Drought forecasting and warning in Africa Some experiences & thoughts from the DEWFORA project Micha Werner 1 & Gilbert Ouma 2 1 Deltares & UNESCO-IHE, the Netherlands 2 ICPAC & University of Nairobi, Kenya Acknowledgements to the full DEWFORA team
The challenges that are ahead for drought forecasting and monitoring are basically data, we don t have enough data The methods for forecasting drought have not been properly done, so there is also a challenge there And lastly the socio-economic aspects of droughts; we have poverty issues within the African communities, that would really interfere with coping with drought, whether we forecast it or not Gilbert Ouma, 2013 ICPAC & University of Nairobi http://africanclimate.net/en/water-africa-changing-climate
The DEWFORA Approach An evidence-based protocol for designing and implementing drought early warning systems What is the science available? What are the societal capacities? How can science be translated into policy? How can society benefit from the forecast?
The situation is that there is not enough water in Southern Africa, the rainy season is short. benefit so we need research to better manage the water resources Jakkie Venter, Area Manager, Department of Water Affairs, Tzaneen area office, Limpopo Province, South Africa http://www.euronews.com/2013/06/17/africa-is-always-at-risk-of-drought/
benefit If there is a drought, the small farms will collapse, they won t function, they will get not profit from their animals if I can get advance about drought, then I will be able to organise some feeds for the cattle, I will be able to see to it that my dam is up to date, you know, water is enough in the dams Stephen Lebotsa, Chairman Mobidibeng dairy Cooperative Limpopo Province, South Africa http://www.euronews.com/2013/06/17/africa-is-always-at-risk-of-drought/
LIMPOPO BASIN CASE STUDY Ø Droughts in the Limpopo: 1982/83, 1987/88, 1991/92, 1994/95, 2002/03, 2004/05, 2006/07 Questions: Do we have the science to provide skilful (hydrological) drought forecasts? and Do these provide variables that are useful to water users? science
METHODOLOGY science Meteorological forcing "Pre-processing" Hydrological forecasting Results Precipitation Temperature (mean, max, min, - ensembles) Biascorrection of precipitation Estimation of potential evaporation (PE) PCR- GLOBWB hydrological model DELFT- FEWS forecasting shell Predicted (ensembles) streamflow, soil moisture, and other hydrological fluxes time time Possible states of the forecast Probability of (not) exceeding thresholds
FORECASTING SYSTEMS science Ø 3 seasonal forecasting systems: same hydro- model forced by different meteorological forecasts FS_S4 FS_ESP FS_ESPcond Meteorological seasonal forecasting system S4 from ECMWF 7 month lead time Hindcast from 1981 to 2010, with 15 ensemble members Climatological bias correction of seasonal forecasts of precipitation Ensemble generated with resampled historical meteorological data Meteorological events that occurred in the past are representative of events that may occur in the future (Day, 1985) Allows to measure the skill that can be expected form only the memory in the hydrology Post-ESP weighting technique (Werner et al. 2004) Uses the El Niño - 3.4 index averaged over the 3 months-period immediately before the issue date of the forecast to weight ensemble members from ESP w λ 1 x i i = 1, k = x k n α
ASSESSING SKILL science Forecast is justified if it supports better decision making then it has value Ø Major use in the basin: irrigation. Meaningful indicators for a better decision making: o standardised Runoff Index (SRI) o agricultural drought (soil moisture) o Reservoir levels: curtailments in irrigation Ø Verification skill scores (assess quality): o ROC Ability of the forecast to discriminate between events and non-events o BSS Relative skill of the probabilistic forecast over that of climatology o Rank histogram How well does the ensemble spread of the forecast represent the true variability (uncertainty) of the observations
RESULTS SKILL ASSESSMENT ROC DIAGRAMS science SRI6-0.5 : Skill in predicting moderate drought at 5 month lead time ROC Score: area under the curve. 1 : Perfect < 0.5 : No skill FS_S4 FS_ESPcond 0 1 0 1 0 1
RESULTS ROCS SOIL MOISTURE & WATER science LEVEL IN RESERRVOIRS Tzaneen Dam Curtailments (hedging) to irrigation & water supply if reservoir levels lower than normal ROCS for: Water Level (WL) < 50th percentile (upper plot), and WL < 37.5th percentile (lower plot) for the FS_S4 forecasts
Can skilful forecasts be provided and do these provide useful variables? Hydrological drought indicators (e.g) SRI-6 can be predicted with skill at lead times of up to 5 months (for the wet season) Skill dominated by initial conditions for 2-3 months Skill dominated by uncertainty in meteorological forcing at larger lead times ECMWF seasonal forecasts (S4) model provides most reliable forecasts but approach using resampled climatology conditioned by ENSO close second Hydrological drought forecast provide variables such as levels in reservoirs can be used e.g. by reservoir operators and irrigation districts Plan curtailments Implement demand reduction actions in response P. Trambauer, M. Werner, H.C. Winsemius, S. Maskey, E. Dutra, S. Uhlenbrook (2014). Hydrological drought forecasting and skill assessment for the Limpopo river basin, Southern Africa. Hydrol. Earth Syst. Sci. Discuss., 11, 9961-10000
LIMPOPO BASIN CASE STUDY science Does seasonal forecasting gain importance in the future due to climatic change? Do critical weather conditions (that may benefit from forecasting) occur more frequently? Can these then be forecast with skill? Critical conditions for subsistence farming Rainfed agriculture (Maize): Dry Spells Dairy farming (cows): Extreme Heat Index Source: http://www.unesco-ihe.org/project-activities/ Project-Portfolio/Small-holder-System-Innovations-in- Integrated-Watershed-Management Concentrate on forecasts for DJF wet season
Method: Climate Change projections & Seasonal Forecasts Possible evolution of the climate Multiple climate models Conditioned on A2 emission scenario Downscaled using Regional climate model over Southern Africa (CCAM) 7-month ahead probabilistic seasonal forecast updated monthly 15 ensemble members 0.5 degrees scale Climatology: ERA-Interim (1978 2014)
Experiment forecast skill science Frequency of events in DJF period Lead time [months] 5 4 3 2 1 Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar
RESULTS FREQUENCY OF EVENTS IN DJF SEASON science Expected number of dry spells longer than 5 days Probability of non-exceedance Expected number of days with Temperature Heat Index > 78 Changing climate
RESULTS HOW WELL CAN WE FORECAST EVENTS? science Skill of forecasts of dry spells longer than 5 days 5 4 3 Skill of forecasting days with Temperature Heat Index > 78 2 1 Skill reduces with lead time [months]
Will seasonal forecasts across Southern Africa gain importance? Dry spell frequency (rainfed agriculture) Can be forecast with some skill 1-2 months ahead in time; No indication of dry spells becoming more frequent in the future Forecasts are already potentially useful in some areas (skill), but will not gain significant importance from the climate change point of view Heat stress conditions(cattle) Expected to become more frequent in the future Can be forecast with skill even up to 4 months ahead Forecasts are already potentially useful, and will become more important in the future H. Winsemius, E. Dutra, F. Engelbrecht, E. Archer Van Garderen, F. Wetterhall, F. Pappenberger, and M. Werner. 2014. The potential value of seasonal forecasts in a changing climate. Hydrol. Earth Syst. Sci., 18, 1525-1538.
GREATER HORN OF AFRICA CASE STUDY society What is the capacity of society to deal with drought? Can a Drought Vulnerability Index assist in prioritisation? 1 Define hazard (meteorological or hydrological) 2 Define the impacts in agriculture, water, ecosystems, health, etc 3 Define exposure (number of people affected) RISK: HAZARD x VULNERABILITY 4 Define coping capacity or adaptive capacity
Methodology: building the DVI society Vulnerability indicators describe social-based responses: anticipation, adaptation and reaction to drought. Social Capacity: capacity of a society to develop knowledge and awareness. Technological efficiency: capacity to develop, export and use eco-efficient technologies. Natural Capital: reliability and vulnerability of water resource infrastructure to confront water scarcity. Economic Capacity: capacity of a system to make investments in development industry, food security and income stabilization DVI = f sc *WI sc + f ec *WI ec + f te *WI te + f nc *WI nc
GREATER HORN OF AFRICA (KENYA) CASE STUDY Establish indicators based on data from County Fact Sheets Seminar with experts (weights) society Spatial resolution: County level
SOCIAL CAPACITY INDICATOR AT COUNTY LEVEL society
RESULTS : SPATIAL DISTRIBUTION OF DVI society Wajir : 0.67 Nairobi : 0.27 Country Average: 0.56 Drought 2010-2011 4,3M people affected in Kenya
Can a Drought Vulnerability Index assist in prioritisation? Proposed method can be applied in assessing drought vulnerability assessment Scale and Country dependent Several improvements can still be made: Indicators considered; Weighting scheme; Traditional approaches in dealing with drought Applied similarly in the Morocco (Oum-er-Rbia basin) Applied at the country level across the African continent E. Mwangi, G. Ouma, A. Opere, M. Faneca Sànchez. 2014. Vulnerability to drought over the Greater Horn of Africa region. Hydrol. Earth Syst. Sci.,(in preparation) Naumann G., Barbosa, P., Garrote, L., Iglesias, A., Vogt, J. 2014. Exploring drought vulnerability in Africa: an indicator based analysis to be used in early warning systems. Hydrol. Earth Syst. Sc. 18(5): 1591-1604 G. Naumann, M. Faneca Sànchez, E. Mwangi, P. Barbosa, A. Iglesias, L. Garrote, M. Werner. 2014. Drought vulnerability assessment for prioritising drought warning implementation. EGU General Assembly 2014, Vienna; 04/2014
African Drought Observatory science edo.jrc.ec.europa.eu/ado/ado.html Meteorologicalforecasting ECMWF S4 Seasonal Forecasts 6-month lead time forecasts Updated monthly Hydrological forecasting PCR-GLOBWB hydrological model DELFT- FEWS forecasting shell
GDIS - Global Drought Information System in collaboration with other (research) projects in Europe, the United States and Australia science African Drought Observatory Pozzi et al., 2013. Towards Global Drought Early Warning Capability: A Framework of International Cooperation for Global Drought Monitoring and Forecasting, Bulletin of the American Meteorological Society. 94(6): 776-785.
Four main challenges remain! Experts from agencies & institutions across Africa interviewed on critical issues that need to be addressed in drought forecasting & warning 40 people, 18 countries
Thank you