Semiarid areas in the ECA region are characterised by high climatic variability.  This is especially manifest in  the  erratic  nature  of  seasonal  rainfall  with  respect to onset,  quantity,  distribution  and  cessation. This constitutes a major constraint to decision making by rainfall dependent smallholder farmers, with respect to production  management.    This  coupled  with  the  fickle  nature  of  the  season  hampers  increased  crop productivity leading to food insecurity.  Climate change-induced variability has only made a bad situation worse by causing shifts in seasonality accompanied by new abiotic stresses. Most agricultural production systems  in  the  ECA  region  are  rain-fed  and  predominantly   subsistence  in  nature,  supported  by  poor management  practices,  hence  is  vulnerable  to the  impacts  of  climate  variability.    Often,  the  smallholder farmers  are  the  ones  mostly  dependent  on  economic  activities  that  are  sensitive  to  the  climate.  Despite agriculture  being  the  mainstay  for  nearly  70%  of  the  region’s  population,  it  continues  to  remain underdeveloped due to inadequate adoption of yield enhancing technologies in almost all countries in ECA.

Climate-responsive  management  techniques  are  likely  to  improve  crop  productivity  in  the  region  and would in turn broaden the scope of product range for rain fed agriculture as each tactical response would target  specific  enterprises,  value  chains  crop  varieties.    The  project  which  was  implemented  in  seven ASARECA member countries - Kenya, Ethiopia, Madagascar, Sudan, Eritrea, South Sudan and DR Congo was  implemented  in  three  levels;  testing  of  technologies  and  appropriate  response  options,  validation  of previously  identified  options  and  upscaling  of  proven  technologies  and  corresponding  response  options. The  implementation  levels  varied  from  one  country  to  the  other  depending  on  previous  research  work conducted in those countries. This is a description of the upscaling level which was implemented in Kenya.

The use of climate information for farm level decision making has in recent years been adopted as a way to address the problem of climate variability and change. The  seasonal rainfall forecast given by the national meteorological agencies has particularly gained prominence among livelihoods practitioners for purpose of informing pre-season activities such as choice of crops and seed varieties, determination of planting dates, livestock stocking and destocking, feed preparation among others. This type of  information  has  however been  available  to  few  persons  while  many  members  of  communities  have  not  had  access  to  such information. The main impediment being lack of access to relevant, understandable and timely information mainly due to lack of capacity and the analogue methods (print media and traditional radio and TV), used to disseminate such information. The development of seasonal forecast informed agro-advisory was earlier tested and validated in Kenya and has continued to give impetus and relevance to the seasonal forecast in the agriculture sector. Over the last few years, many stakeholders and institutions including the ASARECA have  endeavoured  to  ensure  availability  of  relevant,  understandable  and  timely  climate  information  for agriculture especially  in the rainfed arid  and semi arid regions. This  has seen the  increased  need  for the
seasonal climate  information  in the agriculture sector in  Kenya, particularly  for  informing agro-activities during the main crop growing seasons in country.  The response arming approach, which was developed through  action  based  research  and  adaptation  learning  process  has  been  identified  as  an  appropriate methodology  for  linking  seasonal  climate  information  to  local  scale  rainfed  agriculture  enterprises  and other climate dependent environmental livelihoods.

Upscaling Approach for Response Farming

In  Kenya, the usefulness of advisory  had earlier  been  evaluated  in previous project (Making the  Best of Climate:  Adapting  Agriculture  to  Climate  Variability),  in  a  range  of  climatic  conditions.  This  research work focused on scaling up of response farming approaches with an overall aim of developing and testing a system for timely generation and dissemination of agro-advisories to farmers and enhancing the capacity of farmers and extension agents in understanding and utilizing that information. 

Valuable  climate  and  agriculture  nowledge  and  technology  data  and  information  was  compiled  and systematically  organized  to  create  an  information  base.  A  database  of  local  climate  and  corresponding agro-technology data sets as required for characterizing and quantifying impacts and potential responses to climate variability on agricultural  systems was developed. The database was developed to be  interactive and contain local scale location specific knowledge on local climate, value chain and relevant agricultural technologies  which  have  been  developed  to  address  specific  impacts  of  climate  variability  and  extreme events. The  local  scale datasets was  specific  for  locations (ward,  locations, sub-locations or village), for each  of  the  locations  in  Kitui  County.  The  climate  datasets  were  organized  to  accommodate  scenarios based on the three normative rainfall categories of (Above normal, near normal and below normal), often used to describe the seasonal  forecast issued  by the  Kenya Meteorological Service. The scenarios  in the database      further  were  arranged  to  capture  the  two  main  rainfall  seasons  in  Kenya,  long  rains,  March, April, May and short rains, October, November, and December. The local scale seasonal forecast for each season,   form one component of the database. The climate component of the database contain two sections, the  normative  sections  containing  information  on  historical  averages  and  trends  and  the  second  section containing future climate information. The section of future climate information was updated on a seasonal basis, based on the seasonal rainfall forecast issued by  KMS. The section with future climate information contain datasets with expected onset date of seasonal rainfall , normative indication of expected seasonal rainfall,  warning  of  extreme  events,  drought,  floods,  storm s,  ground  frost,  high  temperatures    etc.  , distribution  (spatial/temporal  )  of  seasonal  rainfall  poor/good,  cessation  date  of  seasonal  rainfall.  The agriculture component of the database is organised according to the types of adaptive technologies existing. Guided by available literature the relevant agriculture technologies and recommended actionable options of climate sensitive agriculture practices was identified for inclusion into the database.  The identification of the technologies was done through literature review, collaborated with expert opinion from scientists and practitioners. 

These  agriculture  technologies  and  recommended  actionable  options  of  climate  sensitive  agriculture practices was identified at local scales (ward, location, sub-location or village – lowest administrative area in  Kenya)  and  corresponding  to  respective  agro-climatic  zone.  The  agriculture  technologies  and recommended  actionable  options  of  climate  sensitive  agriculture  practices  further  categorised  into respective  climate  sensitivity  in  accordance  with  the  norm ative  seasonal  rainfall  forecast  categories  of above normal, normal and below normal rainfall.

The  development  of  a  database  of  relevant  location  specific  agriculture  technologies  and  recommended actionable options of climate sensitive agriculture practices was done in a participatory process comprising stakeholders  from  across  value  chains  and  relevant  insti tutions.  The  development  of  components  of  the datasets of technologies and recommended actionable options of climate sensitive agriculture practices of agriculture  activities  was  guided  by  the  normative  rainfall  and agro-ecological  zonings.  These  options consider general impacts of season climate on agriculture for specific locations in each sub-county/ward. The datasets were categorised to include specific scenarios and technologies  for land preparation (tillage types, water conservation, terrace types etc);, specific scenarios and technologies  for crops management, (crop types and varieties: cereals- Maize, sorghum, millet - planting date, pulses- beans, pegion peas green grams cow-peas, dolichos etc);, specific scenarios and technologies for soil fertility (fertilizer application, farm  yard  manure  application  etc);  specific  scenarios  and  technologies  for  water  harvesting  (runoff  
harvesting, insitu water harvesting). The database was developed such that the agriculture technologies and recommended actionable options of climate sensitive agriculture practices would easily be matched in an interactive setting with corresponding climate type and scenario in the climate component of the database. This categorization of technologies enables the system to match the forecast seasonal rainfall potential with corresponding response options of agriculture technologies and recommended actionable options of climate sensitive  agriculture  practices  and  thereby  develop  corresponding  location  specific  seasonal  agroadvisories. This arrangement of the data base ensures corresponding agriculture technologies designed for specific  climates  are  fitted  accordingly  in  the  database  such  that  locally  relevant  agro-technologies  are
matched  with  respective  climate  scenario.  The  information  in  the  database  represents  climate  response options  for the forecast rainfall  following the two rainfall seasons (March- May  and Oct-Dec) in Kenya. When fully developed, the database makes a vital tool for automation of the seasonal scenario and agroadvisory development process. 

The  project  developed  an  IT-based  system  with  the  aim  of  improving  the  efficiency  of  preparation  and dissemination of location specific weather based agro-advisory. The IT system is computer based software which  enables  the  effective  and  systematic  interaction  of  various  components  of  the  database  described above. This automated interaction is used to match the seasonal climate forecast data with corresponding agriculture  technologies  and  recommended  actionable  options  of  climate  sensitive  agriculture  practices, built in the database. 

Every growing season, the seasonal forecasts issued at national level by the KMS was downscaled to local scales  before  the  seasonal  forecast  data  is  entered  into  the  database.  The  automated  system  match  the corresponding  agriculture  technologies  and  recommended  actionable  options  of  climate  sensitive agriculture  practices,  first  interpreted  for  the  local  scales  (sub-locations).  The  automated  IT  system  was used  to    integrate  respective  local  scale  seasonal  forecast  and  the  corresponding  local  scale    agriculture technologies and develop a  Weather Based Agro-Advisory  for the respective  season either  March April May  (long  rains)  and  Oct  Nov  Dec  season  (short  rains)  for  the  local  area(sub-location).  The  system therefore  provide  the  process  of  effective  automated  devel opment of  the  Weather  Based  Agro-Advisory and  provide  timely  and  cost  effective  availability  of  actionable  options  for  use  at  farm  level  during  the respective season. 

Figure 3 Sample of weather based agro-advisory message sent on SMS

A computer based tool was developed to enable effective dissemination of the developed agro-advisories.
For purpose of wide coverage and upscaling of information, the dissemination was done on SMS and email
to  all  persons  and  stakeholders  whose  mobile  phones  and  email  contacts  were  registered  with  the
automated  system.  The  dissemination  utilizes  a  computer  based  internet  (multimedia)  communication
system. The advisory messages were sent to recipients as text messages. The system   utilize a multimedia
approach  which  enables  the  system  to  overcome  the  challenge  of  limited  characters  (140  number  of
characters allowable in text messages in mobile phone), associated with dissemination of text messages on
mobile phones.

Achievements  
The enabling of automation of the dissemination process of agro-advisory involves first selection of users
of the agro-advisory  from all  locations of  Kitui  County.  The persons chosen  include community  leaders,
chiefs, county/sub-county/ward/village administrators, heads of NGO and CBO, agriculture extension from
all  locations  in the county and  leaders of  farmer groups. The leaders were registered with the automated
system such that the system would identify them with their locations institutions and their mobile phone
numbers. At least 3500 persons from each of the 230 sub-locations/villages, within the 40 wards and 8 subcounties of Kitui County, benefitted by receiving the information. These linkages act as intermediaries for
dissemination of the information to the communities. The linkage when completed   provides a network of
at least 15 persons in each village who have access to the automated dissemination of SMS based seasonal
climate information and agro-advisory. This approach enabled community level outreach and access of the
climate  information  in  Kitui  county.  Over  200,000  households  in  Kitui  County  received  appropriate
seasonal  climate  information and  appropriate  local  level  agro-advisory to match the season. Most of the
beneficiaries (> 70%) of climate  information were women,  youth and persons with disabilities. To a big
extent, this information enabled over 450,000 farmers in Kitui  County to prepare their farms and invested
accordingly  in readiness  for Oct-Dec 2013 rainfall  season. Despite the low amounts of rainfall recorded,
farmers were able to achieve sufficient food security levels in that season all attribute to the wise decisions
they made informed by the seasonal forecasts and agro-advisories they received on their phones. 

Quote from Beneficiaries (Mary Mueni Kyome Village Kitui County) ‘This season,I received the forecast on my phone. Use of climate information has eradicated poverty from my house. It has not only saved me from perpetual dependence on food relief but has also given me a stable source of income with which I am able to pay school fees for my children.

Use of climate information in managing my farm has eradicated poverty from my house. It has not only saved me from perpetual dependence on food relief but has also given
me a stable source of income with which I am able to pay school fees for my children.

Insitu water harvesting decisions in the semi arid
regions which require seasonal climate information

The project beneficiaries included different gender categories (men, women, youths, persons with disability
and elderly persons). Specifically, the project improved  the food security, income and livelihoods lives of
over  3500  persons  (1800  women,  900  men,  600  youth,  200  persons  with  disabilities  and  35  elderly
persons) who all were community leaders and received information directly on the phones or emails and
also  acted  as  intermediaries  to  pass  on  information  to  others.  The  project  also  benefited  over  500,000
persons mainly women and youth who received information  from the intermediaries in all 230 villages of
Kitui County. 

Key Lessons

• A farmer having prior information on rainfall potential through forecasting helps to optimize farm level  plans  or  otherwise,  in  case  rainfall  arrival  is  early  or  late  respectively,  This  also  helps  to maximize benefits from good rains, and stabilize yield in case season turns bad.  
• The  automated  weather  based  agro-advisory  is  effective  when  information  is  sent  to  farmers  on their mobile phones. The approach needs to be improved for  replication in other arid and semi arid regions.

Challenges  
1.  Limited capacity among stakeholders to translate forecast into advisories that can easily be taken up by farmers.  
2.  The seasonal dependence nature of project activities did  not allow for complete seasonal cycle of researchable concepts.

Recommendations  
1.  Response farming is not a new concept, but it’s a kind of reimagining farming with a new level of thinking and ‘reengineering of the already existing farm management practices differently.
2.  The automated weather based agro-advisory information sent to farmers needs to be backed up by intensive capacity development among farmers and support stakeholders.  
3.  The database used to develop the automated weather based agro-advisory needs to be up-dated regularly to incorporate new and upcoming technologies and  also include all other data types including crop and livestock relevant technologies.