Barriers of Using Climate and

Weather Forecasts in Drought

Planning and Decision Making
Table of Contents

List of Abbreviations 3

1.0 INTRODUCTION 4
1.1 Background 4
1.2 Anticipatory Approach in Disaster Management 5

2.0 CLIMATE AND WEATHER FORECASTS IN KENYA 6

2.1 Challenges in using climate and weather forecasts 7
2.1.1 Towards Forecast Based Preparedness (ForPAc) project 7
2.1.2 Challenges 8

3.0 RECOMMENDATIONS, IMPLICATIONS AND CONCLUSION 10

3.1 Recommendations 10
3.2 Implications on FbA 11
3.3 Conclusion 11
List of Abbreviations

ASAL Arid and Semi-Arid Lands

DEWS Drought Early Warning Systems
EDE Ending Drought Emergencies
FAO Food and Agricultural Organisation
FbA Forecast-based Early Action
ForPAc Towards Forecast Based Preparedness
KENGEN Kenya Electricity Generating Company
KMD Kenya Meteorological Department
KRCS Kenya Red Cross Society
MAM March April May
LRA Long Rains Assessment
NDCF National Drought Contigency Fund
NDMA National Drought Management Authority
OND October-November-December
SRA Short Rains Assessment
1.0 Introduction drought risk management. To this end, the EDE
established the National Drought Management
1.1 Background Authority (NDMA) with a mandate to provide
leadership and coordinate drought risk
Recurring droughts have significantly impacted management plans, interventions, polices and
livelihoods and economic development in Kenya. stakeholders across national and county levels.
The frequency of these events is increasing; for To address the challenges of financing drought
example, from 1993 the Government of Kenya management, the EDE established the National
has declared drought as a national disaster Drought Contingency Fund (NDCF). The fund
7 times . These declarations were during the allows the pooling of resources from different
droughts of 1992-93,1996-97, 1999-2000, 2005- actors.
06, 2008-09, 2010-2011 and 2016-2017.
Not only has the frequency of the drought Amongst NDMA’s responsibilities is monitoring
events increased to every 2 -3 years, the of drought conditions by establishing and
severity has also increased in terms of the total operating a Drought Early Warning System
population affected and the humanitarian aid (DEWS) in 23 Arid and Semi-Arid Counties.
needed for response (Table 1). According to the The DEWS aggregates data and information
post-disaster needs assessment conducted by monthly from sentinel sites and key sectors
the Kenya Government , the 2008-2009 and the like education, agriculture, health, livestock
2010-2011 drought events affected a total of and health. Currently, the DEWS monitors
3.7 million people and caused $12.1 billion in biophysical, production, access and utilization
damages and losses . The 2016/2017 drought indicators (Table 2). Biophysical indicators are
event affected 23 of 47 counties, where 2.7 used to monitor progression of the drought
million people were declared to be food insecure hazard while production, utility and access
and 357,285 children and pregnant and lactating indicators monitor the impacts of the drought.
mothers were acutely malnourished . For each indicator, thresholds are set to
define three drought stages: alert, alarm and
Following the 2010-2011 devastating drought, emergency.
the Kenyan Government launched a Medium
Term Plan for Drought Risk Management and The use of observed indicators to monitor and
Ending Drought Emergencies (EDE) for 2013- define drought stages and to trigger funding
2017. The EDE commits to end drought as an from the National Drought Contingency Fund
emergency by the year 2022, by strengthening (NDCF) means the drought management system
institutional and financial frameworks for is reactive and not anticipatory. The integration

Table 1:Historical Comparison of Drought Events in Kenya 2

1
‘DROUGHT IN KENYA.’ <https://www.researchgate.net/publication/321461935_DROUGHT_IN_KENYA> [accessed 15 April 2019].
2
The Post Disaster Needs Assessment - droughts in 2008-2011 was done with technical support from the European Union, United
Nations, and World Bank. https://www.gfdrr.org/en/publication/kenya-post-disaster-needs-assessment-droughts-2008-2011
[accessed 28 June 2019]
3
USAID, ‘Economics of Resilience to Drought: Kenya Analysis’, 43.
4
‘Kenya: Drought - 2014-2019’, ReliefWeb <https://reliefweb.int/disaster/dr-2014-000131-ken> [accessed 18 June 2019].
Table 2: Indicators monitored by the NDMA drought EWS (Source: NDMA)

TYPE OF INDICATOR SPECIFIC INDICATORS MONITORED

Biophysical Rainfall estimates
Vegetation Condition Index (VCI)
State of natural vegetation and pasture conditions
State of water sources
Production Livestock production –Livestock body condition, livestock diseases and
pests, milk production, livestock migration patterns, livestock mortality
Crop production – Timeliness and status of crops, implication on food
security
Access Livestock Prices - Terms of trade, cattle/ goat prices
Price of cereals and other food products
Distance to water for households
Distance to grazing areas
Utilisation Health and Nutrition Status – Mid-Upper Arm Circumference (MUAC),
human diseases and epidemics
Coping strategy index
Milk and food consumption

of weather and climate forecast in the system hydrological services and other humanitarian
can shift the system to be anticipatory and agencies. Coughlan de Perez et al. defined FbA
hence better inform actions to reduce the as when a forecast states that an agreed-upon
impacts of droughts on the economy as well as probability threshold will be exceeded for a
on vulnerable communities before the droughts hazard of a designated magnitude, then an
occur. action with an associated cost must be taken
that has a desired effect and is carried out by a
1.2 Anticipatory Approach in Disaster designated organisation .
Management
FbA initiatives seek to jointly develop standard
Despite the availability of early warning systems operating procedures with key stakeholders,
and information, very often communities, where each stakeholder commits to undertake
governments, humanitarian organisations and certain actions when a forecast for an extreme
international aid agencies tend to respond after, event is issued. For example, in 2017 when rains
rather than in advance of a severe drought. were set to fail in the Horn of Africa region, the
And yet there exists a window of opportunity Food Agricultural Organization (FAO) piloted an
between the issuance of early warning early warning early action system in Marsabit,
information and the impacts of drought events Wajir, Kilifi and Kwale . The system initiated early
being felt where early actions can be taken to actions such as, providing veterinary care for
mitigate impacts on society. livestock and distributing animal feeds.

Recognising this window of opportunity, and The FbA approach has the potential to transform
taking advantage of advances in science, data the way governments, humanitarian actors and
and technology, the Red Cross Red Crescent communities view and address crises caused
Movement developed and has piloted an by climatic hazards from reactive to proactive
approach known as Forecast-based early Action actions. Moreover, investment into early action
(FbA), in partnership with meteorological and can be a more effective way of delivering

5
E. Coughlan de Perez and others, ‘Forecast-Based Financing: An Approach for Catalyzing Humanitarian Action Based on Extreme
Weather and Climate Forecasts’, Natural Hazards and Earth System Sciences, 15.4 (2015), 895–904 <https://doi.org/10.5194/
nhess-15-895-2015>.
6
‘CA0227EN.Pdf’ <http://www.fao.org/3/ca0227en/CA0227EN.pdf> [accessed 18 June 2019].
humanitarian aid in the long term by acting three months ahead. Every day KMD produces
before the worst effects of disasters are felt. a 24-hours ahead forecast and a forecast of
According to the FAO, for every 1 USD they average conditions for the next five days. At the
spent per household in early action in 2017, 3.5 start of every week (Monday), a seven days’
USD was saved. However, had the FAO waited average forecast is also produced.
for the drought to hit before responding with At the end of every month, a forecast for the
assistance such as restocking, the response next month is produced showing the likelihood
would have cost an additional 9 USD per of total precipitation in that month being
household6. above normal, normal or below normal (Figure
1). A month or weeks to the main rainfall
2.0 Climate and weather seasons, March-April-May (MAM) and October-
forecasts in Kenya November-December (OND), a seasonal
forecast is produced showing the likelihood of
The Kenya Meteorological Department (KMD) the total seasonal rainfall being above normal,
provides weather and climate forecasts at normal or below normal (Figure 2).
different timescales, ranging from 24 hours to

Figure 1: April 2019 monthly rainfall forecast. Source; Kenya Meteorological Department

7
‘329335905_Extreme_Rainfall_and_Flooding_over_Central_Kenya_Including_Nairobi_City_during_the_Long-Rains_
Season_2018_Causes_Predictability_and_Potential_for_Early_Warning_and_Actions.Pdf’
Figure 2: October-November-December 2018 seasonal rainfall forecast. Source; Kenya
Meteorological Department

The seasonal, monthly, seven days’ and five actions based on the forecast. To support FbA
days’ forecast bulletins indicate that these the forecast skill should go beyond the scientific
forecast should be used in conjunction with one and introduce stakeholder early actions
subsequent forecast updates issued by KMD. and associated avoidable losses and evaluate
This means users need to continuously the forecasts in terms of their potential to
monitor the forecasts provided. This provides reduce expected losses from extreme events.
an opportunity for users track the progress
of the level of certainty in an earlier forecast, 2.1 Challenges in using climate and
and use the latest information to update their weather forecasts
contingency plans.
2.1.1 Towards Forecast Based Preparedness
However, a long-standing issue with the (ForPAc) project
forecasts released by KMD is that they do not In an effort to understand why forecasts
include the skill of the forecast. In the absence produced by KMD are not effectively informing
of this, users do not know how skillful the the drought management process in Kenya, the
forecast is – for example, how many times the Towards Forecast based Preparedness Action
forecast is correct versus hence how many (ForPAc) project (Box 1) conducted a workshop
time they would be acting in vain if they took
in Kitui County, one of the Arid and Semi-Arid
Lands (ASAL) counties in Kenya. The aims of Limited technical capacity to interpret and use
the workshop were to map the current drought probabilistic forecasts in decision making.
management process and to understand the Seasonal climate forecasts, which are crucial
constraints to forecasts from informing the for planning and management of rain fed
process. agriculture with implications on food security,
are presented in form of probability, to cater
2.1.2 Challenges for the inherently chaotic nature of the
Current forecast production is not in sync atmosphere.
with drought planning and decision making
processes The probabilities indicate the likelihood of the
seasonal total rainfall being above normal,
Every year, NDMA conducts two assessments normal and below normal. The tendency is for
namely, the Short Rains Assessment (SRA) and stakeholders to consider the highest probability
the Long Rains Assessment (LRA) which aim to in decisions and actions, often disregarding
analyze and determine the impact of the Long other lower probabilities. Further, it is not
(March-May) and Short (October-December) always known what ‘normal’ rainfall means
rain seasons on food and nutrition security, as for specific areas and the implications of this
well as the food security prognosis for the next amount of rainfall. This makes probabilistic
six months. Climate forecasts could inform the forecast complex to use in decision making,
food prognosis however; the forecasts are not hence technical guidance is required to
available at the time when assessments are interpret and incorporate the forecasts in
conducted. The Meteorological Department decision making. Most decision makers have
issues the Long rains forecast towards Mid- limited capacity to build scenarios that cater for
February while the Short rains forecast is the different probabilities.
issued in September, which are times when the
SRA and LRA have been concluded (Figure 3). For example, in OND and MAM 2018, the
Further, the seasonal forecasts are only for a category that had been forecasted with a
three-month period while the food prognosis low probability of occurrence is the one that
is for the next six months thus leaving a three actually occurred while stakeholders had
months’ gap in information. planned with the category with the highest

Box 1: The Towards Forecast Based Preparedness (ForPAc) Project

The Towards Forecast Based Preparedness (ForPAc) project is a research consortium

made up of institutions in the United Kingdom, Kenya and the Greater Horn of Africa
region, including universities, national meteorological agencies such as the UK Met
Office and the Kenya Meteorological Department and humanitarian organization the
Kenya Red Cross (for full project membership see www.forpac.org)

The project aims to address limitations to taking early action on climate information.
Firstly, by improving the availability of decision-relevant climate information by
piloting research forecasts for extreme flood and drought events. Secondly, the
project is promoting the use of climate information by identifying and addressing
barriers to preparedness action in existing early warning systems in Kenya.

The project’s pilot work on drought is currently focused on Kitui county. This aims
to promote systematic early action in the operational Drought Early Warning System
(DEWS) managed by the National Drought Management Authority (NDMA). Flood
focused case studies are also being undertaken in Nairobi County and the Nzoia River
basin.
Figure 3: Schematic of annual climate, livelihoods, drought risk management and seasonal forecasts
for Kitui county, developed with stakeholders

probability of occurrence. This experience Forecast communication

has made decision makers perceive that the In an effort to get closer to users of weather
forecasts are inaccurate and cannot be trusted, and climate forecasts, KMD decentralized their
in future, this can prevent stakeholders from services to the Counties. However, the County
acting on available weather and climate officers are not autonomous and do not have
forecasts. the equipment to produce County specific
forecast hence they have to wait for forecasts
Inadequate synergies between providers and to be developed at the head quarter for them
decision makers. to tailor the forecasts to their respective
KMD provides rainfall forecasts while as counties.
decision makers are most interested in the
potential impacts of the rainfall. The need This means that the forecast will take long to
for impact information is demonstrated by reach the decision makers at the County level
indices used in drought decision making, which affects the timeliness of decision making.
such as Vegetation Condition Index (VCI) and Further, forecasts mostly do not get to benefit
Standardized Precipitation Index (SPI). Both the most stakeholders because there are often
indices are used to define the drought phases, packaged and disseminated without critical
while VCI is a trigger for funding. VCI compares consideration of who needs the information,
the current state of the vegetation to the state how best they can access, understand and
at same period over a five-year period while use it to take action. The format and language
SPI is the deviation of current observed rainfall of the climate and weather forecasts is
total from a 30-year average. Currently, KMD sometimes too difficult and technical for
does not produce impact based forecasts decision makers to understand. The channels
required by stakeholders because of limited used for communication sometimes are not
knowledge and data on drought vulnerability favorable for decision makers, email is most
and exposure. commonly used channel, it doesn’t provide
a good environment for user feedback and
consultation.
Lack of Standard Operating Procedures for understand the process and are able to provide
integrating forecasts the forecasts need for the process.
The drought management process currently Early release of a skillful October-November-
relies heavily on observed indictors and not December seasonal forecast.
climate and weather forecast hence there are Research that has been carried out by the
no standard operating procedures that spell ForPAc project, has shown that the October-
out how the forecast integration should be November-December (OND) season has
done, the specific information that can be used higher predictability as early as July (Kilavi
and who can be consulted in case the forecast et al., 2018) . In light of this, it will be useful
is not clear. The lack of standard operating for KMD to issue an OND forecast in July, so
procedures for integrating forecast information that it coincides with the Long Rains drought
in the process could be linked to the limited assessment and inform the six months’ food
technical capacity to interpret and use climate security prognosis.
and weather forecasts. Co-develop forecast with users.
Producers of weather and climate forecast
Forecast production is not in sync with need to work closely and regularly with users
budgeting process and decision makers to understand their
The production and release of seasonal evolving needs and to align these needs with
forecasts in February and September doesn’t production of forecasts. Additionally, users
correspond with the annual budgeting process and decision makers should be involved in
that starts in December and ends in June when the design of climate and weather forecasts
the budget is presented. This means money for packages and metrics of presentation. This
implementing actions is not set aside annually will ensure the content, format and language
based on a forecast. The County Emergency of the forecast are easy to understand, and it
Fund, established by the Public Finance will also contribute to better uptake and use of
Management Act of 2012, mandates 2% of forecasts.
total county government revenue for use to
manage events that threaten damage to human Build capacity for communication and
life or welfare and environment. Currently, interpretation.
these funds are only released when any kind of Producers need to enhance their technical
disaster happens. capacity to communicate weather and climate
forecasts to decision makers in a manner that is
easy to understand and interpret and does not
sacrifice the integrity of the underlying science.
3.0 Recommendations, This should go hand in hand with building
Implications and conclusion capacity of decision makers and users for
them to be able to understand, interpret and
integrate forecasts in institutional short term
3.1 Recommendations and long term plans.
Additionally, KMD should continually analyse
Collaboration between users and producers.
the skill of forecasts at all timescales and
The drought decision making process has
communicate the same to stakeholders. This
opportunities for integrating weather and
will inform users how often the forecasts are
climate forecasts. For example, the month
correct and how often they would be acting
ahead forecast could be integrated in the
in vain. This information has the potential to
monthly bulletin in order to inform the sectoral
increase the uptake and use of weather and
recommendations and food security outlook.
climate forecasts.
Stakeholders in drought management should
involve KMD in the process so that they better

8
‘Bridging Forecast Verification and Humanitarian Decisions: A Valuation Approach for Setting up Action-Oriented Early Warnings
| Elsevier Enhanced Reader’ <https://doi.org/10.1016/j.wace.2018.03.006>.
Strengthen the mandate of County Directors of
Meteorological Services.
For forecasts to reach users, especially in
Counties, in a timely manner, the County
Directors Meteorological Services in all
Counties need to be semi-autonomous. This
will give them the mandate to provide County
specific forecasts without having to wait for
the forecasts produced at KMD headquarters.
There is opportunity for this to be integrated in
the draft Meteorological Bill.

3.2 Implications on FbA

For FbA implementation there are three

components that are critical. Firstly, what
early actions can be triggered by a forecast so
as to mitigate the impact of extreme events.
Secondly, what forecast information can
be used to trigger the early action. Thirdly,
deciding whether or not it is worthwhile to act
based on the available forecast information
. The three components are dependent on
available climate and weather forecasts
hence the need enhance the production,
communication, interpretation and use of
forecasts as in the recommendations amongst
stakeholders in the drought management
process.

3.3 Conclusion

More needs to be done to advance the

understanding and use of climate and weather
forecasts in decision making. This article has
pointed to a number of key areas where specific
changes and improvements can be made to
facilitate the integration of this information
into decision making processes. However,
fostering meaningful and sustained links and
collaboration between producers and users of
forecasts is essential for promoting the uptake
of climate and weather forecasts into decision
making. Additionally, uptake of forecasts can
only improve if producers understand the
dynamics of the decision making and tailor
make products for these purposes.
 South “C” (Bellevue) | Red Cross Road, Off Popo Road
P.O. Box 40712, 00100-GPO, Nairobi, Kenya. | Tel: (+254-20) 6003593/6002465/3950000 | Fax:
(+254-20) 3950444
Mobile: (+254) 703 037000/722 206958/733 333040
Email: [email protected] | Website: www.icha.net

South “C” (Bellevue) | Red Cross Road, Off Popo Road

P.O. Box 40712, 00100-GPO, Nairobi, Kenya. | Tel: (+254-20) 6003593/6002465/3950000
Fax: (+254-20) 3950444 | Mobile: (+254) 703 037000/722 206958/733 333040
Email: [email protected] | Website: www.redcross.or.ke