Sugar cane monitoring and analysis using remote sensing

February 22nd, 2018, Published in Articles: PositionIT

There is a need for more reliable and useful information that can aid farmers and various interested and affected stakeholders in optimising their operations. This study employed remote sensing and GIS techniques to monitor sugar cane crop performance and harvesting on the east coast of South Africa.

The South African agricultural industry makes a significant and essential contribution to the national economy. The sugar industry alone generates an annual average income of R8-billion. Direct employment within the industry sits at approximately 79 000 jobs, while indirect employment is estimated at 350 000 jobs crossing numerous industries. The country’s sugar cane is grown by approximately 24 000 registered sugarcane growers, while the sugar is manufactured by six milling companies with 14 sugar mills operating in the sugar cane-growing regions. It is estimated that the industry produces an average of 2,2-million tonnes of sugar per season [1].

For any farmer, achieving the maximum possible crop yield at the lowest cost is the ultimate goal. On the ground however, farmers are faced with numerous daily challenges that can affect this objective. Furthermore, inefficiencies in the daily running costs of the respective mills in South Africa are a significant cost to industry. The knock-on effect of preventable issues, such as supply complications, runs the risk of mills being impacted negatively and overall profit margins being compromised.

Early detection of problems associated with agricultural crops can greatly assist in reducing financial losses and reaching targeted yield and profit margins [2]. Advancements in remote sensing techniques over the last several decades have improved the use of multispectral imagery as an effective tool in determining and monitoring vegetation conditions, crop stress and yield. These methods are not new to the agricultural industry and have already been harnessed in various countries around the world [3 – 7]. Similar methods however have not yet been fully taken advantage of in Africa, where there is a need for more reliable and useful information that can aid farmers and various interested and affected stakeholders in optimising their operations.


This study employed remote sensing and GIS techniques to monitor sugarcane crop performance and harvesting on the east coast of South Africa. Three different but interlinked goals were investigated in this study:

  • Establishing which fields have been harvested and which are still to be cut.
  • Conducting a monthly crop health assessment for each sugar cane field and providing information to growers.
  • Estimating the crop and thereby forecasting the expected yield.

One of the limitations of remote sensing in agriculture has been the cost of satellite imagery. Similar studies in the past have circumvented this through the use of free imagery such as Landsat 8. Where this is viable for certain crop types, it was soon established that the 30 m resolution capability of Landsat 8 would not suffice for sugar cane. Sugar cane panels can often be too small in extent (less than 0,5 ha) and therefore require higher resolution imagery for credible analysis. A further complication is that successful crop monitoring requires clear imagery with timeous and regular coverage.

A new alternative that has become available in recent years is the acquisition of imagery from the Sentinel-2 satellite constellation. Manufactured by Airbus Defense and Space, and operated by the European Space Agency (ESA), Sentinel-2 gives end users globally the capability to acquire free satellite imagery up to a 10 m resolution at regular intervals. According to ESA “the span of 13 spectral bands, from the visible and the near infrared to the shortwave infrared at different spatial resolutions ranging from 10 to 60 m takes land monitoring to an unprecedented level” [8]. This kind of technological advancement that provides free imagery with noteworthy multispectral characteristics, paves the way for greater opportunities in the field of remote sensing, particularly in the area of land management and vegetation monitoring.

Fig. 1: Growth and harvest tracking: (a) RGB imagery, (b) vegetation indices, and (c) Burn Scar Ratio Index.

Fig. 1: Growth and harvest tracking: (a) RGB imagery, (b) vegetation indices, and (c) Burn Scar Ratio Index.

In this this study, the red, green, blue (RGB) and near-infrared (NIR) bands in the electromagnetic spectrum were utilised, primarily for the processing of vegetation indices and the application of unique algorithms. Furthermore, shortwave infrared (SWIR) bands were used at 20 m resolution to offer the capability of using the Burn Scar Ratio Index. This added another layer of data to the overall models, as burning of sugar cane during the harvest process is common practice in the region. Analysis was carried out over a period of 18 months covering the sugar cane growth cycle of 2017.

Harvest Tracking

Fig. 1 depicts a small sample of several sugar cane panels that were examined: Fig. 1a illustrates a standard RGB acquisition; Fig. 1b shows the results after the application of vegetation indices; and Fig. 1c depicts the results of processing an index to establish the burn scars on these respective parcels of land. What this type of technology and these techniques produce is a product from which one is able to establish growing vs harvested panels and the respective dates that fields were harvested or replanted.

Fig. 2: Growth profile of sugarcane panels (East Coast KZN, 2017).

Fig. 2: Growth profile of sugar cane panels (East Coast KZN, 2017).

Fig. 2 illustrates the quantified numerical data that one is able to obtain from the above methods. Based on the clusters of sugarcane panels examined in these examples, a fairly consistent growth trend occurs for the harvest period of 2017. Employing these techniques to a greater area of extent, one is able to ascertain which sugar cane panels were harvested in the region during the growth cycle of 2017 (Fig. 3).

Crop performance monitoring

This month-to-month analyses adds further value to on-going monitoring of crop performance as it allows farmers to track and understand how their crops perform over time, as well as to compare previous crop cycle trends to establish patterns. Another important factor to consider in this regard is how such analysis has the capability of picking up field anomalies that may not be visible on the ground (e.g. infestations, nutrient deficiencies, weather effects, theft, erosion, diseases, and so on).

Fig. 3: Harvested sugarcane panels (red) for growth cycle 2017.

Fig. 3: Harvested sugar cane panels (red) for growth cycle 2017.

In such instances, farmers can be informed of incidents, and are provided with the opportunity to further investigate and take corrective action. Fig. 4 represents examples of sugar cane panels where possible issues have or are occurring for which there could be legitimate reasons that farmers are aware of, or where further investigation is needed to diagnose the occurrence.

 Fig. 4: Identifying potential issues in crops.

Fig. 4: Identifying potential issues in crops.

Yield forecasting

Traditional forms of crop yield forecasting are not new to the agricultural industry and are used extensively in many countries [9]. South Africa is no exception, with government departments, third party institutions, as well as private sector organisations undertaking such projects. The use of remote sensing technologies to achieve yield estimates is however underutilised on the African continent [10].

In order to estimate future crop yield using remote sensing data alone, one needs to ascertain the relationship between the spectral signature of the crops and their yield. This is achieved by analysing historical yield data of the previous growth seasons and correlating it to the crop signature on a month-by-month basis. Typically crop yield prediction accuracies increase the greater the pool of historical data is to draw from. Due to the fact that the Sentinel-2 image service is still relatively new to the market, this study could only go back one growth season.

Fig. 5: Actual versus predicted sugarcane yield forecast.

Fig. 5: Actual versus predicted sugar cane yield forecast.

To enhance the credibility of yield predictions and increase overall accuracies, it is essential to consider other determining factors that affect crop growth and yield [11, 12]. Information such as meteorological and climatic data, soil properties, and farming practices should be combined with the up-to-date remotely-sensed data in order to model crop growth and develop credible yield estimates. In order to account for the diversity of sugar cane varieties, soil, climate, and farming practices, it is necessary to employ and engineer comprehensive, dynamic crop models that take into account these variables. Incorporating the relationship between spectral information and crop biophysical characteristics enhances crop growth, chlorophyll content, and biomass assessments, and makes it possible to produce more robust and reliable yield forecast estimates (Fig. 5).

Key to the above is the continual improvement and refinement of the methodological frameworks to make them more robust so as to provide the most credible and accurate results to clients. Furthermore, collaboration between technology providers, institutions, and other stakeholders in the industry is essential to harness the most relevant data and process it in a manner that can provide the most value to inform decision-making.


Crop monitoring, growth performance and yield estimation are essential for food security as well as optimal financial returns. With the current rate of technological development, the financial feasibility of undertaking such analysis is growing. As more free or cheaper satellite imagery become available there is greater possibility to produce these solutions in a more cost-effective manner. Where growing human populations and declining arable farming land is a mounting concern, these techniques work towards ensuring food security, as well as mitigating negative economic impacts due to inefficient farming and processing practices.


[1] South African Sugar Association (SASA), 2018.
[2] Al-Gaadi K.A., Hassaballa A.A., Tola E., Kayad A.G., Madugundu R., Alblewi B, Assiri F. 2016. Prediction of Potato Crop Yield Using Precision Agricultural Techniques.
[3] Rahman M. and Robson A.J. 2016. A Novel Approach for Sugarcane Yield Prediction Using Landsat Time Series Imagery: A case Study on Bundaberg Region. Advances in Remote Sensing, 5, 93–102.
[4] Huang J., Wang H., Dai Q, and Han D. 2014. Analysis of NDVI Data for Crop Identification and Yield Estimation. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 7, 4374–4384.
[5] Singels A., Jarmain C., Bastidas-Obando E., Olivier F.C., Paraskevopoulos A.L. 2014. Validating water use and yield estimates derived from remote sensing and crop modelling for irrigated sugarcane in Mpumalanga, South Africa. Proceedings of the Australian Society of Sugarcane Technologists 2014 Conference 36.
[6] Morel J., Todoroff P., Begue A., Bury A., Martine J., Petit M. 2014. Toward a Satellite-Based System of Sugarcane Yield Estimation and Forecasting in Smallholder Farming Conditions: A Case Study on Reunion Island. Remote Sensing Journal, 6, 6620–6635.
[7] Ferencz C, Bognar P., Lichtenberger J., Hamar D., Tarcsai G., Timar G., Molnar G., Pasztor S., Steinbach P., Szekely B., Ferencz O.E., and Ferencz-Arkos I. 2004. Crop Yield Estimation by Satellite Remote Sensing. International Journal of Remote Sensing, 25, 4113-4149.
[8] European Space Agency, 2018. Sentinel-2: Plant Health. Available online at:
[9] Food and Agriculture Organization of the United Nations, 2016. Crop Yield Forecasting: Methodological and Institutional Aspects.
[10] Chivasa W., Mutanga O., Biradar C. 2017. Application of remote sensing in estimating maize grain yield in heterogeneous African agricultural landscapes: a review. International Journal of Remote Sensing, 38, 6816 -6845.
[11] Airbus Defense and Space, 2018. Biophysical Parameters.
[12] Prasad A.K., Chai L., Singh, R.P., and Kafatos M. 2006. Crop yield estimation model for Iowa using remote sensing and surface parameters. International Journal of Applied Earth Observation and Geoinformation, 8, 26-33.

Contact Hal Wooding, Wooding Geospatial Solutions, Tel 031 765-1424,