NASA Harvest Partners with FAO to Improve Agricultural Monitoring Across Malawi, Namibia, and Kazakhstan
In 2022, NASA Harvest partnered with the Food and Agriculture Organization of the United Nations (UN FAO) under a United States Agency for International Development (USAID) funded pilot project called "Improved Yield Estimates to Inform Agricultural and Food Security Interventions. This project is aimed at testing and validating satellite-based tools and products to support agricultural monitoring and improve crop yield estimates by Ministries of Agriculture across three pilot countries: Malawi, Namibia, and Kazakhstan.
The joint project led by Christina Justice (Principal Faculty Specialist, University of Maryland; NASA Harvest Food Security and Early Warning Co-Lead) and Jonathan Pound (Economist, FAO GIEWS) aimed to combine survey data collected directly from the field using the mobile application ArcGIS Survey123 with satellite-based agricultural monitoring tools, products and yield forecast estimates to provide comprehensive information on in-season crop development. In addition to these tools developed by a Harvest team, including Ritvik Sahajpal, Blake Munshell, Mike Humber and Inbal Becker Reshef, the NASA Harvest AI and machine learning team led by Hannah Kerner developed high-resolution national-scale cropland masks for Malawi and Namibia and separately a spring wheat crop type mask was developed for Kazakhstan by NASA Harvest. Such maps provide the Ministries of Agriculture a greater capacity to better target agricultural monitoring and assessments. The updated cropland masks have been made available through the NASA Harvest Data Portal for public use.
The information and tools provided to the Ministries of Agriculture through this project have already proved valuable in decision-making and have been used operationally to support food security interventions in both Malawi and Namibia in 2023. Underscoring the effectiveness, these data and tools are continuing to be used operationally, to support regular monitoring and assessments of in-season crop conditions (See FAO News Article From Soil to the Stars for more details on the project and NASA Harvest/ FAO partnership).
Under this pilot project, the team tested different approaches to crop and yield assessments in the three different pilot countries, tailoring the tools and products to meet the needs of the countries.
Malawi: Integrated approach to operational in-season monitoring and yield estimation
In the case of Malawi, several in-season mobile-based assessments were carried out in addition to two sample crop cut campaigns during the 2021/2022 and 2022/2023 cropping seasons using Survey123 to support the Ministry of Agriculture in their regular assessments. The collected information created robust and novel datasets of crop conditions and yields for training and validation of cropland and crop type maps and yield forecast models from NASA Harvest and the broader research community. The data from these rapid agricultural assessments and sample crop cut exercises is being used in parallel with the machine learning and remote sensing-based yield forecast model GEOCIF developed by Dr. Ritvik Sahajpal from NASA Harvest. The GEOCIF model has now been downscaled to the Admin-2 level in Malawi and early yield estimates provided to the Ministry of Agriculture support in-season crop assessments and better prepare for potential food security shocks.
During the 2021/2022 cropping season, several mobile-based rapid crop assessments were carried out mid-season followed by a Crop Cut Campaign in May 2023 with FAO Malawi and the Ministry of Agriculture across three distinct districts in Malawi to represent the large variations in yield levels across the country (Balaka (low yielding), Mzimba North (medium yielding), and Kasungu (high yielding)). Across the three districts, 523 fields were surveyed over the month-long campaign. This dataset of field-level yields from two 5x5 meter sample plots and farmer survey data is now available through the NASA Harvest Data Portal and represents an unprecedented public dataset of sample crop cut yields aimed at supporting the advancement of yield forecast models by the broader research community.
(Left) Weighing maize harvest from sample crop cut plot in Kasungu District Malawi, May 2022. (Right) Harvesting sample crop cut plot in Kasungu district Malawi.
May 2022 Crop Cut Campaign and field sites across Balaka, Kasungu, and Mzimba North Districts, Malawi.
Certificate of completion ceremony for field data collection technicians for the 2022 Malawi Crop Cut Campaign.
During the 2022/2023 cropping season, an in-season mobile-based rapid crop and food security assessment was planned to be carried out in February 2023 at the national scale by NASA Harvest in partnership with the Ministry of Agriculture and FAO building on the work from the previous season. As the rapid assessment was being carried out, Cyclone Freddy hit the southern region dropping six months' worth of rain in just six days. The team mobilized rapidly to update the survey within a week to cover cyclone and flood impacts on agriculture and food security at the request of the Ministry (See NASA Earth Observatory Article for more details on the Cyclone Freddy Rapid Flood Assessment). In just 5 days over 2,000 farmers were surveyed providing critical evidence on flood damage and impact. From this assessment, MoA, FAO and NASA Harvest produced a joint Cyclone Freddy Flood Assessment Report that outlined the impact of Cyclone Freddy on agriculture and helped inform the Ministries flood impact response across the Southern region.
Joint MoA, FAO, NASA Harvest Cyclone Freddy Flood Assessment Report, May 2023.
During the same cropping season, a Crop Cut Campaign in the Kasungu District of Malawi was carried out during the harvest period in June 2023 in partnership with FAO and the Ministry of Agriculture. A newly developed remote-sensing sampling approach was applied to capture in-district yield variability, ensuring that data was more representative of productivity trends in Kasungu, which is one of the largest maize-producing districts in the country and is critical to national maize supplies.
During the two-week crop cut campaign a total of 261 fields were sampled across Kasungu district, Malawi. The crop cut methodology followed the gold standard crop cut methodology taken from the FAO Handbook with updates to include the collection of geolocated quantitative and qualitative data through the digital survey tool Survey123 and additional measurements including moisture meter measurements and different sizes of sample crop cuts to test the how crop cut size affected sample crop cut yields. This crop cut dataset has now been made available through the NASA Harvest Data Portal.
(Right) NASA Harvests’ Christina Justice training Malawi crop cut technicians on sample crop cut techniques for the 2023 Crop Cut Campaign in Kasungu District Malawi. (Left) 2023 Crop Cut Campaign field samples from Kasungu District.
(Left) Malawi crop cut technicians weighing the maize harvest from sample crop cut plot in Kasungu District, Malawi in June 2023. (Right) NASA Harvest’s Blake Munshell and Malawi technicians harvesting maize from the sample crop cut plot in Kasungu District Malawi, June 2023.
In addition to the field-based campaigns carried out across Malawi during the 2022/2023 and 2023/2024 seasons, a critical element of this project was the downscaling and enhanced training of the machine learning and remote sensing-based GEOCIF model to the Malawi context. Under this work, led by Dr. Ritvik Sahajpal, the GEOCIF model was downscaled to run at the Admin-2 level in Malawi and provided to the Ministry with monthly updates of maize yield forecast estimates during the 2022/2023 season and currently for the 2023/2024 cropping season. Initial comparison to field-based surveys indicates that the GEOCIF model adequately captures the variation of yields across the country.
GEOCIF admin 2 level maize yield forecast estimates as of March 11, 2024.
The GEOCIF maize yield forecast results are now being provided to the Ministry of Agriculture in Malawi monthly during the current 2023/2024 cropping season and are being used to support their regular crop estimates.
From the success of these results, other countries in the Southern Africa region are requesting this yield forecast data to support the response to t
Namibia: In-season mobile-based assessments supporting early warning and early action
In Nambia, the team has been working closely with FAO and the Namibian Ministry of Agriculture, Water, and Land Reform to translate their paper-based in-season assessments into mobile-based surveys that can be carried out by extension officers in the field. This pilot project has significantly increased the data collection capacity of the Ministry and allowed for more detailed and accurate information to feed into their regular assessments that are carried out three times a season. Using mobile-based surveys, the Ministry is now able to collect over six times more data during a given assessment (1300 surveys per assessment compared to 200 during a normal paper-based assessment). In the 2022/2023 cropping season, the team worked in close collaboration to translate the three separate assessments into a mobile-based format using Survey123 that was user-friendly and targeted key information needs at various crop development stages. All the data collected during the assessments were organized on online platforms that allow real-time monitoring capabilities supporting early decisions by national agencies to respond to developing concerns.
Building on the pilot project for the 2022/2023 cropping season, the Ministry has continued these efforts supported by FAO and NASA Harvest into the 2023/2024 season and are currently collecting their regular assessments using the novel mobile-based data collection survey developed in partnership with NASA Harvest and FAO using Survey123.
Namibia 2022/2023 Post Harvest Assessment collected in May 2023.
Kazakhstan: Yield forecast estimates to support national assessments
In Kazakhstan, the needs were quite different than the other pilot countries given the country's status as a net exporter of wheat and high producing country. The team worked closely with the Ministry of Agriculture Kazakhstan and FAO to assess the needs and focused more in-depth on tailoring the GEOCIF yield forecast model. To do this, the project undertook the collection of spring wheat yield from combine harvesters by partnering with Growers Union Kazakhstan. Yield data from combines was collected over 100 fields providing detailed information on in-field yield variability and the field level yield from this data has now been made available on the NASA Harvest Data portal for use by the broader research community.
In addition to the collection of critical yield information, the GEOCIF model was trained for Kazakhstan and the results were provided to the Ministry of Agriculture to support their in-season assessments. National yield estimates for 2023 were below average and corresponded with RS-based agrometeorological indicators, and closely tracked USDA yield estimates. Main-producing oblasts were among the most affected by poor precipitation & high temperatures.
Spring Wheat yield Estimates from GEOCIF and USDA
Kazakhstan yield forecast, Aug 2023 (GEOCIF)
GEOCIF crop condition assessment, Aug 2023
(Top) GEOCIF spring wheat yield forecast estimate for Kazakhstan as of August 2023. (Middle) Kazakhstan spring wheat crop condition estimation based on GEOCIF yield forecast estimates compared to the five-year mean. (Bottom) NASA Harvest agrometeorological indicator graphic for high-producing Akmola Oblast in North, Kazakhstan.
Way Ahead
In just two years from inception, this project has been able to provide critical tools and data products to Governments that are now being used operationally to enhance their regular monitoring and assessments in support of agricultural and food security interventions. This has been possible through close collaboration and partnership with the Ministries of Agriculture and FAO for the pilot countries supporting the advancement of methods for early warning of food production shortfalls. The current season in Southern Africa and the ongoing El Nino have made the provision for these tools and products even more critical in support of early warning and quantitative assessment support interventions. Based on the success of the GEOCIF yield forecast results in providing quantitative information on forecast yield outcomes, other countries in the Southern Africa region are now requesting this information to assess the impact of the current El Nino drought situation and to support their agricultural impact assessments and crop estimates.
For more information on the project activities see the following news articles:
FAO and UMD/NASA Harvest Partnership Strengthens Capacity for Crop Yield Estimation and Food Security Assessment in Namibia's Northern Regions
Freddy Brings Lean Times to Malawi
https://earthobservatory.nasa.gov/images/151289/freddy-brings-lean-times-to-malawi?src=ve
From the soil to the stars. Field surveys and satellites help Malawi's farmers manage crops better and access disaster relief more quickly
https://www.fao.org/fao-stories/article/en/c/1655996/
Proving the Value of Earth Observation Data for Small-Scale Agriculture