Use data to deliver sustainable information

Through data collected by mobile devices, IoT and satellites, the NADiRA services support better management practices for smallholder farmers and provides reliable information to stakeholders.


Increase accuracy and lead time of production.


Improve efficiency in monitoring farm operations that precede the planting period.


Provide information about crop growing conditions and anomalies during the growing season.


Improve the efficiency and reliability of effective harvest date.


Forecast crop yield efficiently and reliably for a large number of parcels.

Crop type identification

Crop type identification helps increase accuracy and lead time of production. Crop mapping and identification provide an important basis for many agricultural applications with various purposes such as yield estimation or soil productivity.

Advantage for the stakeholders

Crop type information is collected by the field agents for the parcels under monitoring. It is necessary to get similar information from the satellites about the parcels in vicinity of the parcels under monitoring in order to identify the “fall-back” parcels and the next addressable market.

Parcel preparedness

NADiRA aims to improve the efficiency in monitoring farm operations that precede the planting period. Based on Sentinel satellite imagery data, NADiRA provides an estimated date of parcel preparation and an estimated extent of preparation.

Advantage for the farmer

Assessing farmer’s readiness to sow helps speed up the release of credit.

Advantage for the stakeholders

The parcel preparedness service aims to verify that the dates of critical events on the agricultural calendar are met by the farmer in order to confirm compliance with good practice.

Short report by Anouk VILLE on the results of the NADiRA EO development

The main objective of the B1 service (‘Parcel preparedness’) is to provide information regarding parcel conditions at the beginning of the season to end-users (farmers, wholesalers, cooperatives managers, bankers, insurers, etc.). It helps to estimate the date of i) ploughing and or ii) flooding in the case of rice crops. Optical Earth Observation (E.O.) data help to provide this information based on reflectance characteristics between a bare soil, vegetation and water. Vegetation and water indices (NDVI, NDWI) derived from spectral band combinations allow the estimation of vegetation biomass or the retrieval of water for a given pixel. Flooding detection is based on vegetation and water indices and a Hue-Saturation-Value (HSV) transformation. Vegetation index and HSV are performed to extract water content information for each pixel, then a thresholding on spectral transformations, coupled to parcel boundaries allow to declare whether given parcel is flooded or not. Ploughing detection is performed based on a change detection between two images. This approach detects bare bright soil to bare dark soil changes suggesting a potential ploughing operation. Figure 1 shows an example of results for flooding detection over the Saint-Louis rice region in Senegal.

Figure 1. Raster output for flooding detection

Figure 2 shows an example of results for ploughing detection obtained over Saint-Louis rice region in Senegal.

Figure 2. Result for ploughing detection between two S2 Acquisition dates

In addition to the flooding event detection, a derived service has been implemented. It consists of a temporal synthesis of flooding products delivered during a season. This service covers only rice crops and is made available at the end of an agronomic rice season (irrigated or rainfed). The purpose of this service is to map rice domain for a given season and map the spatial distribution of flooding dates. Figure 3 gives an example of result for this service:

Figure 3. Spatial distribution of flooding dates (2018 irrigated season at Saint-Louis)

Crop status determination

One of the NADiRA objectives is to provide information regarding crop growth conditions and anomalies during the growing season. From an Earth Observation perspective, this information involves the monitoring of canopy development, NDVI and biomass growth, of the crop cycle and phenology for purposes such as crop calendar management, future crop production estimation, fertiliser use management or weed proliferation monitoring.

Advantage for the farmer

This service helps the farmer anticipate risks on yield and field agents may provide advice to the farmer if needed.

Advantage for the stakeholders

This service aims at the continuous monitoring of the crop conditions to identify potential sudden or progressive deviations in the crop growth, in order to implement corrective actions or initiate insurance claim actions.

Harvest date detection

Harvest date detection aims to improve the efficiency and reliability of effective harvest date observations. This service aims to detect, for each parcel, the date when harvesting happened.

Advantage for the farmer

The harvest date is a key event allowing to estimate the ideal time to market, yield and produce quality in conjunction with crop status information and yield forecast.

Advantage for the stakeholders

Receiving reliable information on harvest dates is crucial for stakeholders, e.g. for bankers to release credit instalments or for insurers to settle claims. 

Short report by Anouk VILLE on the results of the NADiRA EO development

The main objective of T1 service (‘Harvest Date Estimation’) is to provide information regarding harvest operations at the end of the season to end-users (farmers, wholesalers, cooperatives managers, off-takers, aggregators, supply chain managers, bankers, etc.). The process for estimating harvest dates and phenological indicator is based on NDVI time series extracted at parcel level. The shape of the NDVI curves allows the estimation of phenological indicators such as: i) date of green-up, ii) maximum growth, iii) beginning of senescence. The method has the advantage to be robust (with a dense and clean time series) and adaptive for each curve.

Figure 4 shows output fitted curves for four parcels. Figure 4 a) and b) are peanut crop in Kaolack (Senegal) in 2019. Figure 4 c) and d) are sorghum crop in Nigeria in 2018.

Figure 4. Time series of NDVI values over two different crops: peanut in Senegal ( a and b) and sorghum in Nigeria (c and d). The yellow marks indicate the start of green up. The orange marks indicate the maximum NDVI value corresponding to the peak of the growing season. The red mark indicates the start of the senescence period.

Result Integration on agCelerant – AgCelerant web interface was developed to evaluate the congruence between EO processing results and field data, allowing displaying Sentinel-2 image in RGB and corresponding EO result analysis as it is visible on the Figure 5. The following interface allows for each service and each MOI to : – Analyse with precision the field situation for a given date with RBG image, – Estimate if the EO processing result (here flooding detection for irrigated rice in Saint-Louis) matches the RBG, – Evaluate the congruence between the declared agronomic operation date (recorded by field agents) and the estimated date derived from imagery.

Figure 5. AgCelerant dashboard example for congruence analysis

Crop yield forecasting

This service aims to efficiently and reliably forecast crop yield for a great number of parcels.

Advantage for the farmer

The estimation of the final crop yield helps farmers organize their harvest and delivery. Optimum harvest and collection period ensure a better quality of produce. 

Advantage for the stakeholders

Crop yield forecasting and the food security information resulting are essential information for stakeholders, to anticipate risk on credit reimbursement, settlement of claims and organization of harvested produce collection.