Being a company that is based in Equatorial Africa, and is using satellite imagery to monitor other parts of the world, its amazing how we can’t get any good imagery for our own backyard. We are building products around satellite imagery and we couldn’t exploit this great data source to our advantage because most of the images for Uganda and most of East Africa is covered in cloud, most of the year. Enter cloud masking tools, SAR and making due with what we can get and drawing conclusions basing on that.
Satellite imagery is one of the sources of “Big data” and also a great avenue for collecting data without massive funding. As Geo Gecko, we think that if properly used, this dataset can be a very good source of insight on what is going on in East Africa hence our interest in exploiting it. We started working with satellite imagery about 3 years ago and the biggest drawback that we experienced and still experience is cloud cover. It would make it impossible to get information since for most of the year, the areas in East Africa are covered in cloud. This information was skewing our analysis and made it difficult for us to draw any logical conclusions for stakeholders.
The first attempt we made for dealing with this was applying the cloud masking tools provided for Google Earth Engine, among other sources and this at least took away the issues with very high spikes in values as a result of cloud cover. We were then able to make our more logical conclusions from the analysis that we ran. This worked for some applications, but we ended up losing a lot of insight from the areas that were taken out.
Enter SAR. So, after working with the masking tools and identifying their strengths and short comings, we decided to dive into the world of SAR. SAR was, and I think still is, the “unicorn” of satellite imagery sources. We tried to adapt SAR for some of our applications and it worked but to a certain extent, as is the case with almost everything.
In this presentation, we will talk about the strengths and shortcomings of these different approaches that we used to see through cloud, what applications they were best suited for and also the applications for which they failed. We shall showcase use-cases specific to our applications as Geo Gecko to make our arguments on how best to deal with this “Cloud” issue.