CitiSat!

Urban populations are growing fast. Cities in developing countries are struggling to accommodate these increasing populations due to limited funding and lack of information about how the city is changing over time. Key decision-makers including city authorities, international donors and civil society are not acting based on timely and/or accurate data.

CitiSat (http://citymonitoring.geogecko.com) provides decision-makers and the public with the critical information that they need to understand what is happening in their city. The data is generated using open source satellite data so is standardized and low cost. The visualization products are simplified to ensure widespread use. The code is open source so it can be replicated in other jurisdictions with minimal effort and cost. The information visualized is green space in the city, building outlines, flood-risk, pollution, urban heat islands and wetlands and how all these indicators have changed over time, with downloadable statistics.

We are responding to the Healthy Cities Challenge: How can urban residents design and live in environments that promote physical and mental health?

The Kampala agglomeration is home to four million people. The city is affected by uncontrolled growth in unfit areas (encroaching on wetlands, across rail tracks) shaped by limited urban planning. This leads to a sprawling city with lack of service delivery on utilities, security, transport, education and health facilities and little green space left. In addition, increasing use of vehicles, burning of trash and in industrial areas in the city lead to increasing levels of air pollution. All in all, this is not a healthy environment to live in and it disproportionately affects young people who are moving to the city to find work.

The majority of cities in Africa are in a similar situation. Over 520 million people on the continent now live in cities and this is set to double by 2050. CitiSat helps authorities plan better and shows them and residents what areas are most problematic and need attention, in that way creating healthier cities. 

High-level decision-makers in city authorities are the primary users. They will use it to monitor change on an ongoing basis from a macro perspective, identify issues early (e.g. wetland encroachment), optimal deployment of limited resources and impact evaluation. Secondary users are the public and civil society who use it to stay informed about the status of their city and hold those responsible to account. The full portal is initially developed for the complete agglomeration of Kampala, including its satellite cities.

City dwellers lives are impacted by cities being planned in better ways, increasing service delivery to poorly serviced areas, curbing pollution and loss of green space. In addition, city dwellers have a way to keep their governments to account regarding the environment they live in.

Geo Gecko regularly works with the Kampala Capital City Authority and will develop the first version of the portal in conjunction with them based on their needs. Test versions of the portal will be shown to members of the public to gauge user friendliness.

CitiSat is a portal (http://citymonitoring.geogecko.com) which provides decision-makers with the critical information they need to understand the changes underway in their city in relation to flooding, change in green space coverage, heat islands, construction and wetland change.

Our product is based on the freely available sentinel satellite data, that is provided by the European space Agency (ESA). The datasets are basically raster images in tiff format and can, depending on the AoI, be scores of GB in size. This size is the reason why we want to store and process everything in the cloud. Training datasets will be collected using smartphones equipped with Open Data Kit. This will be in CSV format and stored on the server. It will detail the location, the land use at that site, the time etc. It will be collected at one point in time. We will also use higher resolution satellite imagery for training data. Sentinel satellite data is available through the Sentinel Hub. The data is collected every five days and is available shortly after capture.

We use running models (built in python), that download satellite data and convert it into logical data. Once downloaded, we use artificial intelligence to extract the relevant data, analyze it and visualize it in a web portal. The raw data can also be downloaded or accessed via an API should subscribers/users wish to use this tool in other systems.

For the green space analysis, we are using a segmentation models, that categorize the different pixels into either being a green space or a built-up area. We use a model to detect pollution (NO2) in the atmosphere by using satellite data. The model calculates the amount of NO2 in the atmosphere and outputs the data for further analysis and logical interpretation by the user. Python and open source tools give us the option of scaling up the project and improving along the way, which is the plan with this funding opportunity.

These models require constant and ongoing refinement. This will be done in various ways. Firstly, we will release the models to the open source community so we can benefit from the contributions of specialists around the world. Secondly, we will integrate newly available datasets to be used as training models and test our results if and when they become available. For example, pollution sensors are being frequently constructed in Kampala.

We take technology and data that is normally only available at high cost, combine it for more insight and deliver it in an understandable and accessible format. In that way, we reach a new group of users that currently have little to no access to this information.

What already exists are heavy city Geographical Information Systems and Google Maps. The former base themselves on data collected by the cities, such as transport routes and addresses of citizens. Such data is hardly collected in African cities. We make a city's situation insightful by using data that is already being collected via satellite, creating instant insight. Google Maps offers high resolution imagery on which buildings and vehicles are visible. Its limitation is that it can just be visually inspected and not processed for analysis, such as to show pollution levels or change in green space.

Our core technology is processing satellite imagery and its related data via cloud computing using machine learning and automating this using algorithms so that it auto-updates and can easily be deployed to new areas. We use the ESA Sentinel satellites that offer the largest publicly available number of spectral bands, thus allowing most analysis.

We subsequently visualise this in an online portal showing green space in the city, building outlines, flood-risk, pollution, urban heat islands and wetlands and how all these indicators relate to each other and have changed over time, with downloadable statistics.

We have coded part of the algorithms needed and are working to improve these and to add to them. The visualisation tools have been built and time will be spent on improving them.

The direct output of CitiSat is twofold: key decision-makers receive the information needed to improve urban living environments and members of the public receive the information needed to hold the authorities accountable. The outcome of this is that urban planning will improve, thus the urban living environment will improve. 

We know from our experience in working with the Kampala Capital City Authority that the information we have given them via earlier versions of the portal, mostly related around green space changes and building outlines created new insights. Whether these insights have improved decision-making is yet to be established as many factors are at play within KCCA.

We also know from experience that civil society organisations, such as Slum Dwellers International have found the information shared previously insightful. It is a well-established fact that information in the hands of CSOs empowers them to hold their government accountable.

Figures represent the populations of the urban areas that will be monitored by CitiSat:

Current: 1.5 million

In one year: 8 million (assuming the full Kampala agglomeration and a second city are added)

In five years: 128 million (assuming 32 cities across Africa are being monitored)

Our goal is to grow to 32 cities across emerging markets using CitiSat within the next five years. We plan to reach this by initially creating the product for Kampala and then offering it to other cities on the African continent. The software is open source and can thus be implemented by anyone. We plan to have a low level of advertising on the site to cover maintenance costs which can be avoided by any city authority by paying a monthly fee. We offer our services to any city authority or donor agency that has specific requests and requires a tweak in the software.

We will continuously monitor how the portal is used by the city authorities and civil society organisations to ensure we keep the portal relevant to them.

The most pressing current barrier is financial as we currently have no funds to develop this further and are thus doing this in company R&D time. We intend to resolve this barrier within the year by bringing in funding from donors and/or investors.

Another barrier is how to effect change within city authorities. We know we can create better insight, but it is not within our power to influence decision-making, such as to put restrictions on imports of old vehicles.

We will overcome the financial barrier by applying for different funds, via contests and directly via donors. In addition, we are starting a round of investment as a company this fall for several satellite processing products and services that are under development. In the meantime, we are using R&D time within our company to continue CitiSat development.

We intend to overcome the barrier of affecting decision-making by making the information we create publicly known. It is not our role, as Geo Gecko, to lobby for different political decisions. However, the information we create empowers civil society, and also Members of Parliament, to hold their government accountable.

9 full-time staff members

The Geo Gecko team understands the opportunities and challenges that working with geo-intelligence in emerging markets offer as we have been working in East-Africa since 2013. We are a multi-disciplinary team with experience in the the development/humanitarian and private sector. Our unique strength is processing satellite data into information while understanding the context on the ground. 

We have a strong commitment to open source software that can quickly and easily be deployed in other locations and a strong commitment to accessible and understandable products and services.

The Kampala Capital City Authority is a former client of ours with whom we continue to engage. We ask them if they have interest in any newly developed products and ideas. They have already shown an interest in the further development of their initial portal into CitiSat.

Our business model is to run and host CitiSat for city authorities and donor agencies based on a monthly subscription model within which we guarantee weekly updates and support. The software is open source, so anyone could implement it. Should this happen, we will still gain income from low-level advertising on these versions. In addition, we will offer tweaks to the original portal for a fee.

Initially, we will need funding or investment to develop the product to a full version. Subsequently, we will bring in revenue from city authorities and some low-level advertising on the open source version from which we will fund maintenance, ongoing support and development of new features.

To be completed.