Expression of Interest (EoI) – Consultancy for Development and Piloting of a Localized, Efficient and Sustainable Earth Observation-based Crop Mapping Pipeline in Rwanda

Expression of Interest (EoI)

Consultancy for development and piloting of a localized, efficient and sustainable earth observation-based crop mapping pipeline in Rwanda

Reference Number: 83466234

Date of Publication: 03.06.2024

0. Context

The Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH is a federally owned international cooperation enterprise for sustainable development with worldwide operations. The GIZ Office in Kigali covers GIZ’s portfolio in Rwanda and Burundi. GIZ Rwanda/Burundi implements projects on behalf of the German Federal Ministry for Economic Cooperation and Development, the European Union and other commissioning authorities in the following priority areas: Sustainable Economic Development, Good Governance, Climate, Energy and Sustainable Urban Development, Digitalization and Digital Economy, Mineral Governance, Peace and Security in the Great Lakes Region.

About the GIZ Digital Transformation Center Rwanda

The program “Digital Solutions for Sustainable Development” (DSSD) aims to promote the development of digital solutions, digital inclusion and professional ICT skills and capacities. In 2019, DSSD opened the Digital Transformation Center Rwanda (DTC) as a hub for innovation and collaboration among public and private sector, academia, and civil society. Over time, other GIZ projects such as the global initiative FAIR Forward – Artificial Intelligence for all, Make-IT in Africa, and the Support Program to the Smart Africa Secretariat joined the DTC.

The Digital Transformation Center is now organized in various topics called Verticals for orientation, and proper collaboration with partners and the Rwandan ICT ecosystem and these verticals are collectively known as the GIZ Digital Cluster. Besides the AI Hub, these verticals include:

• Public Sector Innovation: The public sector innovation vertical is where the government, civil society and private sector all intersect to work together in developing digital solutions and innovation around the challenges experienced in the public sector. In this vertical, the demand is set through a series of collaborations from the government, mainly from the Chief Digital Officers of every sector, The Rwanda Information Society Authority RISA and the Ministry of ICT and Innovation and many other implementing agencies to work together. The solutions developed under the Public Sector innovation vertical revolve around digitization, change management, ICT building blocks, UN Sustainable Development Goals, Smart cities, and efficient delivery of Government services.
• Digital Inclusion & Literacy: The Digital Cluster is actively involved in capacity-building and digital literacy, targeting people in rural areas, women and people with disabilities, ICT professionals, tech startups and employees of political partner agencies. The Digital Inclusion and Literacy vertical works together with several civil society organizations, The Ministry of ICT and Rwanda Information Society Authority RISA to implement joint programs that bridge the skills gap for both public servants and citizens to create an environment conducive to digital transformation and adoption of digital skills for economic development and improved livelihoods.
• Startup Ecosystem Support: The Digital Cluster supports young entrepreneurs and startups through incubator and accelerator programs to build and expand their businesses. These programs provide access to mentorship, funding, and resources to help young entrepreneurs overcome the challenges of starting a business. By fostering collaboration and innovation, these initiatives are helping to create a thriving ecosystem for young entrepreneurs to succeed in today’s competitive marketplace.

About the Digital Transformation Center’s AI Hub

One of the focus areas of the Digital Transformation Center is Artificial Intelligence. Against this background, the AI Hub Rwanda has been founded, bundling all AI initiatives implemented by GIZ in Rwanda. It comprises two projects, the global FAIR Forward program as well as the component for Machine Translation of the DSSD program. The vision of the AI Hub Rwanda is to co-create a vibrant and inclusive ecosystem in Rwanda harnessing the benefits of open and ethical AI for sustainable development. Our mission is empowering our partners through providing open AI training data, capacity building, and development of ethical policy frameworks towards building community-driven AI solutions.

Specifically, we are working in 3 key areas:

• Open data, models and use cases for Natural Language Processing (NLP) and Machine Learning for Earth Observation (ML4EO): We facilitate the creation and open provision of non-discriminatory and inclusive training data and open-source AI applications. Open access to African language data is a key priority to enable the development of AI-based voice interaction in local languages to empower marginalized groups.
• Capacity building: We aim to strengthen the capacities of local actors (local developers, universities, companies and start-ups, governments, as well as civil society) in the development and application of artificial intelligence. For this purpose, we offer “on-the-job” training as well as specific training formats and fellowships.
• AI policy: We advocate for value-based AI that is rooted in human rights, international principles such as accountability, robustness, security and safety, transparency of decision-making, fairness, and privacy, drawing on European experiences such as the EU General Data Protection Regulation (GDPR). Therefore, the project supports the development of effective political and regulatory frameworks as well as policy dialogues at the national and regional level to promote AI adoption for sustainable development and inclusive growth in Rwanda.

Collaboration between MinAgri and RSA

The Rwanda Space Agency (RSA), mandated to regulate and coordinate all space activities in the country, is playing a crucial role in Rwanda’s smart agriculture journey. RSA is developing a Smart Agriculture System that provides near real-time vegetation/crop health monitoring with weekly insights with additional functionality being developed to support “an eagle eye’s view” seamless visibility of the agriculture system from planting, crop area estimation, predictive yield and production estimation for anticipatory planning and end of season assessments. This system is being co-developed with the Ministry of Agriculture and Animal Resources (MinAgri) and identified partners including GIZ to ensure it aligns with decision-making needs and data formats.

MinAgri and RSA collaborate extensively in the field of smart agriculture, with a strong focus on satellite technology and AI. Specifically:

1. RSA acts as MinAgri’s EO (Earth Observation) sustainability partner, advising on suitable EO and other emerging technologies for agricultural decision support, testing technologies for agricultural viability, and streamlining the development and co-development of satellite-based agricultural outputs.
2. Building upon its expertise and mandate, RSA streamlines all satellite-based agricultural outputs through its Smart Agriculture System. This system, co-developed with MinAgri, ensures alignment with decision-making needs and data formats strengthening the capacity to use the outputs by non-technical audiences such as analysts and managers to support agriculture decision making.
3. RSA works with MINAGRI to actively coordinate with satellite and emerging technology partners. This involves prioritizing needs, mobilizing resources, co-developing solutions, and fostering knowledge exchange on best practices. By working together, these partners accelerate the deployment of effective agri-tech solutions for a more sustainable and productive Rwandan agricultural sector.

Smart agriculture in Rwanda

The Agriculture Landscape in Rwanda 

Agriculture is considered the backbone of the economy, contributing to approximately 30% of the Gross Domestic Product (GDP) and employing approximately 80% of the population. The agriculture landscape is highly fragmented consisting of small-holder farmers and mixed highly intercropped production systems which are largely non-mechanized. The sector is characterized by a high population density and agroecological diversity. The sector is vital for food security, rural development, poverty reduction, and economic growth in the country. The GOR has prioritized investments to strengthen resilience and adaptation, modernize agriculture, improve productivity, and ensure sustainable practices to meet the challenges of a growing population and changing climatic conditions. The total arable land is about 1.4 million hectares, which is 52 per cent of the total surface area of the country. However, the actual area cultivated has exceeded 1.6 million ha in recent years. Another 0.47 million ha is under permanent pasture, so well over 70 per cent of the country’s total land surface is exploited for agriculture (REMA).

Implementation Context 

According to the 2020 Agricultural Household Survey there are 2.3 million (80% of households in Rwanda) agricultural households in Rwanda of which 98% are involved in cropping. Agriculture is considered the main activity for 86% of these households. Further, land consolidation and the land law seeking to reduce subdivision of land under 1 hectare is set to increase efficiency in agriculture production in Rwanda.

Digital readiness and data governance

MINAGRI through the digital office has established a unified approach to digitalization through the Agriculture Management Information System (FaMIS). FaMIS adopts a modular approach defining integration requirements to ensure consolidation of data to strengthen decision support analytics. MINAGRI is also in the process of developing its data strategy to strengthen data governance and support the capacities for quality analytics for decision support. The existing infrastructure can support additional systems plug-in. However, RSA as the sustainability partner will host the Smart Agriculture System whose data will feed into FAMIS analytics and supporting near-real-time monitoring of the agriculture season.

Short presentation of the Ministry of Agriculture and Animal Resources

The Ministry of Agriculture and Animal Resources (MinAgri) is the government ministry responsible for the development of the agriculture and livestock sectors in Rwanda. MinAgri has a mandate to promote the sustainable development of a modern, efficient and competitive agriculture and livestock sector, in order to ensure food security, agriculture export and diversification of the productions for the benefit of the farmer and the economy of the Country.

MinAgri is also the key institution responsible for developing and implementing policies, strategies, and programs that promote sustainable agricultural development, food security, and rural transformation. MinAgri recognizes that the effective use of data that is complete, available when required and of high quality is essential to meeting this mandate.

MinAgri’s goals, as outlined in its strategic documents PSTA4 include:

1. Enhancing food security and nutrition for all Rwandans.
2. Promoting sustainable agricultural practices.
3. Increasing agricultural productivity and production.
4. Facilitating access to markets for agricultural products.
5. Supporting the development of rural communities.
6. Strengthening the resilience of the agriculture sector to climate change.

Each of these goals requires the effective use of data that is of high quality, and which can be trusted.

Short presentation of the Rwanda Space Agency

Rwanda Space Agency (RSA) is the National Space agency that was established in 2020 with the mission of developing Rwanda’s space sector towards social-economic development.

RSA mandate is to regulate and coordinate all space activities in the country while also creating an environment that promotes entrepreneurial and industrial development to enable the creation of commercializable products that are globally competitive for local consumption and export markets.

Other key objectives of RSA include designing and implementing capacity building programs in space sciences and technologies, their applications and building highly skilled professionals in the space industry. This is implemented through various partnerships with different stakeholders.

Policy landscape

The implementation of an EO-driven agriculture monitoring system is justified and aligned with several key policies and strategies, including NST-1 (National Strategy for Transformation), PSTA-4 (Program for Strategic Transformation of Agriculture), the Agricultural Policy, and the National Digital Agriculture Strategy. The convergence of digitalization and the use of emerging technologies to strengthen agriculture monitoring and data driven decision making, as emphasized in NST-1, PSTA-4, and the Digital Strategy, presents an opportunity to strengthen agricultural productivity, enhance smallholder incomes, and promote resilience in the agricultural sector.

Rwanda’s strategic framework for ICT through MINICT also supports digital agriculture. The revised National Broadband Policy (2022) prioritizes 4G/5G access, while the Responsible AI policy (2023) promotes ethical development of AI solutions in the sector. Initiatives like the Startups Act and Rwanda Innovation Strategy foster agritech growth.

The creation of digitalization divisions by MINICT within key Ministries proved a game-changer for accelerating digital transformation. This strategic foresight, coupled with the ICT4RAG (ICT for Rwanda Agriculture- 2016-2020), strategy aligning PSTA-4 with digital delivery, provided a solid footing for the sector’s digital journey. ICT4RAG built on three pillars (ICT capability, secure infrastructure, effective governance), delivered tangible results: a data warehouse, improved ministry resources, and a MIS enabling monthly PSTA-4 data collection. Its implementation also introduced the Smart Nkunganire and Kuhangara systems, crucial for the successful subsidy program rollout. ICT4RAG’s learnings informed the development of NDAS (The National Digital Agriculture Strategy-2023-2030), which envisions digitally driven agricultural transformation through user-centric, accessible technologies, empowering farmers and boosting their productivity, income, and livelihoods.

Through MINICT support for an enabling environment for digital delivery in agriculture includes prioritization of Ag use-cases from policy and strategic documents, as is the case of the innovation projects such as the AI voice chatbot for agricultural advisories and the smart agriculture satellite monitoring system.

The web-based Smart agriculture system (SAS)

Among other things, a web-based Smart agriculture system is currently being developed. The Smart Agriculture System envisions a future where agricultural monitoring is seamlessly integrated into the entire agricultural cycle, from pre-planting to harvest. By harnessing the power of satellite imagery, weather data, and climate modelling, the system empowers farmers and policymakers to make data-driven decisions that optimize crop production, enhance resilience to climate change, and ultimately transform Rwanda’s agriculture sector. The Smart agriculture system will encompass the entirety of Rwanda, focusing on CIP (Crop Intensification Program) crop types such as banana, rice, maize, beans, potatoes, and cassava.

In other words, the system’s objective is to provide a singular view of the agriculture season from start to end, by leveraging on a consortium of partners to co-develop and automate analysis. In particular, the following functionalities shall be automated, and their results displayed in custom dashboards:

Its key components are:

• Season progression monitoring (Ready in MVP): Monitor the progress of the season at different levels to assess progress in cultivation and planting. – Completed, but waiting to be validated with ground-truth data
• Crop Type Classification (preliminary results in MVP): The platform will employ machine learning algorithms to accurately classify crop types, providing a clear understanding of crop distribution and patterns.
• Crop Health Monitoring: Utilizing satellite imagery, the system will continuously monitor crop health, identifying potential stressors and enabling timely interventions.
• Yield Estimation (proof of concept for Maize pending scaling): By analyzing crop growth patterns and weather data, the platform will provide yield forecasts, aiding in market planning and resource allocation.

Scope of the SAS

The comprehensive Smart Agriculture System will cover the entire country of Rwanda, with a focus on major CIP Crops. The system will also incorporate weather data and climate modelling to help farmers and policymakers anticipate and adapt to climate change impacts. The ongoing developments are prioritized based on data availability, processing time, and technical requirements.

1. Crop Coverage: The system will focus on major crop types in Rwanda, including banana, rice, maize, beans, potatoes, and cassava. These crops were selected based on their economic importance, production volumes, and environmental impacts. However, the system may also include other crops as data becomes available.
2. Geographical Coverage: The system will cover the entire country of Rwanda, including all provinces, districts, and sectors. This will enable farmers and policymakers to access data and information on crop growth, health, and yield for their specific locations. The system may also provide customized recommendations based on local conditions and practices.
3. Data Sources: The system will mainly rely on open-source satellite imagery and satellite climate data to generate insights and recommendations. The satellite imagery will be obtained from publicly available data and for some tasks like crop-type, high-resolution imagery from commercial providers might be necessary. All this data will be processed using machine learning algorithms to extract useful insights. The weather data will be obtained from remote sensing systems such as MODIS and TRMM.
4. Data Integration: The system will integrate data from multiple sources into a unified database, enabling users to access and analyze data from different sources in one place. The platform will also use data analytics and machine learning models to generate insights and recommendations based on the integrated data.
5. Precision Agriculture: The system will enable precision agriculture practices by providing near real-time information on soil moisture, and crop health patterns. This will allow to optimize agricultural inputs. The system may also provide customized recommendations based on crop-specific requirements and local conditions.
6. Climate Resilience: The system will incorporate climate modelling and early warning systems to help farmers and policymakers anticipate and adapt to climate change impacts such as drought, floods, and extreme temperatures. The system will provide information on weather patterns and trends, recommendations for crop selection, planting dates, and other adaptive management practices.

Users of the SAS

The preliminary scope of users is identified as Government Agriculture sector actors across MINAGRI, RAB and NAEB defined as primary system users.

Stakeholder	Interaction with System	Role/Responsibility
Farmers	Secondary	Benefit from optimized access to inputs from GOR subsidy program, better planning and timely interventions and investment
Extension Officers	Primary	Provide ground validation and contribute to the accuracy of the mapping, estimation, and monitoring data within their assigned regions.
Districts	Secondary	Support district level planning on food security
National Institute of Statistics and Research (NISR)	Primary	Collaborate in data sharing, validation, and integration of NISR reports into the system for comprehensive estimation and monitoring of agriculture parameters.
MINAGRI- Digital Division	Primary	Oversee and support the system’s implementation, provide policy guidance, and ensure that it aligns with the country’s agricultural development goals and priorities.
MINAGRI- Agricultural Modernization Department	Primary	Utilize the system to monitor and evaluate the effectiveness of agricultural modernization programs and interventions, make informed decisions for resource allocation, and track progress towards targets.
MINAGRI- Planning department	Primary	Utilize the system to inform GOR planning and M&E. Support optimal interventions and investments, including engagement with development partners to implement identified interventions.
MINAGRI- Value chains Department	Primary	Monitor Value chains performance and anticipated production and inform trade and supply chain management decisions.
RAB	Primary	Use the system to respond to early warnings, interface with district level decision making actors, and implement interventions to agriculture risks. Also coordinate data collection & validation. Support R&D to improve development and computation of agro-indicators.
Ministry of Finance and Economic Planning MINECOFIN	Primary	Utilize the system’s data and estimations for informed decision-making related to budget allocation, subsidies, and economic planning in the agriculture sector.
International Donor Agencies and NGOs	Secondary	Provide financial and technical support for the development, implementation, and capacity building related to the system, ensuring its sustainability and effectiveness in achieving agricultural development goals.
Technology Providers and Service Providers	Secondary	Collaborate in the design, development, and maintenance of the system, providing expertise in remote sensing, data analytics, machine learning, and GIS technologies.
Allied Private sector delivery partners	Primary	Utilize technology to optimize and improve service delivery. For example, use of the satellite estimates to optimize the delivery of the input subsidy programme delivery by improving assessments of the area under each crop to improve fertilizer and input allocation especially with the foresight of intercropping to reduce double requests on same piece of intercropped land.

Technical description of the SAS

System Architecture:

• Frontend: Javascript/ReactJS – Handles user interface and interactions.
• Backend: Javascript/Express – Manages data processing, logic, and API communication.
• Data Storage: PostgreSQL
• API Gateway: Handles communication with external APIs and services like FaMIS, MINAGRI, and potentially sensor data providers.

Development:

• Development Team: RSA
• Source Code Maintenance: Version control system (e.g., Git)
• Methodology: Waterfall development (more traditional, sequential approach)
• Roadmap

Links:

• Backend: https://github.com/bmurenzi25/RSA-smart-agri-backend
• Frontend: https://github.com/bmurenzi25/RSA-smart-agri-web[1]
• System link: https://smartagri.space.gov.rw/

System Architecture:

• The detailed architecture will be made available to the contractor during onboarding.

System design and functionalities

• User Interface: The system has a user-friendly interface that can be accessed through a web and mobile-based portal. The interface will enable users to access data and information in a visual and intuitive manner, with customizable dashboards and maps. The system will also provide training and support materials for users, including tutorials, manuals, and helpdesk services.
• Data Privacy and Security: The system adheres to Rwanda’s Law Nº 058/2021 of 13/10/2021 relating to the protection of personal data and privacy (DPP Law) and international standards for data privacy and security, including the General Data Protection Regulation (GDPR) and the ISO/IEC 27001 standard. The system will use encryption, access controls, and other security measures to protect user data and prevent unauthorized access or disclosure.

Modules / interfaces

• Web-App: A web-based system for farmers and policymakers to access useful agricultural insights extracted from satellite imagery, and weather data, extracted using machine learning models and statistical learning.
• API-Services: An API for integrating the system with FaMIS at MINAGRI, and with other agricultural technologies and services
• Data-portal: A database for storing and managing agricultural data, including historical records and real-time sensor data
• User-guide: Training and support materials for users, including tutorials, manuals, and helpdesk services.

Current implementation status

• Web –App status
  • Auth Service is completed
  • Core API is completed
  • Front End part is at 80%, gets improved as new functionality come in
• App Extraction Module status:
  • RS module is at 70%
  • Backend is at 80%
  • Optimization: Ongoing

1. Tasks to be performed by the contractor

Overall objectives

The contractor is expected to design, develop, test and finalize a remote sensing-powered crop-mapping tool. In particular, for any given area of interest (AOI), or for any polygon layer (e.g. plots from the agricultural cadastre) this tool should:

Identify (map) cultivated and planted areas and compute areas under each priority crop.

Identify (distinguish) the types of crops grown in those areas, with a focus on identifying dominant and subdominant intercropping.

Operational context

Data sources and sustainability

In order to ensure the project’s long-term and sustainable impact, this tool must be based exclusively on freely available satellite imagery (Sentinel, Landsat, etc.), and on the ancillary datasets (land cover, weather data, cadastre data, climatic models, etc.) that MinAgri can freely have access to. In other words, all the data needed for running the tool must be either freely available (open data) or can be provided to MinAgri for free (e.g. data from other Rwandan Ministries or Agencies). This is especially important because it is expected that the crop mapping tool will be based on machine-learning (ML) approaches, which will require massive stacks of current and past multispectral satellite images. Since neither MinAgri, nor RSA, nor GIZ have the financial resources to procure massive amounts of high-resolution commercial images, the only realistic option is to leverage exclusively the power of freely available satellite images. The contractor is also encouraged, as much as possible, to take advantage of existing APIs or cloud computing technologies (Google Earth Engine, Sentinel Hub API, Copernicus Open Access Hub API, Planet APIs, etc.) that allow to batch-process images without requiring to manually download satellite images². The exact architecture will be agreed upon with MinAgri and RSA as part of the kick-off workshop, but any solution that can streamline and simplify the data flows is seen as a very positive asset.

If needed, it is, however, permitted for the contractors to use high-resolution imagery during this project, with the aim of calibrating and validating the pipeline’s algorithms (e.g. for quality checks, or for digitizing training areas). Additional requirements to use commercial data can be evaluated and approved by GIZ, MinAgri and RSA, with a key consideration on sustainability. The procurement of high-resolution images is to be borne by the contractors and must be included in the budget³.

Expected collaboration between the contractor, RSA and MinAgri 

The contractors are expected to work in very close collaboration with RSA and MinAgri, whose specialists will be involved in every step of the process (co-design of the pipeline and algorithms, co-development, identification of the training areas, testing, quality control). This very close collaboration is mandatory to ensure that MinAgri and RSA will get full ownership and will be able to seamlessly integrate the pipeline into their existing processes and systems. In particular, the crop mapping pipeline must be interoperable with the Smart agriculture system. Sufficient capacity building through the co-development process will be required to support the capacity of MinAgri and RSA to adopt, sustain and scale the pipeline to the envisioned countrywide coverage.

Pilot areas, ground truthing and quality control 

The contractors are expected to test the crop mapping pipeline in 4 pilot areas (incl. collection of reference data), to validate the results against the reference data, and to fine-tune the pipeline, in order to ensure its high quality and robustness. One pilot area of ~600km² is to be selected within each of the following districts: Nyagatare, Musanze, Nyabihu and Ruhango.

Nyagatare District:

• Area: 1,979 km²
• Agro-ecology: Eastern Plateau (altitude 1400-2000m)
• Climate: Temperate to subtropical, with two rainy seasons (February-May and September-December) and average annual rainfall of 1000-1200mm.
• Priority Crops: Maize, potatoes, beans, wheat, and soybeans.
• Landholding Size: Mainly smallholder farms (average 0.7ha), with some larger commercial farms.
• Intercropping: Common, especially maize and beans.
• Dairy farming is also significant, contributing to milk production and income generation. Land degradation and soil erosion are challenges in some areas.

Musanze District:

• Area: 517 km²
• Agro-ecology: Volcanic Highlands (altitude 1800-2500m)

• Climate: Temperate to subalpine, with two rainy seasons (February-May and September-December) and average annual rainfall of 1200-1400mm.
• Priority Crops: Potatoes, beans, wheat, pyrethrum, and vegetables (e.g., carrots, onions, cabbage).
• Landholding Size: Varies, with smallholder farms (average 0.5ha) dominant in some areas and larger commercial farms in others.
• Intercropping: Common, especially potatoes and beans.
• Tea plantations are also present, contributing to export income.

Nyabihu District:

• Area: 592 km²
• Agro-ecology: Volcanic Highlands (altitude 1800-2500m)
• Climate: Temperate to subalpine, with two rainy seasons (February-May and September-December) and average annual rainfall of 1200-1400mm.
• Priority Crops: Potatoes, beans, wheat, pyrethrum, and vegetables (e.g., carrots, onions, cabbage).
• Landholding Size: Primarily smallholder farms (average 0.5ha).
• Intercropping: Common, especially potatoes and beans.
• Similar to Musanze, tea production plays a significant role. Steep slopes and soil erosion require sustainable land management practices.

Ruhango District:

• Area: 620 km²
• Agro-ecology: Central Plateau (altitude 1700-2000m)
• Climate: Temperate to subtropical, with two rainy seasons (February-May and September-December) and average annual rainfall of 1000-1200mm.
• Priority Crops: Maize, beans, cassava, sorghum, and vegetables (e.g., tomatoes, peppers, eggplants).
• Landholding Size: Mainly smallholder farms (average 0.7ha).
• Intercropping: Common, especially maize and beans.

• Dairy farming and livestock rearing are significant contributors to the agricultural economy. Soil fertility and water availability can be challenges in some areas.

Open Source and the Principles for Digital Development

In order to promote successful, efficient and effective digital development cooperation, GIZ signed the commitment to the Principles for Digital Development of the Digital Impact Alliance (DIAL)[2].

The Principles for Digital Development are nine living guidelines designed to help integrate best practices into technology-enabled programs and are intended to be updated and refined over time. They include guidance for every phase of the project life cycle, and they are part of an ongoing effort among development practitioners to share knowledge and support continuous learning. The Digital Principles were created in a community-driven effort, the result of many lessons learned through information and communication technologies (ICTs) in development projects.

Several of the 9 Digital Principles strongly advocate for an Open Source by default approach, which FAIR Forward Rwanda, in agreement with the national partners, encourages. The following is therefore expected from the contractor:

• All components used by the contractor shall be based exclusively on Free and Open-Source Software.
• Any custom development shall be released by the contractor under a Free and Open-Source license. The contractor will also be responsible for making the source code available (e.g. on a GitHub repository), and for ensuring that the source code is appropriately documented.
• All the methodological outputs (scripts, algorithms, training datasets) will be published as digital public goods under appropriate open licenses. The actual results (crop mapping) will be the sole property of MinAgri and RSA, who shall decide on the relevance of opening (or not) the access to those data. This supports the Rwandan AI ecosystem in creating more AI for agricultural innovation by providing easy access to AI and earth observation technology.

Agility

Due to the unpredictability of some key elements, such as weather events influencing the data collection campaign and the cloud cover, but also because some specific needs and requirements will arise during the project, the tenderers are expected to master approaches of agile development and agile project management. Concretely, a tentative workplan and timeline will be developed at the beginning of the assignment, and will be continuously adapted and fine-tuned, according to the forced changes induced by external factors, or by the emerging priorities and requirements.

The contractor is responsible for providing the following services:

Please note that the description of the following work packages does not imply a linear succession. Rather, it is expected that they will vastly overlap, e.g. the development of the crop mapping tool will happen in parallel with the development of the methodology, so that it can be tested, validated and improved early and frequently.

Work package 1: Kick-off workshop

The contractor will organize and moderate a kick-off workshop aimed at initiating the project and clarifying all open questions and topics. If the workshop happens physically, MinAgri, RSA and GIZ will be responsible for the logistical aspects (conference room, inviting the relevant stakeholders, catering, etc.).

Objectives:

• Get an early involvement of all the relevant stakeholders
• Clarify the project management processes (agile vs. waterfall) and tools (e.g. MS Teams Rooms, Kanban board, GitLab repository, virtual whiteboard, etc.)
• Nominate the expert teams from MinAgri and RSA who will actively participate in the conduction of the project
• Clarify the standards and guidelines that the contractor will need to consider (e.g. data models)
• Formalize how and when the contractor will get access to all the necessary national datasets (e.g. agricultural cadaster, national weather data, etc.)
• Discuss all open questions

Deliverables:

• All the project management processes and tools are in place, and access is granted to all participating to the project
• Roles and responsibilities are clearly defined and documented
• The contractor is granted access to all the necessary tools, platforms, documents and datasets
• A precise timeline is developed for the first month of the project. A rough timeline (to be continuously and iteratively updated) is developed for the following months

Work package 2: Development of the crop mapping methodology

The contractor will develop a first version of a crop mapping methodology especially adapted to the specific challenges and priorities identified by MinAgri and RSA (CIP crops, majority of smallholder farms, highly intercropped areas, partial availability of ancillary datasets, constrained financial resources, cloud cover mitigation, etc.).

To this aim, the contractor will work very closely with the expert teams nominated by MinAgri and RSA. It is expected that this methodology will be developed in a very iterative way, with frequent exchange and feedback loops between the contractor, MinAgri and RSA. These frequent exchanges will also serve as capacity development measures for MinAgri and RSA (e.g. on topics such as ML-aided crop detection, object delineation, automated land cover mapping, etc.).

The methodology should aim at creating a robust satellite (and ancillary) data pipeline with the best possible quality for the goals of the crop mapping tool under the constraints of MinAgri’s and RSA’s ability to continuously access the chosen sensors’ data (e.g. optical and SAR data from Sentinel 1 and Sentinel 2).

Specifically, this methodology shall cover the following:

• Identification of cultivated and planted areas
• Identification and quantification of CIP crops

The methodology shall also describe the method for collecting training data and ground-truthing, and address the IT requirements (hardware, development environment, etc.).

Objectives:

• Develop a crop mapping methodology which:
  1. Fits to the context of Rwanda (intercropping, smallholder farming, frequent cloud cover, etc.) and to the specific expectations and needs from MinAgri
  2. Builds upon robust approaches developed and tested in similar countries and regions

Deliverables:

• A crop mapping methodology, in English language, with the following characteristics:
  1. Specifically tuned to the ecological, climatic and agricultural context of Rwanda
  2. Relying exclusively on datasets that can be freely acquired by MinAgri and RSA
  3. Being operational: the methodology shall also address the technical and tactical aspects for future upscaling (data collection campaigns, technical environment, etc.), incl. a tentative plan (roadmap, timeline, resources) for a future nationwide deployment
  4. Being scalable: Defined procedures for model updates, training, and re-calibration based on new data; established workflows for addressing model drifts and performance degradation over time.

Work package 3: Collection of reference data

The contractor will collect reference data, to be used for training the crop mapping pipeline, and for validating the results. In their technical offer, the bidders are requested to explain the method envisioned for the data collection (in-situ, remote, or a mix of both), its tentative timeline (incl. season(s) during which the data collection shall take place), the sampling methodology (number and stratification of samples, ratio between training and validation data, etc.) and the tactical aspects of the data collection campaigns (who will actually collect the data? with which tools? how will the enumerators be trained?, etc.).

The methodological description and action plan of the reference data collection shall include the definition of a sampling unit, the minimum mapping unit, the sampling design, the timing of the field surveys according to CIP crops’ calendar, survey variables, use of digital toolkit and required equipment and personnel, approach to mitigate bias in collected data.

Objectives:

• Collection of the necessary reference dataset, to be used for developing and calibrating the crop mapping pipeline, and for assessing the quality of the results

Deliverables:

• Full dataset of collected reference data
• Methodological description, incl. implementation plan (budget allocated, number of teams and staff involved in the data collection, data collection sheets, lessons learnt and issues, etc.), to allow MinAgri and RSA and consistently repeat the data collection in the future (for updating the data, or for expanding to other areas)
• If the data collection happens in-situ, the electronic forms must also be delivered (e.g. as XLSForm, if the contractor uses Kobo Toolbox)
• All the data used during the development of the tool (e.g. high-resolution imagery used for delineating the training areas)

Work package 4: Development of the crop mapping pipeline

Once the methodology has been approved by MinAgri, RSA and GIZ, the contractor will develop the code and scripts allowing to conduct the methodology over a given area of interest, for a specific timeframe. Typically, the code and scripts will allow the following:

• Automatic fetching and pre-processing of the necessary data (e.g. download of all relevant Sentinel-1 data and Sentinel-2 images with low cloud cover)
• Automatic processing (extraction of the cultivated and planted areas from the Sentinel-1 and Sentinel-2 data, validation against ancillary datasets, segmentation into crop types)
• Export of results (e.g. as shapefiles)

It must be possible to run the code and scripts in at least 2 different ways:

• Locally, through the command line interface, or a simple user interface
• Remotely, through an API. The objective is that a module dedicated to crop mapping can be integrated in the SAS in the future. This future module shall only need to send an API request to the system’s API gateway. The contractor will NOT be responsible for developing this future module. It must however ensure that the API is robust, secure and well documented, so that the integration with the SAS will not pose major challenges in the future

The contractor will work very closely with the IT and EO expert teams nominated by MinAgri and RSA. It is expected that this pipeline will be developed in a very iterative way, with frequent exchange and feedback loops between the contractor, MinAgri and RSA. These frequent exchanges will also serve as capacity development measures for MinAgri and RSA. Technical staff from MinAgri and RSA shall, at any time, have access to the latest version of the source code and of its documentation, and must be enabled to provide feedback, bug reports or pull requests.

The crop mapping pipeline will ultimately be integrated into the SAS and, therefore, the contractor is not expected to develop a sophisticated dedicated user interface. A simple interface (ideally, web-based) will be sufficient, with the aim of testing and demonstrating the pipeline. It should, however, be robust enough to enable MinAgri and RSA staff to work with it until the pipeline has been integrated into the SAS.

The crop mapping pipeline must be built as much as possible on free and open-source components. The use of proprietary components can be acceptable in specific cases (if it brings a tangible benefit for MinAgri and RSA), but only after formal approval from RSA, MinAgri and GIZ. In particular, the licensing model of the proprietary tool must be assessed, to ensure that the pipeline’s sustainability will not be hindered by unforeseen licensing costs or usage restrictions.

Objectives:

• Develop the scripts and programs which allow to run the methodology on-demand (acquisition of the satellite images, processing and analysis, export of the results in GIS formats)
• Develop a simple, yet robust user interface
• Develop an API allowing to run the scripts remotely, so that the tool can be easily integrated into any existing system
• Iteratively deploy the tool’s development versions to MinAgri or RSA servers, to allow for early and frequent testing

Deliverables:

• Full source code of all the scripts and programs
• Appropriate documentation, both for users and developers

Work package 5: Development of the crop mapping pipeline

Once the pipeline has been developed, it will be tested in the 4 pilot areas. The contractor shall run the pipeline in those areas, validate the results against the reference data, and, if necessary, adjust and fine-tune the pipeline (parameters, weights, algorithms, etc.).

Objectives:

• Test the quality and robustness of the tool on the pilot areas
• Validate the results against the reference data
• Adjust the methodology and the tool’s source code, according to the results of the validation

Deliverables:

• Results, figures and data from the pilot areas
• Description of the validation approach, incl. its limitations (e.g. depending on the timing of the project, and on the season(s) during which the reference data collection will take place, the contractor might not be able to validate the yield estimates)
• Results from the validation (deviation between estimates and reality)

Work package 6: Development of the crop mapping pipeline

Objectives:

• The contractor will provide technical support to MinAgri and RSA during the whole duration of the contract, as soon as the pipeline’s first version has been deployed on MinAgri’s servers

Deliverables:

• Varying, as per the requests from MinAgri and RSA (possibly: regular coaching / backstopping, co-development of minor additional features, bug correction, etc.)

Timeline of the assignment

The contractor is expected to conduct all the above work packages from 15.07.2024 until 30.09.2025.

2. Concept

In the bid, the bidder is required to show how the objectives defined in Chapter 2 are to be achieved, if applicable under consideration of further specific method-related requirements (technical-methodological concept). In addition, the bidder must describe the project management system for service provision.

Note: In their offer, the tenderers are expected to strictly adhere to the structure outlined under Technical-methodological concept.

Technical-methodological concept

Strategy: The bidder is required to consider the tasks to be performed with reference to the objectives of the services put out to tender (see Chapter 1). Following this, the bidder presents and justifies the strategy with which it intends to provide the services for which it is responsible (see Chapter 2). The bidder should avoid repeating information from the description of the implementation approach. In particular, the bidder is expected to cover at least the following aspects within its strategy:

• Short description of past experiences: the tenderer shall explain how they have successfully conducted crop mapping projects in similar contexts. In particular, the following aspects must be explicitly addressed: dealing with smallholder farming and inter-cropping mapping, ensuring countrywide scalability, conducting broad-scale in-the-field data collection campaigns, leveraging open-source tools and methods, relying on freely available satellite imagery, developing APIs and web platforms.
• Approach towards crop mapping: given the specific context of Rwanda (inter-cropping, smallholder farming, cloud cover, …), describe the (tentative) methodology that will be developed for crop mapping. Indicate the main challenges to be expected, as well as the envisioned strategies to address them.
• Upscaling: yield predictions: although yield predictions are not part of this assignment, they are seen as the logical evolution of the crop mapping pipeline in a near future. Therefore, the tenderers are requested to shortly explain how their approach can flow into yield prediction models, e.g. by showcasing previous projects that they have successfully implemented in this field.
• Technical considerations: indicate explicitly the tools, technologies and datasets that you expect to leverage, as well as the key technical and architectural aspects to ensure that the pipeline and its API will be robust, scalable and secure. Also indicate how the simple, yet robust user interface will be developed, incl. tentative wireframes.
• User-centricity: describe the envisioned strategy for continuous development and iterative deployment of the crop mapping tool. Indicate how the users and stakeholders will be enabled and encouraged to frequently test the tool, and how their ideas, wishes and feedback will be collected and considered for the further development iterations.
• Approach towards data collection: describe the approach that will be used for collecting data in the 4 pilot regions. Address the methodological aspects (sampling methods, matching the data collection campaigns with the plants growth seasons, etc.), as well as the technical (mobile apps, mobile devices, connectivity in the field, etc.) and tactical ones (number and profile of enumerators, training of the enumerators, incentives, transportation and logistics, etc.).
• Sustainability: describe the actions that will be taken to ensure that MinAgri and RSA will be enabled to use, and further improve the crop mapping pipeline autonomously on the long term.

The bidder is required to present the actors relevant for the services for which it is responsible and describe the cooperation with them.

• What approaches / methods will be put in place to ensure a very close cooperation, enabling very frequent feedback loops
• What tools will be used to support this cooperation (e.g. for chats / conversations, for code management, for file sharing, etc.)

The bidder is required to present and explain its approach to steering the measures with the project partners and its contribution to the results-based monitoring system.

The bidder is required to describe the key processes for the services for which it is responsible and create a schedule that describes how the services according to Chapter 2 are to be provided. In particular, the bidder is required to describe the necessary work steps and, if applicable, take account of the milestones and contributions of other actors in accordance with Chapter 2.

The bidder is required to describe the role(s) that MinAgri and RSA staff will fulfil during the project . In particular, what concrete contributions are expected from MinAgri and RSA, and how will those be coordinated and integrated within the project.

The bidder is required to describe its contribution to knowledge management for the partner and GIZ and promote scaling-up effects (learning and innovation). In particular, explain how the documentation (technical documentation and user documentation) will be developed, tested, maintained and shared with the partners.

Project management of the contractor

The bidder is required to explain its approach for coordination with the GIZ project.

• The contractor is responsible for selecting, preparing, training and steering the experts (international and national, short and long term) assigned to perform the advisory tasks.
• The contractor makes available equipment and supplies (consumables) and assumes the associated operating and administrative costs. The equipment procured will be handed over to the designated cooperation partners of GIZ at the end of the project.
• The contractor manages costs and expenditures, accounting processes and invoicing in line with the requirements of GIZ.

The contractor reports regularly to GIZ in accordance with the AVB of the Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH from 2018

The bidder is required to propose a tentative workplan and timeline for conducting the project, and to explain what methods and approaches will be used to efficiently, continuously and empirically adjust and adapt the workplan. In particular, the approach and processes for continuously adjusting the workplan to changes and emerging opportunities must be described.

3. Personnel concept

The bidder is required to provide personnel who are suited to filling the positions described, on the basis of their CVs (see Chapter 7), the range of tasks involved and the required qualifications.

The below specified qualifications represent the requirements to reach the maximum number of points.

Team leader

Tasks of the team leader

• Overall responsibility for the advisory packages of the contractor (quality and deadlines)
• Coordinating and ensuring communication with GIZ, partners and others involved in the project
• Personnel management, in particular identifying the need for short-term assignments within the available budget, as well as planning and steering assignments and supporting local and international short-term experts
• Regular reporting in accordance with deadlines

Qualifications of the team leader

• Education/training (2.1.1): Postgraduate university degree (Master’s) in Geoinformation, ICT, environmental sciences, or a relevant field
• Language (2.1.2): proficiency in English
• General professional experience (2.1.3): 15 years of professional experience in the field of crop mapping
• Specific professional experience (2.1.4): 10 years closely collaborating with high-level national and international institutions
• Leadership/management experience (2.1.5): 7 years of management/leadership experience as project team leader or manager in a company
• Regional experience (2.1.6): 5 years of experience in projects in Central or Eastern Africa
• Other (2.1.8): 5 years of professional experience with agile development approaches

Short-term expert pool with minimum 2 maximum 5 members 

The technical assessment will be conducted against the qualifications of the expert pool as a whole

Please send a CV for each pool member (see below Chapter 7 Requirements on the format of the bid) for the assessment.

Tasks of the short-term expert pool

• Develop the crop mapping methodology
• Implement the methodology into a crop mapping pipeline
• Develop and maintain the API and the user interface
• Ensure the quality and robustness of the code
• Maintain the code and the documentation in an environment accessible to MinAgri and RSA (GitHub, GitLab, or similar)
• Ensure the capacity development of MinAgri and RSA (backstopping, training, etc.)

Qualifications of the short-term expert pool

• Education/training (2.6.1): 2 experts with Postgraduate university degrees in ICT, environment sciences or similar fields
• Language (2.6.2): All experts with proficiency in English
• General professional experience (2.6.3): 1 expert with 15 years of professional experience in crop mapping, 1 expert with 5 years of professional experience with machine learning for agricultural applications, 1 expert with 10 years of professional experience in the development of web platforms and APIs, 1 expert with 7 years of professional experience in the design, conduction and coordination of field data collection campaigns
• Specific professional experience (2.6.4): 1 expert with 5 years of professional experience with code versioning tools
• Regional experience (2.6.5): 2 experts with 5 years of experience in projects in Central or Eastern Africa

4. Costing requirements

Assignment of personnel

Per-diem and overnight accommodation allowances are reimbursed as a lump sum up to the maximum amounts permissible under tax law for each country as set out in the country table in the circular from the German Federal Ministry of Finance on travel expense remuneration (downloadable at https://www.bundesfinanzministerium.de).

Travel

The service provider is required to calculate the travel by the specified experts and the experts it has proposed based on the places of performance stipulated in Chapter 2 and list the expenses separately by per-diem allowance, accommodation expenses/allowance, flight costs and other travel expenses.

The bidder is required to calculate the travel by the specified experts and the experts it has proposed based on the places of performance stipulated in Chapter 2 and list the expenses separately by daily allowance, accommodation expenses, flight costs and other travel expenses.

Specifically, if proposed experts are from outside of Rwanda, where the travel costs are to be included in the offer, the contractor will calculate the following tips:

• 2 return trips to Rwanda for a maximum of 4 experts for a maximum of 80 days
• 80 days of travel within Rwanda for a maximum of 4 experts

Specification of inputs:

Fee days	Number of experts	Number of days per expert	Total	Comments
Team Leader	1	30	30
Short-term expert pool	1	200	200	200 days altogether, to be split between the members of the expert pool
Travel expenses (applicable only for proposed expert from outside of Rwanda)	Quantity	Price	Total	Comments
Per-diem allowance in country of assignment	80
Overnight allowance in country of assignment	80
Transport	Quantity	Price	Total	Comments
International flights Flight ticket for Travel to Rwanda (If applicable)	8			If applicable: Travel to the place of service delivery Rwanda (2 return trips to Rwanda for a maximum of 4)
Other travel expenses if applicable (visa,transfer to/from airport etc ..)
Other costs	Number	Price	Total	Comments
Flexible remuneration	RWF 8.290.020	A budget of RWF 8.290.020 is foreseen for flexible remuneration. Please incorporate this budget into the price schedule. Use of the flexible remuneration item requires prior written approval from GIZ.
Procurement of materials and equipment	RWF 11.053.360	The budget contains the following costs: smartphones / tablets and Sim cards for the field data collection, high-resolution satellite images (for collecting reference data and / or for calibrating the model). The equipment will be handed over to the designated cooperation partners of GIZ at the end of the project.
Enumerator related costs	1	Up to RWF 20.725.050	Up to RWF 20.725.050	The budget contains the following costs: (daily fee), travel-related costs within Rwanda for the data collection campaign.

5. Inputs of GIZ or other actors

GIZ and/or other actors are expected to make the following available:

• Logistics for workshops: GIZ, RSA and MinAgri will provide the logistics (conference rooms, workshop material, coffee break) for the kick-off workshop and other workshops, should they take place physically

6. Requirements on the format of the bid

The structure of the tender must correspond to the structure of the ToRs. In particular, the detailed structure of the concept (Chapter 3) should be organised in accordance with the positively weighted criteria in the assessment grid (not with zero). The tender must be legible (font size 11 or larger) and clearly formulated. It must be drawn up in English.

The complete tender must not exceed 10 pages (excluding CVs). If one of the maximum page lengths is exceeded, the content appearing after the cut-off point will not be included in the assessment.

The CVs of the personnel proposed in accordance with Chapter 4 of the ToRs must be submitted using the format specified in the terms and conditions for application. The CVs shall not exceed 4 pages each. They must clearly show the position and job the proposed person held in the reference project and for how long. The CVs must also be submitted in English.

Please calculate your financial tender based exactly on the parameters specified in Chapter 5 Quantitative requirements. The contractor is not contractually entitled to use up the days, trips, workshops or budgets in full. The number of days, trips and workshops and the budgets will be contractually agreed as maximum limits. The specifications for pricing are defined in the price schedule.

7. Submission of the offer

7.2 Technical Proposal

• Self-declaration of eligibility for the award (see annex form) with administrative documents of the company.
• A cover letter expressing your interest in this assignment
• Technical Proposal (attached template for technical proposal MUST be used)
• Up to date CVs of proposed experts

7.3 Financial offer:

Financial offer indicates the all-inclusive total contract price, supported by a breakdown of all costs as described in the specification of inputs. The costs must be in RWF and VAT excluded.

Please submit electronically your EoI (technical & Financial offer) in 2 separated emails and should be in PDF files to this email ONLY: RW_Quotation@giz.de until latest June 24^th, 2024. Please you must write on each email subject this sentence: 83466234-Technical/financial offer, without this sentence, your offer may not be considered

Hard copies are not allowed this time

GIZ reserves all rights

Annex:

• Eligibility assessment grid
• Self-declaration of eligibility for the award
• Technical Proposal template
• Technical assessment grid
• Price sheet

8. List of abbreviations

AVB General Terms and Conditions of Contract (AVB) for supplying services and work 2022

DSSD Digital Solutions for Sustainable Development

DTC Digital Transformation Center Rwanda

ToRs Terms of reference

AG Commissioning party

AI Artificial Intelligence

AN Contractor

API Application Programming Interface

AVB General Terms and Conditions of Contract for supplying services and work

CIP Crop Intensification Program

EO Earth Observation

FK Expert

FKT Expert days

KZFK Short-term expert

GOR Government of Rwanda

ICT4RAG ICT for Rwanda Agriculture- 2016-2020

MinAgri Ministry of Agriculture and Animal Resources

MinICT Ministry of ICT & Innovation

MVP Minimum Viable Product

NAEB National Agricultural Export Development Board

NISR National Institute of Statistics and Research

PSTA4 Strategic Plan for Agriculture Transformation 2018-24

RAB Rwanda Agriculture and Animal Resources Development Board

REMA Rwanda Environment Management Authority

RSA Rwanda Space Agency

SAS Smart Agriculture System

[1] The GitHub repositories are not publicly available. Bidders can however be added to the repo upon request.

[2]https://digitalprinciples.org/endorse/endorsers/