Satellites and evolving technology could improve fertiliser use efficiency and crop yields for farmers in developing countries.
There needs to be a change in agricultural aid policy if remote sensing is to benefit the developing world and improve fertiliser use efficiency. Farmers need accurate information on the plant nutrient status of their crops, and they need the wherewithal to deliver the nutrients accurately. Technology transfer is the answer, the process of transferring skills, knowledge and technology to enhance activities in a wider sector.
In most of the developing world, farmers lack the means, facilities or infrastructure to obtain the nutrient status of their crops and soils and are forced to use locally available artificial fertilisers and manures. Other than nitrogen, they have no means of correcting a mineral deficiency which may be limiting production. Important food crops, for example rice, are often yield-limited by potassium in areas of South America, Asia and Africa. Small additions of potassium chloride have doubled yields in some cases.
The means to measure nitrogen and some plant-quality parameters remotely, using vegetative indices, has existed for some years. Some of the spectral bands from which the indices are derived are satellite mounted, and the number of satellites, the number of bands and the spatial resolution of the imagery has improved rapidly in recent years.
Most satellite imagery is operated commercially and processed for many uses. In terms of agriculture, some imagery is supplied free of charge – notably NASA provides ortho-rectified Landsat imagery for free. But this imagery needs to be commercially processed to get useful information such as vegetative indices and surface temperatures at a resolution of 30m. Most of the earth is covered every 16 days.
Although satellites offer a means of obtaining remotely-sensed imagery, this information is often of little value when cloud cover obscures the area of interest.
This would limit the use of imagery from satellites passing an area of interest at discrete time intervals of more than a day or two. Should funding allow for imagery from many satellites to be processed, the information gleaned would be limited to predicting regional crop failure or tracking drought, disease or pest infestations.
Disseminating information which would benefit an individual farmer would likely be part of a regional exercise, such as spraying locusts or selecting storm cloud activity for cloud seeding with silver iodide to reduce the impact of droughts.
Predicting the concentration of plant minerals is now possible using hyperspectral remote sensing. These sensors are currently too heavy to be mounted on commercial satellites but they operate perfectly well from fixed wing aircraft.
They provide a data cube of contiguous spectral data from the visible range to infrared through to short wave infrared. Nitrogen can be estimated in the infrared to visible red edge part of the spectra from which vegetative indices are derived, the other minerals can be predicted in the short wave infrared where leaf biochemical activity can be differentiated.
Flying hyperspectral sensors over regions of interest and mapping the nutrient status and crop quality parameters has the potential to improve fertiliser use efficiency and improve yields. Fertilisers would not be applied if they were not needed, and yields could be boosted by supplying minerals necessary to increase crop production.
Hyperspectral sensing also allows for the mapping of soil zones in terms of texture and moisture-holding capacity, as well as salinity in soils and water supplies, which could improve fertiliser application strategies and improve water use efficiency.
In addition, hyperspectral sensors are able to give the quality indices which multi-spectral sensors provide, so mineral status can be studied alongside these parameters. This would indicate what is limiting production and allow comparisons between regions producing similar crops.
Processing hyperspectral data is currently a complicated process which needs to be automated to enable large data sets to be processed in the cloud. Plant tissue samples, which may be dried, will need to be analysed by wet chemistry and by proximal sensors, to calibrate the remote-sensed data.
At the time of writing this is believed to have only been successfully undertaken on New Zealand pasture, through hyperspectral imaging, where the same algorithms have been validated and used on locations hundreds of kilometres apart.
A change in agricultural aid policy is essential if remote sensing is to benefit the developing world. Additional funding is needed for obtaining, calibrating, automating and disseminating remotely-sensed multispectral and hyper-spectral data. Understanding and addressing the factors limiting crop yield offers the best opportunity to maximise the benefits of fertilisers and increase yields.
The transfer of information to the farmer is easy by comparison. Wireless communications by smart devices should allow farmers to benefit from the technology transfer. Once a system is in place to allow the best use of gathering and processing information, the rest is simple.