TOOLS AND DATA NEEDS FOR FORECASTING AND EARLY WARNING Professor Richard Samson Odingo Department of Geography and Environmental Studies University of Nairobi, Kenya
THE NEED FOR ADEQUATE DATA AND APPROPRIATE TOOLS INTRODUCTORY REMARKS There has been a marked increase in the awareness about extreme weather events, which are linked to the process of climate change. This increased awareness has thrown a challenge to traditional methods of data gathering, and for the use of standard meteorological tools for weather forecasting.
TOOLS IN COMMON USE The commonest tool for manipulating past and present climatic data sets is Time Series Analysis Data records from various sources taken on an hourly, daily, weekly, monthly, or annually are taken as the basis for forecasting decisions The meteorological elements include rainfall, temperature, and humidity They are analysed for their temporal behaviour as a basis for future forecasting
USE OF TIME SERIES FOR FORECASTING Trend analysis help to establish seasonality, cyclic features related to ENSO Events and signals, inter annual patterns, as well as random occurrence of the data being analysed All data re usually subjected to standard statistical analysis to reveal their components which can form the basis of a forecast An example of Time series Graph is given in the next slide
WHAT TIME SERIES DATA CONTAIN Time series data show the temporal behaviour of rainfall or temperature for a given plave, and when the series is long enough it can be used for indication not just variability, but also change. Time series can be used to comment on past climate as well as future climate
Examples of climate variability in Africa
SEASONAL TO INTER ANNUAL TIME SCALES It is now possible to predict climate with improved accuracy in time spans ranging from one season to even more than one year in advance The process of early warning has been facilitated by the possibility for long term prediction Predictions are based on well known weather generators such as ENSO Events and others
YEAR TO YEAR RAINFALL VARIABILITY FOR Different Locations and seasons: N. Kenya
WEATHER SYSTEMS USED FOR PREDICTION THE INTER TROPICAL CONVERGENCE ZONE(ITCZ) MONSOONAL AND ASSOCIATED WIND SYSTEMS SUB TROPICAL ANTICYCLONES CYCLONES SQUALL LINES EL NINO SOUTHERN OSCILLATION(ENSO)
WEATHER SYSTEMS USED FOR PREDICTION II NORTH ATLANTIC OSCILLATION EXTRA TROPICAL WEATHER SYSTEMS NORTH ATLANTIC OSCILLATION THE INDIAN OCEAN DIPOLE OTHER LOCALISED WEATHER GENERATORS
LONG TERM WEATHER FORECASTS AND EARLY WARNING From the point of view of the users of climate information it would be ideal to have rainfall forecasts 3 6 months in advance. Unfortunately this is not always possible. Climate sensitive sectors demand this information for early warning. Such sectors as Agriculture, Food Security,Health,Water Rsources management, and Livestock management would fare better with this information
MORE ON EARLY WARNING NEEDS THE DEMAND FOR EARLY WARNING CLIMATE INFORMATION IS TO ENABLE SUSTAINABLE DEVELOPMENT TO TAKE PLACE, AND TO BENEFIT SUCH OTHER AREAS AS CLIMATE CHANGE ADAPTATION AS WELL AS DISASTER RISK REDUCTION
PREDICTION OF ENSO EVENTS SCIENTIFIC MODELS TO PREDICT EL NINO SOUTHERN OSCILLATION AND LA NINA ARE NOW AVAILABLE IN EASTERN AND SOUTHERN AFRICA. ENSO IS A MULTI ANNUAL CYCLE IN THE TROPICAL PACIFIC OCEAN WHICH HAS TELECONNECTIONS THROUGHOUT THE TROPICS
PREDICTION OF ENSO EVENTS CLIMATE ANOMALIES IN AFRICA ARE ASSOCIATED WITH DIFFERENT EL NINO PHASES CF. MODOKI IN 2009/2010 WHICH WAS ATYPICAL NOW FOLLOWED BY LA NINA During El Ninos, much of Southern Africa often experiences low seasonal rainfall and poor crop yields. NB>Most models used cannot predict weather events beyond 10 days data badly needed
FORECASTING SEASONAL RAINFALL ENSO signals are still very much used for seasonal forecasting in Eastern Africa. They are combined with other techniques to forecast the three month rainfall features in the Greater Horn of Africa countries which fall under ICPAC. The type of forecast obtained may be limited, but it is accompanied with local knowledge using historical perspectives as well as putting historical data in a
SEASONAL RAINFALL FORECASTS Historical data is put in a probabilistic framework by involving local experts who know the seasons in their territory, and the historical picture for correctly interpreting the data. Thus the NMHS are involved through the DMCs, and a consensus forecast is arrived at after looking at what the weather is doing in other parts of the world for the coming season.
OTHER DATA SOURCES FOR SEASONAL FORECASTS It has been emphasised that although models could not predict weather events more than 10days in advance, they could forecast seasonal conditions. This explains why during the COFs the entire sub region tries to arrive at a consensus forecast, which takes into consideration El Nino La Nina, Sea surface Temperatures, and all the other weather generating events. The outcome has been good
DYNAMICAL MODELING APPROACHES Dynamical approaches are now available at all the global meteorological centres as well as research/monitoring institutions that make routine global forecasts. The experience in Rastern Africa has been to use these to sharpen the sub regional forecasts, and to make them more relevant. Combined with local expertise they have served the subregion very well.
REGIONAL CLIMATE MODELING To address the local needs, regional climate models have been found to be extremely useful. They have provided a way of addressing the local needs of dynamical model products, and when converted to local scales, they have proved to be very useful.
1961 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 5 4 3 2 1 0-1 -2-3 -4 The ECHAM4.5 model simulation of Interannual variability during the season October to December over Eastern Africa. Years ec45mean ec45m15 CRU_OBS ST_OBS Rain Index
SATISFYING USERS NEEDS For forecasts to be in demand they must satisfy specific users needs: Power generation Agriculture and Food Security Malaria Outlook Forum Water Management Livestock Management
THE COF AND SEASONAL FORECASTS ICPAC has now had experience of hosting 26 COFS, and the technique for the preparation of seasonal forecasts has proved very successful. It has involved virtual discussion among the NMHSs in the Sub Region, followed by group work and training of the forecasters at the ICPAC Headquarters. Methodologies have been standardised before a consensus regional forecast is made
COF FORECASTS II Comparison is then made with other global and regional forecasts This is followed by the preparation of a consensus global forecast and sub regional forecast During the COF proper the consensus forecast is deliberated upon by other experts and users A final forecast is then issued to the NMHSs
SECTORAL PARTICIPATION IN OUTCOME During the COF, the following sectors and users fully participate: Agricultural risk mapping management Water Resources Group Malaria Outlook Group Food/famine Early Warning group Media group Other End users
NATIONAL LEVEL OUTCOME FOLLOWING THE REGIONAL FORECAST THE NMHSS TAKE OVER AND DO A NATIONAL FORECAST WITH COMMUNITY CLIMATE INFORMATION NETWORK IN MIND A NATIONAL FORECAST IS THEN ISSUED ONE OR TWO WEEKS AFTER
HOW THE COFS HAVE BEEN RECEIVED FORECASTS HAVE BEEN WELL RECEIVED BY THE CONSUMERS IN GHA EVEN THOUGH THERE ARE STILL PROBLEMS IN THE INTERNATIONAL ARENA TOO IT HAS BEEN CONCLUDED THAT FORECAST USE MAY BE AN EFFECTIVE WAY FOR AFRICA TO PREPARE FOR CLIMATE CHANGE, ECONOMIC DEVELOPMENT, DISASTER REDUCTION AND CLIMATE CHANGE ADAPTATION
SEASONAL CLIMATE EARLY WARNING IN THE GHA SUB REGION, FOOD EARLY WARNING IS PREPARED BY FEWSNET PARTLY USING THE COF MATERIAL. THE IMPROVED ACCURACY OF THE COFS HAS MADE IT POSSIBLE TO ISSUE FOOD EARLY WARNING DURING THE COFS. MANY RISK MANAGEMENT STRATEGIES CAN NOW BE PUT IN PLACE SOON AFTER THE COFS
CONTRIBUTION OF REMOTE SENSING TO DATA NEEDS Satellite technology has added a major source of data for weather forecasting,and the investigation of climate Data obtained are estimates, but combined with ground truth they form an essential part of forecasting tools The global satellite coverage is so complete that hourly monitoring of weather events is now the order of the day
. Figure 30a: Example of satellite observation systems within the WMO weather and climate monitoring system (WMO, 2002)
. Example of satellite observation systems within the WMO weather and climate monitoring system (WMO, 2002)
Current Food Security Status: Affected Population and Underlying Issues 15 16 Million 0.13M Recurrent climate extremes (drought & floods) in marginal areas 1.3M 6.4M Conflict and civil insecurity Transboundary diseases 0.7M 2.7M 3.2M Hyperinflation/Market disruption Declining Pastoral Terms of Trade IDPs and refugees Poverty/Malnutrition Source: FEWS NET Policies
SUMMARY OF DATA SOURCES Atmospheric Observations GCOS and GUAN and GAW Oceanographic Observations SSTs, Winds Waves, salinity etc. Terrestial Observations Satellite Observations