For any questions or comments about this report, please reach out to our team members:
Daniel Osgood: deo@iri.columbia.edu
Max Mauerman: mmauerman@iri.columbia.edu
Dante Salazar: dante@iri.columbia.edu
Led by the World Food Programme (WFP), the FbF/AA project in Ethiopia is an initiative aimed to strengthen national capacities for climate risk management.
In Phase 1 of the project, IRI conducted a remote technical training on forecast generation, a workshop on using forecasts for anticipatory action planning, and provided technical improvements to the tools used to generate and visualize forecast information. The technical training focused on building the Ethiopian Meteorological Institute (EMI)’s capacity for seasonal forecast generation and updating data libraries, while the policy workshop focused on reviewing anticipatory actions and setting forecast trigger rules for the Oromia region. Finally, on the technical support side, IRI provided a re-designed AA dashboard for EMI to maintain as part of the local hand-over of responsibility, and provided additional tools to explore decision rules based on both forecast and observational data.
In Phase 2 of the project, IRI finalized the AA dashboard tools to be handed off to EMI, and conducted an intensive in-person workshop with the EMI on the entire end-to-end AA process: From setting up a locally tailored objective seasonal forecast using IRI’s pyCPT2 tool, to the steps of generating, verifying and uploading routinely issued forecasts to the EMI server, and finally to the steps of creating and updating an AA dashboard using RMarkdown.
Following this training, EMI was able to successfully generate and communicate seasonal forecasts for AA during the 2024 October-November-December (OND) season. Following this, IRI worked on revisions to the AA dashboards to make them more robust and tailored to EMI’s needs, and finalizing documentation of the forecast generation and verification process for posterity.
# Deliverable
Status
Terms of Reference available here: https://drive.google.com/file/d/1aWRj_3lsPZnnr8xs92rL_e1yiBX6Dw4g/view?usp=sharing
Objectives
The main objective of this workshop were to present and review the forecast generation process using CPT (PyCPT) for both Oromia and Somali regions. The specific objectives are to: 1. Provide technical training for EMI experts and staff for partners in both Somali and Oromia regions. 2. Co-develop tailored drought forecasts for Oromia Region, MAM season. 3. Provide technical support for the tool updating process with forecast. 4. Conduct capacity assessments with a pre and post training approach. 5. Create documentation for decision-making tool usage.
Expected outcomes
Develop knowledge and capacity on the following aspects: * Operational seasonal forecasting * CPT and PyCPT use * Forecast Verification * Forecast Tailoring * Forecast Interpretation * Forecast fit into global forecast (or other reference data) * PyCPT V2.0 Installation
Workshop participants
Central office (HQ): EMI experts and partners from the national office were be invited to attend so existing capacity can be leveraged on the technical assistance as well as the data library management.
Regional Offices: The workshop was be open to EMI experts and staff from Somali and Oromia regions, as well as the staff in charge of data library management and operation, to ensure local capacity building for AA system support and implementation.
Workshop Logistics
Workshop Recordings
Technical Outputs
Technical Note available here: https://drive.google.com/file/d/1ftyhu7048HzNZx3zvsyODkGsnlLS29u2/view?usp=sharing
MAM Jupyter Notebook available here: https://drive.google.com/file/d/1r3GwrosmNyFbf2ZyRIELZ0gFFJIYl6Nh/view?usp=sharing
Python configuration file available here: https://drive.google.com/file/d/1aYsKoJQ-XdKv-8vCfYZtJLsdEnOq2XI6/view?usp=sharing
The IRI a workshop on the review and refinement of Anticipatory Action Protocol (AAP) for Oromia region on 20th and 21st March, 2024, focusing on planning actions for the coming October-November-December (OND) seasonß.
The full presentation and materials for this training sessions can be found in the link below. Please note that figures and forms were used to illustrate and collect past years’ drought impact severity for reference and development of the trigger system for the region. The document below contains reference materials for all the concepts covered in the workshop, including points for participant feedback.
Link to full materials: https://fist.iri.columbia.edu/publications/docs/ethiopia_aa_oromia_ond_workshop/
Workshop goals:
You can find an example of the draft AA protocol developed for MAM season here:
https://fist.iri.columbia.edu/publications/docs/eth_oromia_fbf_monitor_2023
You can find the previous MAM workshop materials here, including a guide to the AA Design Tool:
https://fist.iri.columbia.edu/publications/documents/Ethiopia%20FbF%20Presentation.pptx
Section 2 of the workshop pertained to the generation of a preliminary list of known drought years, a process previously completed during the MAM Season workshop. The regional Technical Working Group (TWG) was able to compile this list, which serves as the basis for understanding drought risk and making informed decisions. The historical analysis of drought occurrences and impacts on specific sectors and communities is essential for stakeholders.
To enhance the utility of this list, years were ranked for comparison against historical data, and impacts within the focused sector were discussed. Consideration was also be given to multiple sectors or sub-regions of Oromia to account for variations in the timing and severity of climate events. If necessary, the TWG can revisit the process to revise the list for the OND season.
In Section 3, the process of setting triggers for an FbF/AA project was discussed, emphasizing the importance of collaboration between climate scientists, experts, and stakeholders. Triggers were established based on the expected frequency and severity of a hazard, such as drought, and its impact on the ground. The trigger setting process followed an impact-based approach, which involved validating the performance of a drought forecast, determining appropriate lead times, and assigning severity levels to Early Actions (EA). The probability output of the forecast determines whether an EA will be triggered.
A decision maptool was developed to assist in selecting variables to support the development of Standard Operating Procedures (SOPs). Historical drought years, from Section 2, would be used to identify SOPs based on severity levels, and the map tool is utilized to set triggers for different severity levels of drought. This process involves determining the frequency of historical drought events to establish trigger frequencies for forecasts. Additionally, observational triggers, such as previous month’s vegetation cover, may be used to capture activities of greater concern based on preceding dry seasons and as validation checks on the forecast as the season progresses.
A proposed methodology for reviewing Early Actions (EA) and Standard Operating Procedures (SOPs) was shared. It involves several key steps. 1. Stakeholders and partners engage to agree on feasible activities for implementation in preparation for potential drought years, considering sectors, implementing agencies’ capacities, and budget requirements. 2. Timelines are established for each early action to ensure timely execution before the rainy season, alongside defining budgets and geographical coverage. 3. Trigger frequencies are determined based on factors like drought severity and budget distribution, ensuring early activation of actions in response to severe events and appropriate allocation over the program’s duration. This methodology aims to optimize Early Action plans, aligning them with forecast information to enhance preparedness and response to drought events.
To better inform anticipatory action planning and the handoff of operational forecast responsibility to the EMI, IRI produced several updated codebases and tools related to AA. These include (1) a dashboard for AA review and monitoring, designed to be maintained by the EMI, (2) updated decision support tools to explore the potential of incorporating both forecasts and observational conditions into AA, and (3) technical improvements to the existing educational and data exploration tools.
Following detailed feedback and discussion with WFP HQ, RBN, and the Ethiopia CO, as well as input from EMI, IRI developed a redesigned approach to anticipatory action publication and review. This resdesign approach is meant to consolidate and formalize each step of the forecast review and dissemination process, with the goal of standardizing approaches and making things as clear as possible for the EMI.
The dashboard hosted on the EMI website can be found here, along with a mirror on the IRI website in the event that the EMI website is unavailable due to server downtime or maintenance:
To make this process more robust, IRI is working to port the RMarkdown dashboard creation scripts to run on the EMI server instead of relying on local machines. This system is currently functional as a local application (see links in User Guide), but is awaiting EMI input before it can be run on their server.
Following feedback from WFP, IRI created an expanded set of decision support tools to accommodate the more sophisticated AA rules being developed in Somali and Oromia regions. These rules may incorporate both forecast data and observational data; hence, there is a need for data exploration tools that can explore options for combining each type of data.
The decision support tools are accessible via the links below:
IRI made updates to the existing map tools that have been used for education and decision support, including updating the list of bad years and the region shapefiles to their most current versions, and incorporating the most recent seasons’ forecasts. In addition, IRI integrated new sources of vegetation data to replace the MODIS product, which is being phased out as of 2024.
The maptools continue to be found at the following links:
AA Design Tool - Somali Region OND https://iridl.ldeo.columbia.edu/fbfmaproom2/ethiopia-ond?mode=0&season=season2&predictors=pnep+rain&predictand=bad-years&year=2022&issue_month=jul&freq=30&severity=0&include_upcoming=false&position=%5B6.875%2C43.875%5D&show_modal=true
AA Design Tool - Somali Region MAM https://iridl.ldeo.columbia.edu/fbfmaproom2/ethiopia?mode=0&season=season1&predictors=pnep+rain&predictand=bad-years&year=2023&issue_month=feb&freq=30&severity=0&include_upcoming=false&position=%5B6.875%2C43.875%5D&show_modal=true
AA Design Tool - Oromia Region OND https://iridl.ldeo.columbia.edu/fbfmaproom2/southern-oromia-ond?mode=0&season=season1&predictors=pnep+rain&predictand=bad-years&year=2022&issue_month=sep&freq=30&severity=0&include_upcoming=false&position=%5B5.7%2C40%5D&show_modal=true
AA Design Tool - Oromia Region MAM https://iridl.ldeo.columbia.edu/fbfmaproom2/southern-oromia?mode=0&season=season1&predictors=pnep+rain&predictand=bad-years&year=2023&issue_month=feb&freq=30&severity=0&include_upcoming=false&position=%5B5.7%2C40%5D&show_modal=true
Following participant feedback from the July 2024 EMI workshop, IRI implemented several requeseted technical improvements to the PyCPT2 forecast software, including:
Following a workshop at the EMI offices in Addis Ababa on July 1-5, 2024, EMI was able to successfully generate, verify and communicate the first AA seasonal forecast for the OND 2024 season, issued in mid-July.
Objectives
This mission aims to achieve the project’s overall goal through the provision of specialized services for building national capacities for AA in Ethiopia. To achieve this goal, IRI is working with WFP to fully hand over the developed FbF map room with associated map tool to Ethiopian Meteorological Institute (EMI), to undertake further improvement of FbF map room for drought forecasts in the Oromia Region and to provide capacity development training for EMI Oromia region experts.
Expected outcomes
Specific activities addressed by this mission include:
Workshop participants
Central office (HQ): EMI experts and partners from the national office were be invited to attend so existing capacity can be leveraged on the technical assistance as well as the data library management.
Regional Offices: The workshop was be open to EMI experts and staff from Somali and Oromia regions, as well as the staff in charge of data library management and operation, to ensure local capacity building for AA system support and implementation.
Workshop Logistics
The mission took place in person, at EMI HQ in Addis Ababa, from July 1st through 5th (5 working days). The mission covered the technical details of generating, validating and uploading forecasts using the PyCPT2 software; the division of responsibilities between agencies for routine AA monitoring and seasonal assessment; and a review of the Anticipatory Action Protocols. As part of these efforts, participants were asked to work on a forecast project specific to the region they are responsible for.
Technical Outputs
Workshop materials are available here:
Results of the pre- and post-training capacity assessment can be found in the “Capacity Assessment and Next Steps” section.
To support the sustainable handoff of the AA tools and processes to EMI, IRI developed an end-to-end user guide, available here:
https://fist.iri.columbia.edu/publications/docs/Ethiopia_AA_Manual_2024/
The following plots show the change in participants’ self-reported technical capacity in a number of key areas, before and after the in-person training.
On average, participants reported improvements in all areas of technical capacity, particularly in the area of using IRI’s PyCPT2 software for forecast generation and verification, in which the modal participant response went from a 2 (beginning) to a 4 (proficient) or 5 (expert). Participants reported that the detailed discussion of forecast configuration tailoring and model selection using PyCPT was particularly useful, and that they would appreciate ongoing support during the rest of the year and potentially follow-up trainings on topics like sub-seasonal forecasts (which are out of the scope of this project, but could be covered in future funded projects).
Operational seasonal forecasting
PyCPT use
Forecast verification
Forecast tailoring
Forecast interpretation
Cross-comparison of local and global forecasts
PyCPT2 installation
Operational seasonal forecasting
PyCPT use
Forecast verification
Forecast tailoring
Forecast interpretation
Cross-comparison of local and global forecasts
PyCPT2 installation
Next steps for potential future work after this project include: