You are on page 1of 18

TITLE: Effects of Seasonal Forecasts on Farm Decision Making for Small Holder

Farming systems in Semi Arid Parts of Kenya




Authors *W N Githungo
a
, R G Kinuthia
b
, K Kizito
c
, KPC Rao
d



AFFILIATION INSTITUTIONS
Kenya Meteorological Department
a
Email: william_ndegwa@yahoo.com

University of Nairobi
b
email: rgkinuthia@gmail.com;

Kenya Agricultural Research Institute
c
email: kwenakizito@yahoo.com

International Centre for Research in Semi Arid Tropics ICRISAT
d
email:

k.p.rao@cgiar.org


ABSTRACT
Impacts of adverse effects of climate change to farming communities can be reduced by
availing climate forecasts to farmers to enable making of informed farming decision and
adaptation. The needs and demand for climate information vary according to the production
systems and hence the usability of forecasts depends on the characteristics of the farmers and
their geographical location. Meteorological services in the Greater Horn of Africa (GHA)
region issue seasonal climate forecasts on regular basis as part of their operations. The lack of
a comprehensive profile of users of the forecast has left a gap between the information that is
useful to farmers and what is actually provided for public consumption. Difficulties in
interpreting and applying the forecasts include, mismatch between the variables in the
forecasts and the operational needs of farmers and simple lack of comprehending what is
entailed in the forecasts. Downscaling seasonal forecasts to local scales, promises to deliver
benefits of seasonal forecasts for smallholder farmers in the semi arid, districts of Kitui,
Mwingi and Mutomo in Kenya for improved on-farm decision-making.
Key Words: climate, downscale, information, farming, decision-making

1 INTRODUCTION
Seasonal climate forecasting can increase preparedness and lead to better social, economic
and environmental outcomes within agricultural production systems,. The production of rain-
fed crops in semi-arid tropics exhibits large variation in response to the variability in seasonal
rainfall. There are several farm-level decisions such as the choice of cropping pattern,
whether to invest in fertilizers, pesticides, the choice of the period for planting, plant
population density for which the appropriate choice (associated with maximum production or
minimum risk) depends upon the nature of the rainfall variability or the prediction of climatic
variables for a specific year. The purpose of this paper is to describe the framework used by
a team of researchers implementing the research project Managing risk, reducing
vulnerability and enhancing agricultural productivity under a changing climate to
assess the potential in the use of downscaled climate forecasts in agricultural decision-making
and to summarize what was learned in the research process. Climate forecasting is one of
many risk management tools that play an important role in agriculture decision-making.
Agrawala et al. (2001) has decried the fact that only a few examples of seasonal climate
forecasts are being used successfully by vulnerable groups, despite international efforts to
improve societal responses. OBrien et al. (2000) has indicated that one of the primary
reasons for this scenario is that forecast information does not specifically target vulnerable
groups, which results in poor availability of information and therefore a barrier to ease of
access to the information. The forecast information is often not tailored to suit target farmers
in content and delivery style, which means that they may end up not being able to access it
even if it was availed to them (Patt, 2001; Broad et al., 2002). In the Greater Horn of Africa
(GHA), region, seasonal climate forecasts are more widely available at present than they
were a decade ago. Dialogue between producers of information, researchers and different
categories of decision makers has been enhanced by outreach activities through the GHA
Climate Outlook Forums (GHACOF) and the Kenya Meteorological Departments awareness
and sensitization services. However, gaps still exist between information provided and
specific information desired by decision makers. The challenge eliminating this gap rests
with both the providers of information, who do not always understand the users needs, and
users of the seasonal predictions, whose capacity for interpretation and comprehension of
the forecasts is low. Despite the availability of relatively reliable forecasts from the Kenya
Meteorological Department, farmers seldom use this information for farm level decision-
making (Hansen, 2002; Hammer et al., 2001,). This is mainly due to lack of adaptability of
the information to the local needs and difficult in accessing the information on time and in a
format that farmers can easily understand. This gap is mainly caused by lack of capacity for
interpretation and use of the forecasts (Hansen et al, 2009). Providing location specific and
easy to understand climate forecasts can therefore enhance farmers capacity to use climate
forecasts to manage risk. Downscaled and location-specific forecasts may assist small holder
farmers in taking proper decisions at the farm level.


2 OBJECTIVES
The broad objective of the project was to develop an effective method for communicating
seasonal climate forecast information to small scale farmers.
Specific objectives include:
(i) To enhance farmers knowledge on available climate information and their ability to
anticipate extreme climate events;
(ii) To enhance farmers capacity for observing climatic parameters and their use in
guiding farm activities; and
(iii) To build farmers capacity to interpret probabilistic seasonal forecasts for purposes
of on-farm decision making.
3 LITERATURE REVIEW
3.1 Traditional Strategies to Manage Climatic Risk
From historic time, to date, the climatic considerations have always remained key to the
settlement patterns and development trend of humankind. Throughout human history,
societies have expected seasonal changes similar to the local historical averages and a certain
amount of variation around these averages, but, despite their efforts to forecast these
variations, they have not typically counted on much skill in predicting them. Thus, they have
conditioned themselves to expect climatic surprises and only hope to to deal with the impacts
reactively. The newly developed scientific skill in climate forecasting may fundamentally
change the ways social systems cope with or react to climatic variation by reducing the
magnitude or frequency of climatic surprises through provision of forecast information that
enable adequate lead time to prepare for climatic events.

Climate variability has a major influence on agricultural production in smallholder farming
systems of Kenya. Most communities in arid and semi arid parts of Kenya are predominantly
agro-pastoralists and climatic variability has a direct bearing on their livelihood support
opportunities with rainfall being the most important climate parameter. Over the years,
smallholder farmers in Kenya had their climate monitoring methods, which were passed on
from generation to another generation Nyakwada et al., (2003). The indicators used for
prediction included the environment around them such as plants, birds, animals, insects, stars,
the moon, winds, clouds, lightning patterns and heat and humidity. Indigenous knowledge
and tradition still play a role in climate risk management in Kenya. Farmers in Ukambani
community (Kitui, Mwingi, Machakos, and Makueni) of Kenya, where the project is being
implemented, have indicated up to twenty (20) rainfall indicators used at different stages of
any given rainfall season. Detailed explanations of various seasonal indicators among various
Kenyan communities are well documented by Nyakwada et al., (2003).
3.2 Seasonal Climate Forecasts
Growing understanding of ocean-atmosphere interactions and advances in modeling the
global climate systems now provide a usable degree of predictability of climate several
months in advance for many parts of the world (Goddard et al. 2001). Combined with the
ability to systematically quantify agricultural management responses via simulation analysis,
the enhanced understanding and modeling of climate issues offers an opportunity to improve
climate risk management (Meinke and Stone 2004). Integrating seasonal climate forecasting
with agricultural system analysis can increase its effectiveness (Hammer et al. 2000; Meinke
et al. 2001). The recent improvement of seasonal climate forecasts has meant that forecasts
on how much rain to expect over the season, and predictions of seasonal variability of rainfall
are now widely available. The seasonal forecast is based on the fact that lower-boundary
forcing, measured by sea surface temperatures, drives future atmospheric perturbations
(Murphy et al. 2001). These boundary conditions evolve slowly and so enable predictions of
rainfall to be made (Palmer and Anderson 1994).
Seasonal forecasts are probabilistic Fig 1 and rainfall is often forecasted as the chance of
being above normal, below normal or near normal. The normal amount of rainfall is
the middle third (tercile) of the average rainfall for the past number of years of rainfall data
used to develop the forecast. The forecast is usually issued for a period of one to three months
and suggests the probability of a given amount of rainfall expected over that period. The
forecast gives an indication of date of onset and cessation of rain. The forecast however does
not give much weight to the temporal distribution, such that if the amount of rainfall forecast
were to fall over a few days, the seasonal forecast would still be correct but the impact could
be catastrophic (Agrawala et al. 2001). It is this probabilistic nature that needs to be given
attention in dissemination of forecast information so that farmers do not assume it to be
deterministic.
In the agricultural sector, forecasts have provided information for agricultural decisions
relating to dryland farming, irrigated farming and livestock management (Marshall et al.
1996; Hammer et al., 2000). The types of decisions making that seasonal forecasts can
support include both operational short-term decisions, and tactical and strategic longer-term
decisions. A strategic decision for smallholder farmers might be to maximize total crop yield.
The accompanying operational decision might involve deciding what variety of crop to plant
in order to achieve maximum yields. A forecast for below-normal rainfall could encourage
drought-resistant seeds to be planted instead of long-maturing varieties that require more
moisture and might fail completely without adequate moisture. If the forecast provides better
than best-guess information about the rainfall in the succeeding season, it allows better
decision making and maximization of conditions (Walker et al. 2001). The opportunity to
manage variation is a strength of seasonal forecasts that is just as important as decreasing
negative impacts of climatic variability. The most useful forecast information, according to
the farmers, are the early warning on anticipated poor season, the commencement of the
season and adequacy of anticipated rains (Phillips et al, 2001). It is probable for people
living in low rainfall zones that seasonal forecasts for wetter years are of greater value than
warnings of a poor season (Phillips, 1998). Above all the forecast should be stated and
presented in a language and in terms that the target end users understand Unganai (2000)

3.3 Current State of Seasonal Climate Forecasts in Kenya
The seasonal forecast issued by the Kenya Meteorological Department is based on the
consensus seasonal rainfall forecast developed by ICPAC. The forecast is based on
considerations of the prevailing atmospheric-oceanic conditions and an ensemble of seasonal
climate forecasts generated by leading global climate prediction centres; European Centre for
Medium Range Forecasts (ECMWF), UK Meteorological Office (UKMO), International
Research Institute (IRI), and the regional IGAD Climate Prediction and Applications Centre
(ICPAC). This consensus forecast is produced as part of the GHACOF process, and is based
on discussions by many interested parties. Consensus on the long-term prospects of each
rainfall season is established through regional climate outlook fora. The fora attracts climate
experts from global and regional climate prediction centers. The seasonal climate forecasts
are based on empirical diagnostic analyses with Global Sea Surface Temperature (SST)
anomaly patterns, the sate of El-nino Southerly Oscillation (ENSO), North Atlantic
Oscillation (NAO), Indian Ocean Dipole (IOD) and upper level stratospheric winds such as
the Madden J ulian Oscillation (MJ O), being the main predictors. The forecasts are issued in
terciles i.e. below normal, normal and above normal with the probability of rainfall being in
each of three categories as the forecast. These forecasts are are normally issued for relatively
large homogeneous rainfall regions (over 9000km
2
) extending over three month periods. The
probabilities indicated are weights of expected climate categories of above normal, normal
and below normal.

The definition of the boundaries between the forecast regions is a process of this discussion.
The forecast is composed of categorical information of rainfall performance during the
season for large regions (the least is over 9000km
2
). This is indicated in terms of above-
normal, normal and below -normal categories. The categories are further described by
weights indicating the probability of occurrence of each category of forecast. This
probabilistic categorization is what users find difficult to interpret and understand. Down-
scaling the seasonal forecast is the process of interpreting the indicated probabilities into
categories of rainfall amount which users could find easier to comprehend or understand.

Current seasonal climate forecasts do not provide adequate information required for farm
level decision making. For example theres no information about the intra-seasonal character
of the rainfall season i.e. the extent of wet/dry periods within the season. The demarcation of
areas is often difficult for users located on the border line of the forecast spatial coverage
since the forecast changes drastically each time. Besides, There is limited understanding of
climate forecasting science among agricultural practitioners and extension agents and this
makes it difficult for smallholder farmers to interpret and use forecasts on their own. A
typical example of a seasonal forecast is shown in the figure 1. This covers the period
between October and December 2009. The national weather forecasts in Kenya are
disseminated through radio/TV, printed press, and the internet. Experience has shown that
radio broadcasts are the most efficient means of communicating climate forecasts to rural
communities in the country, while the Internet is least efficient method of climate forecast
dissemination.

4 MATERIALS AND METHODS
4.1 Farmers Climate Information Requirements
Assessments based on surveys and focus group discussions on stakeholder climate
information needs was carried out in the Kitui, Mwingi and Mutomo districts of semi arid
Eastern Kenya to investigate the utilization of climate information in agricultural
management. The assessments were done to determine the way the terminology used in
information dissemination and precision of the information influence decision-making. The
following were identified as information requirements for decision making in on-farm
agricultural management: Onset date for the main rains, quality of the rainy season or total
amount of rainfall, cessation date for the main rains, temporal and spatial distribution for the
main rains, timing and frequency of rainy and dry periods or wet and dry spells and
agronomic recommendations in terms of which crop varieties to grow.
4.2 Downscaling: Quantifying Probabilistic Seasonal Forecasts of Estimated Rainfall
Amounts to Local Scales
Statistical downscaling involves the application of relationships identified in the observed
climate, between the large and smaller-scale, to climate model output. It assumes that the
relationships between predictors (large-scale variables) and predictands (small-scale surface
variables) do not vary under climate change conditions. The process required to adapt model
outputs to end-user demands is complex. The FACT-FIT agro-climatological toolkit was used
for down-scaling the seasonal forecast issued by the Kenya Meteorological Department. The
FACT-FIT agro-climatological toolkit utilizes a technique based on Monte Carlo simulation
for adjusting existing mean climatologic statistical parameters of rainfall to match forecast
information. The resultant new parameters define the probability of events for the forecast
interval and are used to quantitatively perform adjustments to describe climatologically
derived probability distribution of rainfall in proportion to the forecast probabilities Barnston
et al., (2000). This technique, termed the Forecast Interpretation Tool (FIT), estimates new
distribution parameters defining the probability distribution for the forecast interval. This
distribution can be used to predict the likelihood of specific events during the forecast
interval at specific stations or for rainfall fields so long as there is a reasonable climatological
distribution. Details of the methodology and approach for seasonal estimation of rainfall are
document by Husak et al (2010). The toolkit allows the users to generate probability maps,
indicating locations that are likely or unlikely to attain critical rainfall amounts (thresholds).
Fig 2 shows downscaled rainfall amounts for Kitui, Mwingi and Mutomo districts in Eastern
Kenya for the March to May season in the year 2009. The threshold is set based on long term
climate averages of the relevant location or specific crop water requirements. The maps
generated are used to determine geographical areas that are at risk of experiencing significant
rainfall amount deficit or surplus or increased likelihood of drought or flood occurrence
depending on the areas specific climatic vulnerability. Farmers are more comfortable
interpreting indicative rainfall amounts rather than probabilistic categories of forecast.
Applying this process to the seasonal forecasts for two seasons each year in March April May
(MAM) and October November December (OND) seasons for the years 2008 and 2009,
helped to develop downscaled climate information for Kitui, Mwingi and Mutomo
districts.Tables 1 and 2 show the respective seasonal forecast and corresponding rainfall
amounts in the three districts. The downscaled seasonal forecast and identified suitable
localized agriculture practices for the season are compiled into an agro-advisory bulletin for
each location.
4.3 Integrating Seasonal Forecasting In On Farm Decision Making
The project prepared the participation of farmers in two phases: The first socialisation
phase began in late 2007 and focused on increasing farmer knowledge on climate and the
use of seasonal forecast information to develop a cropping strategy. Farmer group leaders
were trained on interpretation of probabilistic seasonal forecasts with a special ttention being
given to leaders of farmer groups who would be expected to train other farmers, especially in
geographic locations with low literacy levels. The second phase covered the year 2008 and
2009 and it mainly focused on operation and capacity-building for farmer groups to integrate
climate and forecast information into their farming activities. In order to prepare for the
implementation of the two phases, agricultural extension agents were trained by staff from
Department of Meteorology to act as intermediaries/trainers. Topics covered in the training
for the extension staff included:

1) Elements of weather and climate including the difference between weather and
climate
2) Rain formation processes
3) Understanding terminologies used in seasonal forecasting
4) Understanding probability concepts
5) Use of climate forecast information in planting strategies
6) Quantifying the economic benefits of using climate forecast information

Farmers in the project research area grow a mixture of fast maturing maize, pigeon pea, green
grams, sorghum and millet as their staple crops. The community members already have
access to the seasonal rainfall forecasts developed at the bi-annual Greater Horn of Africa
Climate Outlook Forum (GHACOF). The GHACOF forecasts are downscaled to national
level and disseminated, by the Kenya Meteorological Department through radio and print
media, with the former being the preferred mode of dissemination. The forecasts contain
categorical probabilities of rainfall estimates for the two crop growing seasons in the study
area: long rains season (MarchMay- MAM) and short rains season (OctoberDecember-
OND). These categorical estimates are in the form of probabilities for rainfall totals falling in
the ranges of below normal (a range defined by the 10 driest of the past 30 seasons), normal,
or above normal (a range defined by the 10 wettest of the past 30 seasons). Beginning in
February 2008, a series of seasonal participatory climate forecast workshops in each project
site were held, designed to assist the group of farmers in each study site to better understand
the forecast and to be able to apply it to their farm management decisions. The workshops
took place in a location chosen by the farmers but always were held in the homes each of the
farmers on rotation basis. Often the meeting lasted 2 -3 hours, and was conducted in English
and Kiswahili and interpreted into the local language. The workshops followed a common
format, designed to assist farmers in applying the forecast information:

They were were asked to comment on the previous seasons rainfall data, and whether it
was in tandem with their recollection of the forecast.
The forecast performance was measured against rain-gauge records on farmers fields and
compared with farmers own historical knowledge of local rainfall quantities and
estimates of actual ranges of rainfall.
Farmers were then asked to comment on the success of their management practices in the
past year, given the rainfall received.
Farmers were asked to offer their insights into the coming years rainfall, based on their
interpretation of local traditional rainfall indicators.
The forecast for the coming season was explained to farmers, in terms of the probabilities
for below-, about-, and above-normal rainfall levels.
The forecast was downscaled, using the FACT-FIT agro-climatological toolkit.
The information used to generate the forecast was explained in simple terms and
questions were encouraged, including a discussion on El Nino phenomenon.
And finally, a discussion was facilitated between farmers and the local agricultural
extension service officers on the appropriate farm management practices for the coming
year, taking into account the forecast, the local indicators and available planting seed.
4.4 Traditional practice and Prudent Preparation and Decisions of a Growing Season in
Kitui, Mwingi and Mutomo Districts
Farmers decisions for various farm operations in the three study districts is so much
influenced by the conventional local practices in the region for each of the two seasons - long
rains season or the short rains season. There exists knowledge on potential actions/responses
and approaches for each of the two seasons among the locals in the study area
4.4.1 Traditional Allocation of Resources in Different Seasons
Farming Practices

During the March, April May (MAM) season, farmers concentrate on drought resistant and
fast-maturing crops such as green grams, maize and also tend to increase the herd size
livestock. The acreage under crop is low compared to the short rains season. Onsets of the
rains determine the crop species and variety to be sowed, while planting depths are also
determined by the amount of rainfall received before planting.

During the October December season (OND), the most reliable rain season, there is
increase in acreage and planting of late maturing crops and agro-forestry species. The
converse is true. Should the farmers anticipate bad season, based on traditional indicators,
they plant less acreage and resort to fast-maturing crops like in MAM season. Crops for this
season include maize, pigeon peas, beans, dollicos, cotton, pumpkins, and establishment of
pastures restocking of livestock and spraying against ticks.
Soil/water conservation
During OND season the main soil and water conservation activity is rehabilitation of water
harvesting facilities, while during MAM season farmers in Kitui and Mutomo districts
concentrate in establishment and construction of water harvesting structures. In Mwingi
district, soil/water conservation establishment is done in J une-early October. In some areas,
soil/water conservation works is done in J an-Feb
Nutrient management
During J une J uly August September (J J AS) season application of manure in preparation for
OND season is the main nutrient management activity. Use of inorganic fertilizers is
practiced during actual planting in OND season.

In general, the traditional practices used as the baseline measurement for prudent decisions by
farmers and which were considered to be highly influenced by climate information as
analyzed in this paper include: (i) Choice of crop and variety, where farmers would be
expected to invest in crops which optimize output in good seasons or minimize risk in low
rainfall seasons (ii) Level of investment in land, where farmers are expected to
increase/decrease land under crop during wet/dry forecast (iii) Investment in farm input,
where farmers make choice concerning farm iputs that are sensitive to climate information
such as seed species and variety and soil and water management practices. The observed
trend is increase in use of chemical fertilizer during wet season and decrease of the same
during anticipated dry season and increase in use of manure during anticipated dry season and
decrease in use of the same during anticipated wet season, and (iv) Acqusition of Credit,
where farmers would be expected to go for financing of farm inputs during good seasons.
Only one group (Kaveta FFS) had an internal credit facility mechanism which according to
group by-laws was only accessed for farm related activities, and the same was monitored.
5 RESULTS & DISCUSSIONS

The use of forecasts was associated with varied decisions made by individual farmers results
of which have been analyzed below for each season.
5.1 March-April-May season 2008

This was the first season of the project to tests farmers capability and acceptance of use of
season forecast information in on-farm decision making. Indications show that only 12% of
the farmers in Kitui and 8% in Mutomo used the decisions suggested in the agro-
meteorological bulletin developed by the research team. The season forecast indicated normal
to below-normal rainfall, yet several farmers, did not show signs of utilizing drought resilient
crops or varieties. Fig 3 shows the land utilization for the various seasons, low land
utilization was recorded during the March May 2008.
5.2 October-November-December season (OND) 2008

This was the second season of trials of the integration of seasonal forecasts in on-farm
decision making. Forecast indication showed normal to above normal rainfall. Farmers were
enthusiastic and went out and increased land under crop and invested in seeds and fertilizer
Fig 4 shows the mount of fertilizer used by farmers in Kitui district, while Fig 5 shows the
amount of fertilizer used by farmers in Mutomo. The season was however disappointing as it
turned out that the actual temporal distribution of rain was bad and the season was
characterized by long dry spells which caused water stress on the crops resulting in crop
failure.
5.3 March-April-May season (MAM) 2009

The MAM 2009 season was the third season of the project in the field. The meteorologist had
a hard time explaining the outcome of the previous season, and also gaining confidence of
farmers given the failed forecast of OND 2008. However this was one of the seasons which
had greater success in risk management actions among farmers. Forecast indications of
below-normal rainfall were well interpreted in the downscaled rainfall amounts via the
FACT-FIT toolkit. It was clear among the participating farmers that the season was going to
be dry and probably a severe drought. Farmers responded positively to the forecast
information by restraining investments, and went for low risk investments such as less land
on crop, use of farm yard manure in place of chemical fertilizer. Farmers in Mutomo
increased the amount of land on millet and sorghum which are more drought tolerant crops.
5.4 October-November-December season 2009
The OND 2009 season was greatly pronounced over the media as an El Nino year. Farmers
eagerly awaited the seasonal forecast information from the meteorological department to
enable them confirm their expectations. The forecast indication of near-normal to above-
normal rainfall for all the locations of the study sites was well received by the farmers.
Farmers reacted positively to forecast and invested greatly in land, labour and farm inputs.
Conspicuous during this season was the heavy investments in chemical fertilizer, unlike in all
the previous seasons.

Farmers expected huge gains from their farm activities. Most farmers went for greater
investments in form of high yielding crop varieties and hybrid seeds. There was a problem
however in the distribution of rain in Mwingi and Mutomo districts, with long dry spells in
between rainfall episodes causing water stress on crops.

Credit is one aspect of agriculture which opens up financial risk. Fig 6 shows the credit
acquisition of farmers in Kitui. Many authors have regarded farmers as risk averse. Credit is
the most liquid investment a farmer may make, however despite the perceived risk, credit
acquisition among the farmers in Kitui have also been observed to follow a similar trend as
responce to climate information. Approximately 87% of the farmers in Kitui, acquired credit
during the OND 2009 season when the forecast indicated above normal rainfall influenced
by El Nino phenomena. It is a big deviation when compared to the previous seasons when
only 24% acquired credit in OND 2008 when there was an indication of above-normal
rainfall. The fact that the above-normal forecast was associated with El Nino may have
motivated farmers to react positively.
6. CONCLUSIONS
Seasonal forecasts benefit farmers most when they give them the opportunity to take
advantage of good weather conditions. Downscaled seasonal forecasts can make the
difference between optimizing on climate opportunities and risk management in arid and
semi arid areas. Final evaluation of farmers confirmed long held opinion that integrating the
effectiveness of climate and forecast information into their cropping strategies had not been
effectively achieved. The key challenges identified were translating climate information into
user-friendly language for farmers and integrating this into effective adaptation to climate
change and variability. Utilization of land during the four seasons of the project had some
distinct impressions of forecast guided decision making. Forecast information had greater
influence on decisions of amount of land put on various crop types and seed varieties.
Seasonal forecasts benefit farmers most when it provides opportunity to take advantage of
climate conditions. Onset of rain and expected rainfall amounts is most important information
for arid and semi arid lands. Downscaling climate information may be expensive, but it
provides an informative approach for seasonal forecasts and farm-level decision making.
Capacity building among farmers and extension agents on interpretation and use of seasonal
forecast is necessary in order to reap the benefits of seasonal forecasts.



7. ACKNOWLEDGEMENT

The project Managing risk, reducing vulnerability and enhancing agricultural productivity
under a changing climate is supported by the Climate Change Adaptation in Africa (CCAA)
program, a joint initiative of Canadas International Development Research Centre (IDRC)
and the United Kingdoms Department for International Development (DFID). The views
expressed in this paper are those of the authors and do not necessarily represent those of
DFID or IDRC.


























8 REFERENCES
Agrawala S, Broad K and Guston D H (2001) Integrating climate forecasts and societal
decision making: challenges to an emergent boundary organization Science,
Technology and Human Values 26 45477
Barnston A, He Y, Unger D. 2000. A forecast product that maximizes utility for state-of-the-
art seasonal climate prediction. Bulletin of the American Meteorological Society
81(6): 12711279.

Broad K, Agrawala S. The Ethiopia food crisis: uses and limits of climate forecasts. Science.
2000;289:16931694.

Goddard L, Barnston A, Mason S. 2003. Evaluation of the IRIs net assessment seasonal
climate forecasts: 1997-2001. Bulletin of the American Meteorological Society
84(12): 17611781.
Gregory J . Husak,a J oel Michaelsen, Phaedon Kyriakidis, J ames P. Verdin, Chris Funk, and
Gideon Galu (2010) the forecast interpretation tool a monte carlo technique for
blending climatic distributions with probabilistic forecasts, International J ournal of
Climatology
Hammer, G.L., Hansen, J .W., Phillips, J .G., Mjelde, J .W., Hill, H.S.J ., Love, H.A., &
Potgieter, A.B. (2001). Advances in application of climate prediction in agriculture.
Agricultural Systems, 70, 515553.
Hammer G L (2000) A general systems approach to applying seasonal climate forecasts in
Hammer G, Nicholls N and Mitchell C eds Applications of seasonal climate
forecasting in agricultural and natural ecosystems the Australian experience
Kluwer Academic Publishers, Dordrecht 5165
J ames W. Hansen. Realizing the potential benefits of climate prediction to agriculture: issues,
approaches, challenges. Agricultural Systems 74 (2002) 309330
Meinke, H., Baethgen, W.E., Carberry, P.S., Donatelli, M., Hammer, G.L., Selvaraju, R. and
Stckle, C.O., 2001. Increasing profits and reducing risks in crop production using
participatory systems simulation approaches. Ag Systems, 70: 493-513.
Meinke, H. and Stone, R.C., 2004. Seasonal and inter-annual climate forecasting: the new
tool for increasing preparedness to climate variability and change in agricultural
planning and operations. Climatic Change, in press.
Nyakwada W- Ocholla A B C, Mango C, Onyango M, . Musembi D.K, Nyakwada C
(2003), The traditional rainfall indicators used to monitor and predict rainfall in
Kenya, Drought Monitoring Centre, Nairobi, DMCN Report
OBrien K, Sygna L, Nss L O, Kingamkono R and Hochobeb B (2000) Is information
enough? User responses to seasonal climate forecasts in Southern Africa CICERO
report no. 200003
Palmer T N and Anderson D (1994) The prospects for seasonal forecasting a review paper
Quarterly J ournal of the Meteorological
Society 120 75593
Patt A, Gwata C. Effective seasonal climate forecast applications: examining constraints for
subsistence farmers in Zimbabwe. Global Environ. Change. 2002;12:185195.
doi:10.1016/S0959-3780(02)00013-4
Phillips, J .G., 1998: Comparing Optimal Use of ENSO Forecast Information in two
socioeconomic contexts. Invited Talk at the American Association of Geographers
Annual Meeting, Boston, MA, March 1998.
Phillips, J .G. et al, 2001: Current and Potential Use of Climate Forecasts for Resource-poor
Farmers in Zimbabwe. American Society of Agronomy, Crop Science Society of
America, and Soil Science Society of America, Madison, USA. Impact of El Nino and
Climate Variability on Agriculture. ASA Special Publication no. 63
Roncoli, C, C J ost, P Kirshen, M Sanon, KT Ingram, M Woodin, L Some, F Ouattara, BJ
Sanfo, C Sia, P Yaka and G Hoogenboom (2009). From accessing to assessing
forecasts: an end-to-end study of participatory climate forecast dissemination in
Burkina Faso (West Africa). Climatic Change 92:433-460.
Unganai, L.S., (2000): Application of long range rainfall forecasts in agricultural
management: a review of Africas experiences. Proceedings of the International
Forum on Climate Prediction, Agriculture and Development, International Research
Institute for Climate Prediction
Wilks DS. 2000. On interpretation of probabilistic climate forecasts. J ournal of Climate
13(11): 19651971.

Ziervogel, G (2004). Targeting seasonal climate forecasts for integration into household level
decisions; the case of smallholder farmers in Lesotho. The Geographical J ournal, Vol.
170, No 1, pp 6-21 18 ibid


















Table 1 Categorical Forecast and Corresponding Downscaled rainfall amounts for Kitui

SEASON Forecast
category
Category of
Estimated rainfall
Amounts for the
Season
Probability of
Occurrence of
indicated
amounts
Actual Rainfall
amounts received for
the Season
(as monitored by
farmers)
MAM
2008
normal to below
normal
240 -280mm 70% 80% 340.3
OND 2008 Near normal to
above normal
280 -300mm

40% - 50% 227
MAM
2009
Below normal 150 240mm

60% - 70% 203.6
OND 2009 Near normal to
above normal

260 -300
20% - 40% 596

Table 2: Categorical Forecast and corresponding downscaled rainfall Amounts for Mutomo

SEASON Forecast
category
Category of
Estimated rainfall
Amounts for the
Season
(downscaled by
FACT-FIT)
Probability of
Occurrence of
indicated
amounts
Actual Rainfall
amounts received
for the Season
(as monitored by
farmers)
MAM
2008
normal to
below normal
100 -200mm 10% -30% 140.3
OND 2008 Near normal to
above normal
260 -280mm 40% - 60% 341.4
MAM
2009
Below normal 200 -260mm 20% - 40%

70.7
OND 2009 Near normal to
above normal
240 -280mm 50% -60% 291.6


Figure 1 GHA Seasonal Climate Forecast MAM 2009











Figure 2 Downscaled seasonal forecast Kitui District MAM 2009

Figure 3 Land Under Crop in Mutomo











Figure 4 Amount of Fertilizer used by Farmers in Kitui





Figure 5 Amount of Fertilizer used by Farmers in Mutomo





Figure 6 Proportion of Farmers Seeking Local Credit in Kitui

You might also like