UKRC NERC funded

University of Reading
CARD, University of Malawi

ScSh 2018-04-12 at 15.37.38.png

SATWIN-ALERT Moving beyond rainfall in drought insurance design

The Satellite data for Weather Index Insurance-AgricuLtural EaRly warning system (SatWIN-ALERT) programme is designed to support the insurance industry in moving beyond rainfall totals when developing drought insurance insurance contracts.  We are working with a range of operational insurance programmes across Africa to:

  1. Design and co-develop with stakeholders a practical framework for managing basis risk, which can be incorporated into well-established and trusted systems.
  2. Exploit new Earth observation datasets, including remotely sensed soil moisture, probabilistic rainfall estimates and new seasonal forecast technologies for improved early warning of drought, and subsequent prediction of drought related hazard to agriculture.
  3. Utilise and assess novel participatory methods for assessing farmer perceptions of basis risk, and its effect on decision making, especially the impact that crowdsourcing might have on insurance design in Malawi.
  4. Work with partners put in place operational and sustainable long term solutions for anticipating meteorological basis risk - in the form of scripts that can be run within or alongside trusted systems.

The social impact of weather insurance for agriculture

CCAFS funded

Website here



Exploring the social impact of weather insurance

Click here for more

Scaling-up index insurance, or assessing its impact is not a simple matter: 

  • At a farm or community level, insurance must fit into the ever-shifting landscape of culture, context and farmers’ existing risk management strategies.  Agricultural insurance is able to reduce poverty traps and allow farmers access to new agricultural technologies, but it equally has the power to entrench existing inequalities, leaving the most powerless even worse off.   
  • People also have many complex reasons for buying insurance that are influenced by marketing and delivery as much as the product itself.  If one wants to use insurance to achieve some goal (such as increasing the uptake of climate-adapted seed), then it is often necessary to have a holistic understanding of the farming system, rather than just studying an insurance product in isolation.   
  • There must be collaboration between farmers, finance experts, global reinsurers, meteorologists, agronomists, banks, seed companies, technology experts and many others for an agricultural insurance program to scale and reach greater numbers of farmers.  These stakeholders do not often meet, believe that they speak different ‘languages’, and may have competing (and conflicting) expectations about what an insurance program is trying to achieve.  These competing expectations can easily lead to missed opportunities or unintended consequences.

I am specifically leading the aspects on business models and impact assessment, working with Pula Advisors to map the insurance landscape of Malawi and to use tools like R-shiny to visualise that data.


CCAFS Funded

Website here

SIREC, University of Ghana


From satellites to gender dynamics for Ghanaian agricultural insurance

Click here for more
(Dr Susana Alo interviewing potential insurance customers in North East Ghana)

CASCAID aims to bring state of the art climate information tools to agricultural decision making across West Africa.  This includes building on agri-business relationships, understanding how crowdsourced data can transform agricultural value chains, pulling climate information into agricultural extension via the wonderful PICSA programme and creating platforms to exchange climate information.
My aspect of the programme has been to work with Ghanaian Meteorological Service and the African Institute of Mathematics to validate and understand the fitness for purpose of different satellite rainfall products in Ghana.  I have also been working with sociologists at the University of Reading to create a qualitative toolkit to explore the gender dynamics of insurance purchases in the northern part of the country.

Towards a Global Flood & Flash Flood Early Warning Early Action System Driven by NASA Earth Observations and Hydrologic Models


University of Chicago

ScSh 2018-04-12 at 16.31.28.png

NASA GEO Linking forecast and response for flash floods

(article here & see below for the story that inspired this project)

In January 2015, heavy rainfall over Lake Malawi basin led to high water levels and subsequent flooding downstream on the Shire River. As levels approached bank-full, persistent convective rainfall occurred over southern Malawi, setting all-time daily rainfall records, causing urban flash floods in Blantyre. This led to a complex and evolving scenario for disaster across the nation.

The riverine floods were better forecast and monitored and as a result, a snapshot of the full disaster situation was incomplete and the majority of initial resources (food, medical items) were distributed amongst the districts impacted by riverine floods, even though areas impacted by flash floods had more dire conditions (Kruczkiewicz and Cen 2015). Of the 300 casualties, the majority lived in areas impacted by flash floods.

It is notable that several forecast products were available, in addition to heavy rainfall warnings from several sources, but as most of the protocols were set up to deal with the longer term flooding, it was hard to use these effectively.

We feel the initial response to the Malawi-Mozambique floods could have been more efficiently allocated if disaster management protocols had been conditioned on flood type and forecast characteristics, potentially resulting in decreased impact on lives and livelihoods. Furthermore, a district or regional exposure map for various types of floods would have been useful to indicate where populations may be more resilient to particular types of floods, informing planning and prioritization of disaster response resources.

Flash floods are often left behind in the discourse about early warning.  They are often grouped together with slow-onset floods (riverine for example), leading to operational protocols that are insufficient (send boats!) and a disconnect between flood forecasts and flash flood impacts.  At the same time, there is also a building disconnect between flood impact data and response (NGOs, government agencies) vs the hydrological data and monitoring of the floods themselves.  Including flood impact data in forecast evaluation would lead to improved metrics and better link forecast products to response protocols.

This project aims to address both of these issues with a 5 element solution

  1. Develop a global flash flood dataset & maps of flash flood risk, vulnerability and exposure: We will collate traditional and ‘new’ flash flood databases (impact databases from NGOs etc) into a new dataset, which includes user driven map attributes such as new variables (e.g. cost/deaths/ vulnerability) and an examination of appropriate spatial techniques for mapping flash floods.
  2. Stakeholder engagement to co-develop matrices outlining the relationship between forecasts and preparedness actions: A Technical Working Group (TWG) will be established to include private sector, humanitarian, government and academic representation. The TWG role is to first identify existing Standard Operating Procedures (SOPs) for floods, then to explore necessary constraints for a flash flood Early Warning Early Action.
  3. Intercomparison of flash flood forecasts, with hazard parameters, evaluation metrics and lead time directly informed by the disaster managers: The first global flash flood model intercomparison and validation will be performed, utilizing the user defined metrics & case studies from E2 and the new observational dataset from E1. The outputs of this work will be stored in an Open Data Commons format.
  4. Co-development and real-time demonstration of an EWEA framework for flash floods: We will formally reconvene the TWG to present the results of the intercomparison for the tailored case studies to our end-user community, working with them to explore suitability for forecast based action and potentially develop SOPs.

We will be focusing on two case studies: one in the Rohingya refugee camps in Bangladesh (which has exceptional flash flood vulnerability) and one to be determined in Latin America.