Different types of disasters occur in the world, some natural and some man-made. Peru is not exempt from it, and some of the natural disasters that occur in our country are with floods and landslides, mainly due to the 'El Nino' and 'La Nina' phenomenon. This situation motivated us to propose the present work.

FloolBox is a platform that contains two main components: The Checklist and an Easy-to-Use Image Segmentation Tool.

Checklist

Many decision-makers do not have the knowledge and tools necessary to develop a suitable risk reduction and disaster management plan. The FloolBox Checklist is a tool that provides the different steps to follow as well as the links to sources of information, images, and data in general, necessary to develop this plan.

The list of steps of the Checklist was developed, taking different sources of information, among them the Sendai Framework for Action [1], the UNISDR Technical Guidance [2], the proposal of Tchankova [3], flood analysis [4], among others.

• Sendai Framework for Action has seven global goals for 2030 [1]:
• Reduce mortality caused by disasters.
• Reduce the number of people affected.
• Reduce economic losses (GDP).
• Reduce damage caused by disasters in infrastructure and essential services.
• Increase considerably the number of countries that have strategies.
• Considerably improve international cooperation and increase the availability of early warning systems.

The priorities for action are: Understanding disaster risk, strengthening disaster risk governance, investing in disaster risk reduction for resilience, and increasing disaster preparedness in the recovery, rehabilitation, and reconstruction aspects.

The UNISDR Technical Guidance describes the collection and use of uniform and clear indicators of mortality (number of people) [2]. Addresses important aspects of data collection that the Member States should consider to develop a robust methodology to measure mortality.

Our proposed Checklist contains the methodology followed for the simulation, evaluation, and the formulation of a risk management plan, which consists of the following six steps:

1. Selection of the study area.
2. Collection of initial information.
3. Establishment of design costs (flow).
4. Hydraulic simulation in 2D.
5. Establish criteria for the elaboration of maps.
6. Post-processing of the information.

Easy-to-Use Image Segmentation Tool

Being able to identify the areas affected by floods is an essential task for good post-disaster management. In this task, satellite images are of great help, as they give a global vision of the entire area under study [5]. An important task is the segmentation of the image to identify the most affected areas. Traditional methodologies use feature extraction techniques to characterize images, as presented in [6]. Currently, deep learning techniques are used to perform segmentation tasks, as shown in [7]. However, both to extract characteristics and to program deep learning algorithms, it is necessary to have technical knowledge that is often difficult to get. In this sense, the present work proposes using data compression as a method for the task of segmenting. Today, almost everyone has the experience and knowledge necessary to compress files with a simple click. This aspect makes the proposal easy to use and of great importance for people without technical knowledge.

Data compression, which is based on information theory, will be the primary technique for developing the proposed algorithm. When one compresses an image, the compression rate depends a lot on the visual complexity of the image, a visually complex image is compressed less than a visually more detailed image.

In a satellite image, the compression of entire areas could serve as a segmentation technique. 

The steps to image segmentation using data compression are:

• Divide the image into patches of 64x64 pixels.
• Compress each patch.
• Collect the size of each compressed patch.
• Apply a clustering process based on the compressed size of patches.

The compressor used was the JPEG through the imwrite MATLAB instruction. JPEG compression is an image compression technique based on the discrete cosine transform (DCT) [8]. One can have lossy compression and lossless compression.

If the image has only one type of landcover, urban area, for example, one expects to have a similar compressed size of the different patches, but if one finds different compressed sizes, it could be an anomaly indicator that indicates the affected zones.

The final stage is the grouping of the patches according to their size after compression. Unsupervised classification techniques can also be used for this stage. In the present work, the K-means technique was used, a method of grouping k groups in which each observation belongs to the group whose mean value is closest. [9]

In the present work, the MATLAB instruction kmeans was used.

The tools provided for the challenge by the different agencies involved in the NASA Space Apps Challenge were helpful. For our proposal, we use the following resources to define the FloolBox Checklist:

• Hyogo Framework for Action 2005-2015
• Sendai Framework for Disaster Risk Reduction 2015-2030.
• UNISDR Technical Guidance

For satellite images, we had a great variety that offers: Jaxa, NASA, ESA, World Bank, GEOSS, among others.

As tools to prepare and develop the idea, we use:

• Miro (for brainstorming and final block diagram of the solution)
• GoDaddy (to have a domain and hosting for the website of the proposal)
• WordPress (to design the website)
• MATLAB (to develop the algorithm for the image segmentation tool).
• GitHub (as a collaborative repository of our code).
• QGIS (for geographic information system)
• Among others.

Video: https://drive.google.com/file/d/15viXxuZ8HMl9OoYfdqw_IyJmBWOxi3mB/view?usp=sharing

Slides: https://drive.google.com/file/d/1uwKR0SyMlwTh-xciCo-eAcXyC2hPGthY/view?usp=sharing

The results obtained for the different components of the FloolBox platform are described below:

FloolBox Checklist

After applying the described methodology, a complete list of the steps to follow to implement a risk reduction and disaster management plan was obtained. This list includes each step's objectives, the actions to be taken to meet each objective, the necessary tools to use, link to information, and general data that will support the plan, and finally, one can also find useful indications. In the following link, one can find a complete table with the steps and the links to the information necessary to develop each stage:

https://docs.google.com/spreadsheets/d/1MVhC1kYI0mTz9cQ8PGDFWpLXspWSJ4tU48Cup1g3L0A/edit?usp=sharing

FloolBox Easy-to-Use Image Segmentation Tool GUI

For the image segmentation tool, a graphical user interface (GUI) has been developed in which the entire compression algorithm has been programmed. In this GUI, one can enter the patch's size into which the image will be divided for compression, one can indicate the number of classes for results, and one can also save the obtained result as an image.

In the following link one can find the complete code for the developed GUI:

https://github.com/avidrg/FloolBox

Figure 2: FloolBox Easy-to-Use Image Segmentation Tool

FloolBox Website

As a third result, a website for the proposed work has been created. On this website, one can find all the information related to the project, the objectives, the methodology, the results, the contact details, and, mainly, the tools developed such as the FloolBox CheckList and the FloolBox Easy-to-Use-Image Segmentation Tool.

In Figure 3, one can see a screenshot of the website; it can be found at https://floolbox.co/

Figure 3: FloolBox website

1. Aitsi-Selmi, Amina, et al. "The Sendai framework for disaster risk reduction: Renewing the global commitment to people’s resilience, health, and well-being." International journal of disaster risk science 6.2 (2015): 164-176.
2. UNISDR. "Technical guidance for monitoring and reporting on progress in achieving the global targets of the Sendai framework for disaster risk reduction." (2017).
3. Tchankova, Lubka. "Risk identification–basic stage in risk management." Environmental management and health (2002).
4. Avid Roman-Gonzalez, Brian A. Meneses-Claudio and Natalia I. Vargas-Cuentas, “Flood Analysis in Peru using Satellite Image: The Summer 2017 Case” International Journal of Advanced Computer Science and Applications(IJACSA), 10(2), 2019. http://dx.doi.org/10.14569/IJACSA.2019.0100246.
5. Alvan Romero, Nancy, Francesca Cigna, and Deodato Tapete. "ERS-1/2 and Sentinel-1 SAR Data Mining for Flood Hazard and Risk Assessment in Lima, Peru." Applied Sciences 10.18 (2020): 6598.
6. Mohamed, Rziza, and Mastere Mohamed. "A hybrid feature extraction for satellite image segmentation using statistical global and local feature." Proceedings of the Mediterranean Conference on Information & Communication Technologies 2015. Springer, Cham, 2016.
7. Khalel, Andrew, et al. "Multi-task deep learning for satellite image pansharpening and segmentation." IGARSS 2019-2019 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2019.
8. Abuzaher, Mazen, and Jamil Al-Azzeh. "JPEG Based Compression Algorithm." International Journal of Engineering and Applied Sciences 4.4 (2017).
9. Gan, Guojun, and Michael Kwok-Po Ng. "K-means clustering with outlier removal." Pattern Recognition Letters 90 (2017): 8-14.