart summarizes the map 30 - 40% No dominant type information. In summary, drought is the most probable flooding. Indian 20 - 30% Drought Mombasa 10 - 20% Mombasa Locust Infestation ! natural disaster in every Horn country. Locust infestation ! Ocean 0 - 10% Flood U.S Department of State

COMBINED is the second most probable disaster in Djibouti and 5°S Eritrea, while floods are second most probable in 5°S [email protected] 35°E 40° 45° 50° Ethiopia, Kenya, and Somalia. 35°E 40° 45° 50° http://hiu.state.gov

HUMANITARIANHIU INFORMATION UNIT