20 SAS–2026 A © NISR 2.2.3. Data collection tools 2.2.3.1. Survey questionnaires The SAS utilizes two main questionnaires: The Screening Questionnaire and the Plot Interview Questionnaire. The Screening Questionnaire is designed to gather information on the plot, focusing primarily on aspects such as land use, plot area, and crops cultivated. Conversely, the Plot Interview Questionnaire is specifically designed to collect detailed information from the sampled plots, covering crop production, agricultural inputs used, and applied agricultural practices. 2.2.3.2. Data collection applications The SAS data collection applications were based on three main software applications: • Arc GIS field map, which utilizes GIS software and an external GPS linked to tablets via Bluetooth to accurately measure crop areas. • CSPro software, known for its efficiency in census and survey administration, this software facilitates data collection, entry, and management processes. Csentry data collection tool was developed by an IT staff specialized in the SAS survey, enabling data collection from sampled plots and large- scale farmers. • Survey123 is used to collect screening data for large-scale farmers. 2.3. Data quality assurance Data quality assurance is achieved through a comprehensive approach, involving enumerator training, continuous data monitoring, supervision of data collection activities, and data cleaning throughout the season. 2.3.1. Training of enumerators Prior to data collection, enumerators participated in a training session held at the NISR training center from November 24th to 28th, 2025. The training covered the overview of the new upgrade of the SAS, data collection procedures and ethics, screening procedures, the plot interview questionnaire content, and the use of data collection applications, including Survey 123, Arc GIS field map, and CSEntry. 2.3.2. Fieldwork monitoring 2.3.2.1 Monitoring attendance and performance of enumerators Effective monitoring of enumerator attendance and performance is vital for ensuring operational efficiency and contribution. During the 2026 SAS Season A, the monitoring system used GPS metadata, capturing location and GPS time, data which differs from device time and cannot be modified by the user. Whenever the enumerator transmits data to the server, the metadata is included, enabling analysis of attendance (including start and end times), data collection locations, and performance metrics such as the number of tasks completed.
21 SAS–2026 A © NISR 2.3.2.2. Attending the sample location and use of high precision GPS The SAS collects data from observation points grouped into square segments measuring 300 by 300 meters. Enumerators were required to collect data within a one-meter distance buffer around each observation point, a specification enforced to ensure accuracy. Any observation recorded outside this buffer is marked as an error and rejected by the central database. For plot area measurement, high-precision GPS units are employed along with correction services, achieving sub-meter measurement accuracy at a 95% and addressing precision-related challenges. 2.3.2.3 Field Monitoring Dashboard A field monitoring dashboard used is an online web application offering a visual representation of real-time data collected from various field operations. It provides a centralized and accessible platform for monitoring and managing activities, resources, and performance in the field. 2.3.2.4. Field supervision To ensure data quality throughout 2026 season A, an intensive field supervision was implemented. The first field supervision was conducted from December 14th to 24th, 2025, and a subsequent field supervision during the harvesting phase, from February 1st to 14th, 2026 by a team of NISR staff. Throughout both phases, supervisors were deployed to all districts to provide continuous oversight and support to field personnel. Their responsibilities included providing technical guidance, monitoring the execution of data collection activities, and ensuring compliance with the data collection ethics as well as completeness of the workload, among others. 2.3.2.5. Data Editing During the 2026 Season A, a monitoring system involving the GIS tools and data editors was used to ensure quality assurance. The data collection was monitored using dashboard and Google Sheets. Editors conducted daily follow-ups to clean data, identifying and rectifying discrepancies using STATA do files based on logical patterns and feedback from training sessions, aiming to provide a cleaned raw dataset for further analysis. 2.4. Data processing and analysis process The analysis involved several steps from organization of raw dataset, data management, cleaning, checking for outliers and dealing with missing data to ensure the quality and cleaned dataset before tabulation. 2.4.1. Data management process SAS data are collected electronically using tablets and are then transmitted directly to the NISR servers. The data analyst team downloads the data and imports it from CSPro into STATA software for further examination, including checking, cleaning, and tabulation. An exploratory analysis of the dataset is conducted for all variables to assess the completeness of the sample, identifying missing data or incomplete observations. Any identified cases are sent back to the field for verification and completion. Exploratory techniques, including descriptive statistics (such as summary statistics, frequency tables) and graphical methods (such as histograms, box plots, etc.) are employed to detect missing values, incomplete data, and potential abnormalities or outliers within the dataset.
22 SAS–2026 A © NISR 2.4.2. Detecting outliers and dealing with missing values 2.4.2.1. Missing values and duplicates observation During data collection, built-in validation rules within the CSPRo application detect missing, omitted, or skipped variables. Error messages appear on the tablet screen during interviews when enumerators skip questions that require responses. After completing the interview but before sending data to the servers, an error message notifies users if any questions remain unanswered or if duplicate questionnaire IDs are identified. Once data is downloaded from the servers and imported into STATA, the data analyst merges the area dataset with the crop dataset and conducts preliminary checks, cleaning, and necessary transformations before analysis. A do-file is developed to check the completeness of data for screening and plot/harvest datasets. A team of data analysts reviews the data on a daily basis, and any inconsistencies identified are communicated to field workers for correction and clarification. 2.4.2.2. Detecting and dealing with outliers Outliers are assessed for all quantitative variables, including crop production, fertilizer quantity, seed quantity, agricultural input prices, irrigation costs, and other related expenses. Two approaches are employed to detect outliers for variables such as crop production and input quantities, while a single approach is applicable for the remaining variables. The first approach involves comparing the value per hectare of land to the standard optimum quantity provided in the guidelines issued by the Ministry of Agriculture, known as “AGENDA AGRICOLE,” for the same land size. Any values exceeding 1.5 times the standard values are flagged as potential outliers and are subsequently sent back to field workers for verification and confirmation. The second approach utilizes statistical processes to detect outliers. In SAS, various statistical methods such as standard deviation are utilized in combination with graphical methods, including normal box plots, to identify possible outliers within the dataset. 2.4.3. Methods for Estimating Area and Yield 2.4.3.1. Estimation of area Approach The National Institute of Statistics of Rwanda (NISR) adheres to and applies methodologies and guidelines outlined by Food and Agriculture Organization of the United Nations (2017) and the East African Community (2022), regarding area and yield estimation. Among the several methods proposed, NISR has opted to use high-precision GPS devices to measure crop area, due to its high accuracy and efficiency compared to alternative methods. For yield measurement, NISR relies on farmer estimations. 2.4.3.2. Process of measuring the area After the identification of the plot boundaries, the enumerators mark GPS point locations at approximately every three meters, as well as at each corner of the plot, while moving around its perimeter. A polygon is then obtained when the starting and final points are connected. The area is finally computed automatically by GIS software linked to the enumerator’s GPS device, based on the resulting shape.
23 SAS–2026 A © NISR 2.4.3.3. Process of measuring the yield Yield data are calculated by considering both the plot and crop areas, alongside the crop production reported by the farmer within the sampled plot. This calculation involves dividing the total production, converted into kilograms, by the estimated crop areas measured in hectares. 2.4.4. Data analysis The survey data are analysed using STATA software, which provides robust capabilities for data management, including importing, cleaning, merging, and manipulating datasets. These features facilitate the preparation of data for analysis. Additionally, STATA enables the development of tabulation commands and the generation of survey tables, graphs, and charts for inclusion in survey reports. Furthermore, SPSS and STATA softwares are utilized to estimate survey sampling errors, thereby ensuring the accuracy and reliability of the survey results.
Chapter 24 SAS–2026 A © NISR This section highlights the key results of SAS 2026 Season A related to crop area (including physical land use, cultivated area, and harvested area), yield, production, agricultural inputs, and agricultural practices in Rwanda. 3.1. Agricultural land use Figure 1: 2026 Season A - Agricultural land use (in thousands of hectares) Source: NISR, SAS 2026 The total land area of the country is estimated to be 2.376 million hectares, of which 1.376 million hectares (approximately 58% of the total land area) are used for agricultural purposes. In 2026 Season A, 1.043 million hectares were allocated to Seasonal crops, 0.501 million hectares to permanent crops, and 0.084 million hectares were allocated to permanent pasture. (See district details in Table 9).
Total land area 2,376 1,376 1,170 1,176 1,043 501 84 117 12 Agricultural land Arable land Physical cultivated land Area under seasonal crops Area under Permanent crops Area under Permanent pasture Temporary fallow land Temporary widow and paste SURVEY FINDINGS 3
25 SAS–2026 A © NISR 3.2. Crop area, yield and production estimates for major crops 3.2.1. Yield and Crop area for major crops Figure 2: 2026 Season A - Yield of major crops (MT/ha) Source: NISR, SAS 2026 Maize: The national average yield was 2 tons per hectare, with small-scale farmers harvesting 1.9 tons per hectares and large-scale farmers harvesting 4.3 tons per hectare. The cultivated area was estimated at 245,405 hectares, representing an increase of 0.5 % from Season A of 2025 Beans: The national average yield was 700 kilograms per hectare. The cultivated area was estimated at 327,907 hectares, representing an increase of 0.2 % from Season A of 2025. Paddy rice: The national average yield was 4.1 tons per hectare, with small-scale farmers harvesting 2.5 tons per hectare and large-scale farmers harvesting 4.2 tons per hectare. The cultivated area was estimated at 17,209 hectares, indicating a decrease of 0.6% from Season A of 2025. Irish potato: The national average yield was 8.9 tons per hectare, with small-scale farmers harvesting 8.9 tons per hectare and large-scale farmers harvesting 10.2 tons per hectare. The cultivated area was estimated at 55,310 hectares, indicating an increase of 1.5 % from Season A of 2025. Sweet potato: The national average yield was 8.8 tons per hectare. The cultivated area was estimated at 96,217 hectares, representing an increase of 16.7 % from Season A of 2025. Bean Maize Paddy rice Sweet potato Irish potato Banana Cassava Large scale Farmer Small scale Farmer National 13.7 10.2 8.9 8.9 8.8 4.1 4.2 1.9 2.0 4.3 0.7 2.5 11.9
26 SAS–2026 A © NISR Cassava: The national average yield was 13.7 tons per hectare. The harvested area was estimated at 41,628 hectares, while the cultivated area was estimated at 236,357 hectares, representing a decrease of 4.6 % from Season A of 2025. Banana: The average yield was 11.9 tons per hectare. The harvested area was estimated at 112,010 hectares, while the cultivated area was estimated at 267,676 hectares, representing a decrease of 0.3 % from Season A of 2025. 3.2.2. Production of major crops Maize: The production was estimated at 488,622 metric tons, representing a 1.5% increase compared to Season A of 2025. The highest maize production was recorded in the Eastern Province, particularly in the districts of Nyagatare, Kirehe, Gatsibo, and Kayonza, as illustrated in Map 5 (see details in Table 15). Map 5: Distribution of Maize Production by District, Season A 2026 Source: NISR, SAS 2026 Nyarugenge Gasabo Kicukiro Nyanza Gisagara Nyaruguru Huye Nyamagabe Ruhango Muhanga Kamonyi Karongi Rutsiro Rubavu Nyabihu Ngororero Rusizi Rulindo Gakenke Musanze Burera Gicumbi Rwamagana Nyagatare Gatsibo Kayonza Kirehe Ngoma Bugesera Nyamasheke DRC Tanzania Uganda Burundi Ü Maize Production Season A Maize Production in MT National Parks Open Water 0 25 50 12.5 Km 20000 and above 10000 - 20000 5000 - 10000 Less than 5000