Satellite data: useful or useless?

The sheer number and scale of technological solutions available daunt even the bravest amongst us. The question, then, is whether they are ego trips, or are they any good in solving real risks and challenges?

Satellite and earth observation data is used by farmers, when consuming their daily dose of weather forecasting. The fact is that a similar combination of data can also improve transparency on agricultural activity and output, as it helps to assess crop conditions and environmental risks.

Using this data, algorithms can even be developed to assess farmers’ creditworthiness. This, in turn, could allow access to the necessary resources to improve production. To make this information as simple as possible, Andries Wiese, Head of Agri at Hollard Insure, spoke with Reinhard Kuschke, CEO of Agriseker, and credited him for the insights and information he provided.

Risk transfer mechanism

The principal risk transfer mechanism in the fruit and field crop sectors is the crop insurance industry.

Over the past decade, insurers have implemented satellite data and derived information to support their objectives of monitoring, planning and post loss evaluation. As with any newly introduced application on existing processes and procedures, it had to prove its value to justify its continued use. As a monitoring tool, one can employ satellite data to track a variety of indices to detect and evaluate conditions which are important to agricultural production: from droughts to the extent of highly destructive hailstorms.

The application of this data source, in its widest sense, is limitless. An array of indices and parameters are available. One can keep track of cropping conditions to evaluate the extent of hail losses or active veldfire events. In the case of fire detection, and hail loss, a relatively simple evaluation can be implemented to determine the extent of the loss. This will give a clear picture of the affected area, but not quite the quantification of monetary losses.

You must be able to determine whether a loss was suffered, due to an insured peril, and to what extent such peril impacted the cropping production. To any user, not just the insurance industry, there must be a direct correlation between what you are measuring and the outcome of the event.

The use of satellite data

Firstly, one should be aware of the basis risk involved in spatial resolution. Basis risk exposes the uncertainty of a value between two adjacent measurements. This was considered as the biggest critique towards implementation of index insurance. An individual farmer will never accept the outcome of an interpolated value, should it not reflect his experience of the event he insured. It did not rain on his fields, for example, but the weather stations around him recorded rainfall events. By just applying the index, it looks as though it rained on his fields - which is not true. The use of satellite data will overcome this problem to a certain degree, depending on the spatial resolution of the data and the pixel size of the data sensor.

 Secondly, the current offering in satellite data is vast. This number multiplies when the spectral capability of the sensing platforms is considered. You need to evaluate the type of data available to quantify your need or objective. Spectral bandwidth available, can provide virtually any solution. This ranges from determining soil moisture, evaluating field size and burn scars to near infra-red and infra-red sensing to evaluate plant growth and by combining various bandwidths. The variables required to run crop growth models can also be achieved.

A third consideration should be the temporal resolution of the return in data. This is the intervals of receiving sensed data. Some platforms provide data sub-hourly, such as meteorological satellites evaluating fast moving weather systems. Some platforms will only perform a bypass every 10 days, and some take even longer.

Evaluate against benchmark

All environmental data needs to be evaluated against some benchmark. If it is an absolute value, then it needs to be calibrated against a standard. If it is an index derived from data acquired across various platforms, then it also needs to be evaluated against its own history to derive useful statistics. By converting radar data to a soil moisture index (as an example) and evaluating this value over a period will help determine whether a dry period is at play, or not.

Lastly, in raw format, none of this data is useful. Reinhard stresses the basic difference between data and business intelligence. The data must be converted into information which the end-user can interpret. This process requires highly specialised data scientists, with access to computing facilities, which can process huge amounts of data on the fly by applying various algorithms. Many agencies have a boutique of shelved products, ready and pre-processed, which are accessible to the public.

It is worthwhile to evaluate the considerations above when one needs to appraise the value proposition of using satellite data as a monitoring or evaluation tool. The success rate of risk management and mitigation, in this regard, is proportional to data quality and its ability to help you address the problem. Andries Wiese Head Hollard Insure.