Thank you. We will get back to you soon.


Introduction

From the earliest film images taken by the extraordinarily ambitious Corona satellite campaign, to side on views of Mount Fuji from worldview 3, to Europe’s own Rosetta mission exploring far flung rocks, images acquired from satellites will always interest me. As an Earth observation (EO) data scientist at KisanHub, I have the privilege of indulging this interest on a global scale. Remotely sensed imagery offers us the opportunity to watch the world breath, and in doing so provide accurate information as to how healthy it is. Delivering this wealth of information to growers around the world in a timely fashion, and ensuring it's accuracy, is the core of my role.

At KisanHub, using all available data sources is one of the ways our novel software can aid intelligent decision making using machine learning tools. One of the biggest advantages of the machine learning age is that we can assimilate these datastreams with relative ease. Within this post I hope to introduce readers to what growers can expect from EO data in the near future, and how I am involved in delivering these data.

Rapid revisit satellites

A recent boom in EO has lead to something of a space race in the discipline, with companies like planet and earth-i competing to get satellites to low earth orbit at a rate not seen previously. This has allowed scientists to access daily imagery of some parts of the world. I often draw an analogy in weather data, where geostationary satellites revolutionised weather prediction by giving daily imagery on both local and global scale.

In my opinion this will be similar with the current rush to low earth orbit. Daily imagery can be combined with weather information and ground truth logs to refine models in a way not previously seen. The increase in resolution will lead to a boom in variable rate application, optimisation of irrigation services and crop yield forecasting. Making use of all available information is a priority for feeding a hungry planet. At KisanHub, these diverse data can be centralised and leveraged to offer these services, and my role includes assimilating these satellite derived information into existing environmental data accurately and efficiently, as well as using the newest tools and data sources available to assure the best quality product is delivered to growers. On a day to day basis, this involves developing algorithms and software tools to automatically check image quality, and derive useful information from them to deliver to platform users.

Urthecast are developing technology to deliver video from space, which would allow for more robust data to be generated for the agricultural industry. Have a look at London from UrtheCast. 

Aside from the private sector, publically available resources have never been better. The European Space Agency’s Sentinel missions intended to reveal details of the earth’s climate cycles over a variety of sectors. For land monitoring, the Sentinel 2 mission was dreamt up, which, aside from it’s very high image quality, provides color data which can reveal things which the human eye can’t detect, such as where crops are water stressed or where nitrogen content is low. NASA’s Landsat series has provided a similar service since the early 1970s, so EO for crop modelling can be backcast to include these historical data. KisanHub integrate datasets such as these in building yield prediction and crop health models.

ESA are using machine learning to distinguish between crop types in satellite images

Unmanned aerial vehicle (UAV) data

One major drawback of using EO images is that clouds aren't see through. This unavoidable fact leads to gaps in the dataset, which can often coincide with the timing of important events. The silver lining is that machine learning tools allow us to reliably fill in these gaps until we can get an accurate view of the earth's surface, so when the lights get turned off we can still say, with a given certainty, how things are developing.

The KisanHub platform is capable of handling data from UAV, which will undoubtedly play an intimate role in the future of precision agriculture, given that it doesn’t face the limitation of clouds covering the land surface. This is another sector which is undergoing rapid growth, as the demand for UAV services come from a range of industries. When these data are handled correctly, seamless images can be created between the satellite and UAV imagery records, allowing for both datastreams to be used in tandem. Looking to the future at KisanHub, I am involved in planning how best to integrate these resources, and how we can advise growers on what they need to ensure the best data is provided to them.

Summary

My role as an EO data scientist involves working with diverse datasets in order to deliver the most accurate products given the input data to growers on our platform. Given the incoming wave of EO data which will become available on account of the technologies discussed, I believe EO data will become an invaluable tool for growers to not just use, but rely on to make the best decisions possible. Growers globally can look forward to precise and accurate image data in the coming years, to help them estimate yield, get early warnings about pest and diseases changing the health of their crops and save money using variable rate applications of resources on the farm. I look forward to providing these tools to growers both now and in the future.



Call us today +44 1223 276907

Latest tweets

Stay connected

KisanHub apps