top of page

EOPAM - A new drought index-insurance model for African pastoral regions - phase 2

Updated: Apr 8, 2019

We aim at supporting African Risk Capacity developing and implementing new models for drought index-insurance targeting the extensive pastoral areas of Africa.

 

RESEARCH OVERVIEW

Existing approaches in ARV for agricultural drought monitoring are based on the Water Requirement Satisfaction Index (WRSI), an indicator of crop/pasture performance related to the availability of water to the vegetation during the growing season and based on satellite rainfall estimates (RFE). Although WRSI can be parametrized for grasses, it was not specifically targeted at extensive pastoral areas, possibly limiting the ability of the model to accurately assess drought-related impacts on livestock and pastoralist welfare.


During the first phase of the project ILRI, with technical support from JRC and the University of Twente researchers, has developed for ARC a specific methodology to determine a pastoral index based on the Normalized Difference Vegetation Index (NDVI) satellite imagery. NDVI, which is a proxy of green biomass/vegetation cover, is currently used in several risk management initiatives for operational drought monitoring in Africa (e.g. FewsNET-USAID, NDMA-Kenya, ASAP-JRC, VAM-WFP, etc.), including index-insurance over pastoral areas in East Africa (www.ibli.ilri.org). The work acknowledged that NDVI is overall a reliable indicator of vegetation condition for regional scale monitoring and outperforms the currently used index in ARC for drought detection in African rangelands. However, to actually provide an insurance product to cover drought risk in the pastoral areas of


PROPOSED SOLUTION & RESEARCH ACTIVITIES

WP1 Sensitivity analysis of cNDVI drought model to input parameters

The cNDVI model demonstrated to overperform WRSI model in detecting drought conditions over pastoral areas, thus reducing the overall basis risk of the product. However, considering that the customization process is an essential step in ARC country engagement, it is also important to evaluate the sensitivity of the proposed model to changes in key inputs that could be customized by end-users and their implication for the model accuracy (i.e. basis risk): the seasonality (defined from NDVI-based phenology thresholds), the selected mask (i.e. used for rangeland identification), the benchmark period for anomaly calculation. This would allow the definition of the operational modalities to perform the customization of the index.


WP2 Definition of an alternative processing chain based on a different satellite sensor

A critical operational requirement for any ARC index-insurance product is data continuity, in order to guarantee the fulfillment of the contractual terms of the active policy. MODIS NDVI satellite imagery was selected as the most suitable data source. However, the long-term continuity of the MODIS mission cannot be guaranteed, and an alternative data source should be identified, including a comparative analysis to understand the potential implications of the data source change on the product value and basis risk. The main candidate datasets are SPOT VEGETATION/PROBA-V NDVI/fAPAR products at 1km resolution freely distributed by Copernicus (https://www.copernicus.eu/en), one of the main data portals of the European Space Agency (ESA). An alternative to be considered is to use geostationary satellite data (MSG/Meteosat). One candidate dataset will be jointly agreed by the proposer and the ARC technical team at the beginning of the project. Then the comparative analysis will be carried out.


WP3 Proposal of an impact/payout model suitable for pastoral areas

The drought model needs to be coupled with an impact/payout model to estimate the indemnity payments. The current approach developed by ARC uses a vulnerability model to estimate the number of people affected by a given drought event. Alternative approaches, such as the one developed under for IBLI, linearly scale the indemnities to the drought index without measuring the impact of the drought with a quantitative variable (e.g. people affected, tons of forage production loss, ect.), but rather using a relative indicator. The total sum insured is based on the estimated cost the client/beneficiary is expected to use to mitigate the drought impacts. The latter approach has pros and cons. The main pros is that by keeping the model framework very simple, the risk of error propagation is reduced. The model is also easy to understand and require no additional inputs. On the other hand, the lack of a direct quantification of the impact could be seen as a limitation from some stakeholders and might hamper the process of validation.

This WP will investigate alternative options for the indemnities calculation suitable for the African pastoral areas and propose a new impact/payout model. The current ARC vulnerability model will be used as a benchmark, while alternative approaches will be explored based on the IBLI experience (adapted to a macro-level product) and on the literature review.


WP4 Technical support for implementation and training

ARC plans to launch the new pastoral product by 2019. This would require (i) that the technical implementation of the product would be completed and tested; (ii) that appropriate modalities for stakeholder engagement and customization are designed.

This work package aims at (i) supporting ARC in the technical implementation of the product by assisting the developers and (ii) assisting the operational implementation of the new product through selected case studies (e.g. Kenya, Ethiopia, Senegal, Mauritania). This will be done through at least a technical Workshop with developers, at least one Training of Trainers to ARC Technical Working groups (TWGs), and with support of at least a technical workshop (e.g. in Kenya) for product launch and customization (Dec 2019)


 

RESEARCH PARTNERS:

African Risk Capacity, University of Twente, Joint Research Centre


TIMELINE:

April 2018 - December 2019


FUNDED BY:

African Risk Capacity


CONTACT:

Francesco Fava

bottom of page