The team developed a prototype product which considers the NASA and Partner Agencies SMP/RMP program and verifies the images while also providing the capability for further analysis. The project adds value to the existing EO dashboard program by giving ground source data in the form of webcams and other images, and population movement data for those 120 countries in the SMP/RMP program. This is important because it allows the radar images captured to be compared directly with images on the ground. Our program is designed to provide the longitude and latitude coordinates for each webcam and takes a snapshot of that webcam at exactly 6pm each evening, to correspond with the satellite passing overhead and capturing radar images. Those ground truth images are stored and capable of being viewed retrospectively from the date we began running the program. The project is designed to match the space/time window for each of the 120 countries in the NASA data. Further analysis capability is offered in the form of a prompted input field where questions can be asked around what has been seen in the images, and what implication this has for human behaviour and activities during the COVID 19 recovery phase. The product and analysis capabilities are able to be adapted to suit the requirements of a number of agencies.

Project team: VM Project Leader/Health Scientist (UK), AK Python Coder (India), GF Video/Animation (Italy), IK programmer (Turkey(under 18 yrs))

This project was designed by VM as she had been in rural Vietnam during the start of Covid and then travelled to the UK 6 months into the lockdown. When she arrived in the UK she found that Vietnam was being congratulated for its handling of the pandemic and lockdown process, due to good statistics being produced by the country. As a health scientist, her experience of what had happened in Vietnam made her question how this praise could be given, purely based on statistics. She believes that it is important to understand what is going on 'on the ground' and to understand the people element of the COVID 19 pandemic and recovery as statistics do not always represent the true experience. This project is designed to build evidence and data around countries approaches to lockdown and recovery, and the social and behavioural changes taking place, in place of self-reported statistics.

How we developed this project

VM formed the team and developed the project idea. She then sought out team members through the EO Dashboard channels where she located 5 suitable members by the end of day 1. 3 of those members had no further contact on day 1 or day 2 and had not contributed any work. A new team was formed at the end of day 2 after consulting with the Tech staff of the Hackathon, carrying across the two contributing members and the team leader and adding a new python programmer. One further member dropped out due to personal issues and VM sought a new python programmer over day 3 and day 4, again through the EO hackathon channel. On day 4 a second suitable python programmer joined the team and that brought the team count back up to 5. Communications were carried out via Discord (between VM and the programming team), and the EO chat for dashboard development and videos. Video calls also took place between VM and GF using whereby. By day 5 the team consisted of four members who remained until the end of the project. Project documentation was developed by VM, debug logs were developed by IK, and AK led the programming. GF developed the program introduction video.

PHASE 1:

First CDC and WHO open source data was collected 0938 on 25/06/2021 from https://covid19.who.int/table and https://covid.cdc.gov/covid-data-tracker/#global-counts-rates by VM. Data files are stored at https://drive.google.com/file/d/1S1xQfqSqHiinrYFm32DX0gjr0xd_2JxQ/view?usp=sharing and . Due to the size of the CDC data file, it was not able to be opened and all data was manually entered by VM. These two data sources provided information on the count and rates per 100,000 of COVID 19 infection organized geographically. VM created a spreadsheet collating the CDC and WHO data, which can be found here https://docs.google.com/spreadsheets/d/1CK4LA3zT1U5Y-ksd31ktdLEjsB7Xl4Mi1Z-8I1LRXaI/edit?usp=sharing. The spreadsheet was collated by taking the transmission count per 100,000 of the population from both the CDC and WHO data and averaging it. This averaged data is located in Column H, sheet 1. To further improve the robustness of the data, the average would benefit from being taken across 5 data sets. This would need further time outside of the scope of this project. Once this data was collated, countries were identified according to their level of transmission per 100,000. It was determined by VM that infection rate under 500 per 100,000 population was a yellow alert, infection rate under 100 per 100,000 population was an amber alert, and infection rate under 50 per 100,000 population was a red alert. These alerts can be increased or decreased to widen or reduce the size of the project as resources allow. It is suggested that under 1000 be an appropriate threshold for yellow alert, under 500 for amber alert, and under 150 for red alert be applied for an increased project scope. 37 countries were identified as falling into the yellow alert. 12 countries were identified as an amber alert. 10 countries were identified as a red alert. Following this VM created a copy of the spreadsheet, and renamed it Copy of IDENTIFIERS - SLOWDOWN + WHO CDC collated data, which can be found here https://docs.google.com/spreadsheets/d/10YmvNWQ32jQ1B2oLAb26CKozTn_Joz_luaYMfMKDiOo/edit?usp=sharing. SMP compatibility was then considered and can be located on sheet 3 SMP/RMP Compatibility. The compatibility with the SMP listed countries was assessed using a simple yes/no method through comparison with the slowdown project information. VM used Document: Copy1 of Slowdown Project Information provided by a Subject Matter Expert to confirm the countries were either in the SMP data or were not. Once this was done, the countries which met the criteria of being assessed as an alert AND being a yes SMP country were isolated and added to a final spreadsheet Anomalies for Dashboard Integration which can be found here https://docs.google.com/spreadsheets/d/1fauthlrTNKg0jpjo9IO3lak6EJo4SlRoYEckkLlwHXQ/edit?usp=sharing. This document was provided to GF and Sauro for dashboard integration development and project video production. 

Following consultation with a Subject Matter Expert, it was decided that the project would use two baseline cities and two ANOMALY countries for the purpose of the prototype. VM selected LA and Dallas as the two US baseline cities, and Tanzania (red alert) and Panama City (no alert) for the case studies. Project would drastically increase in capability and potential by focusing on cities rather than countries. It would also increase the accuracy and integration with the slowdown and recovery data collected by NASA, JAXA and ESA. Collating the large amount of data accurately is outside the scope of the current project, due to time limitations. AK was notified of the country and city selections. Saorao and GF were notified of the country and city selections. 

PHASE 2:

VM developed a three-point approach to triangulate image sources and create a picture of the two data sets. The first point uses NASA RADAR images from the SPM/RPM program, the second uses GROUND TRUTH data and the third point uses Facebook STAYPUT data. These were selected to act as a ‘confirmation of truth’.

AK began work with VM on the development of the groundtruth data python program using discord as the communications medium and github to host the program. First AK and VM had an in depth discussion about the program concept and design. It was identified that a Deep Learning program utilizing crowd counting in images would be highly beneficial to the project, but was out of the scope of this project due to a lack of programmers and time limitations. Script which was capable of being adapted for the purpose was identified. AK then located a number of webcams which may suit the purpose of the project and sent these to VM for input. Webcams for all four of the countries/cities were agreed upon. AK sourced cameras https://worldcam.eu/webcams/south-america/colombia/19096-bogota-panoramic-view, https://www.see.cam/co, https://en.world-cam.ru/cams/webcam-bogota-watch-online/the-colombian-capital-webcam-bogota-watch-online/ for Colombia, and https://www.earthcam.com/search/ft_search.php?term=Panama+City for Panama City. Colombia was then changed to Tanzania by VM as a red alert country case study was preferential. AK sourced the link https://www.webcamtaxi.com/en/tanzania/zanzibar/page-beach.html . AK developed initial code for the ground truth data with the intention to automatically capture images from each of the webcams at precisely 6pm to sync with the 6pm timing of the satellite image capture. Some issues with the program were discovered such as the instability of the local cameras, and the appearance of google ads instead of the video feed where cameras that were doing live uploads via google chrome. The Los Angeles webcam used was https://myearthcam.com/starlinetours, which is currently down. The Texas webcam was https://www.earthcam.com/usa/texas/elpaso/?cam=elpaso. AK was joined by a second programmer IK, who worked on installing ad blockers and debugging the program. The code as of midday on day 6 (28/06/2021) can be located here https://drive.google.com/file/d/1HaeEoMKCCfm92E3etKGx4bLgin5Tq7UQ/view?usp=sharing. These cameras have been selected as placeholders. Further cameras can be located and utilized for the project, and can be mapped using coordinates allowing integration with the SMP maps. The use of webcams in this phase of the project has been approached as a demonstration of potential capability due to time constraints. The live code is on Github at this address https://github.com/adityakm24/NasaSpaceApps-EO and is managed by AK, and accessed by VM and IK. On 27/06/2021 the decision was made by VM that the program would run continuously until a manual stop occurred, rather than being a manual capture each day. This decision was made after consultation with a Subject Matter Expert who mentioned that backward retrieved images would be of value to the project. The captured images are currently stored in the Github folder. There is the potential for images to be organized and archived, and available using a date selection to retrieve backwards dated images in the future. This could be integrated with archived SMP images. 

At 1400hrs on 28/06/2021 VM made the decision to pull AK and IK from the webcam programming to focus on EO dashboard integration. A previous member had not produced any work on the EO dashboard after 48 hours, so the call was made to divert the programmers to the Dashboard development. Previous member in question stopped responding to contact and ghosted the team. The remaining team IK, AK, and VM then reviewed the Earth Observation Covid 19 dashboard. We used the link https://earthdata.nasa.gov/covid19/explore/global?map=-73.942%2C42.3994%2C1&layers=co2-diff&date=2020-12-15&lState=co2-diff%7C0%7C0 which was provided to us in the EO Challenge communications. Meetings took place between VM, AK and IK to discuss the planning and capability of the dashboard integration from a programming perspective. Due to issues in accessing the Earth Observation Dashboard information a decision was made by VM for the programmers to begin coding the background program and user operability was discussed. Once this was finalised, a decision was made by VM to use placeholders to represent the RMP/SMP data and the Movement Range Maps due to limited personnel, and problems with the programmers computing resources. With further time and resources the systems and data could be fully integrated into each of the sources within the program. Due to the limited time and resources available, this was put on hold and more essential tasks prioritized. The team was unable to gain complete access to the Earth Observation Dashboard and VM ran the KMZ files provided using Google Earth Pro. AK was unable to access Google Earth via his IPad so the complete integration of the data was not able to be considered further at this point. With further access to the Earth Observation Dashboard, and further resources, complete integration into the dashboard would be a capability. At 1630 on 28/06/2021 the dashboard integration phase began. AK and IK began work on the presentation of the three data sets and the analysis fields linked to the geographical locations. VM continued with project documentation and image sourcing.

Webcams were reviewed by VM for lat/long capability on the evening of 29/06/2021. After a meeting with a Subject Matter Expert VM began analysis of webcams in all 120 countries in the SMP program country list, mapping these manually to their lat/long coordinates and street addresses. This was done through assessment of key landmarks, shops, restaurants and features on the streets. VM is aware there is a 30m accuracy window with the accuracy requirements of the lat/long coordinate. Rome webcam videos were collected by VM with each correlated to coordinates. 

Europe

Italy

Rome

http://www.albergodelsenato.it/webcam.php

Piazza Del Pantheon

Piazza della Rotonda, 73, 00186 Roma RM, Italy

41.89920831448809, 12.477381619964236

https://www.skylinewebcams.com/en/webcam/italia/lazio/roma/fontana-di-trevi.html

Trevi Fountain

Corner of Via del Lavatore, 44, 00187 Roma RM, Italy

41.900851859131755, 12.483836881601148

https://www.skylinewebcams.com/webcam/italia/lazio/roma/piazza-di-spagna.html

Spanish Steps

Piazza di Spagna

41.905635042515755, 12.482185307346164

https://www.hotelhasslerroma.com/en/il-palazzetto-footer-pages/webcam

Spanish Steps

Hotel Hassler Roma

41.90604077354032, 12.483649793465002

https://www.facebook.com/197761626911992/videos/230567388684465

Colosseum

These locations were provided to IK and AK to add to the python camera capture program. Further webcams were sourced for the list. The webcam lat/long coordinates list can be seen here https://docs.google.com/spreadsheets/d/1I0zBk9Egu_9AQroBe6LQF0GWf9byWcG_tFu9v2cNgbs/edit?usp=sharing. More time is needed to complete the full collection of 3-4 webcams for each of the 120 countries. For the purpose of the hackathon, the project provides an example selection of what can be done and what is available. A subject matter expert provided KMZ files for Rome. VM passed these files to IK and AK for integration into the python 6pm image capture. 

VM provided AK and IK with the facebook open source population location data. This can be found here https://data.humdata.org/dataset/movement-range-maps. IK began the production of a geographically organized list of the data for the comparison process. Due to the size of the data the team did not have the resources to analyse or model the data, and VM made the decision to narrow the selection to Rome to match with the case study.  

PHASE 3:

Integration of the three options into the dashboard program began on 29/06/2021. The code has now been developed for the three options to be shown side by side. Due to technical errors with Google Earth and other software the decision was made by VM that the .TIF files for each of the cities would be used as placeholders for the real SPM data. The team then worked to incorporate these files into the program. The final decision was made by VM that the cities for the project prototype would be Rome Italy, Houston Texas, LA California and Panama City Panama. The following webcams and lat/long coords were provided to the programming team.

USA

California

Los Angeles

https://www.earthcam.com/usa/california/losangeles/hollywoodandvine/?cam=hollywoodandvine Hollywood and Vine, View of Vine Street 34.100719613940214, -118.3264301744061

https://www.earthcam.com/usa/california/losangeles/hollywoodblvd/?cam=hollywoodblvd

Hollywood Blvd 34.102116361103114, -118.33937689580797

https://www.westland.net/beachcam/

Venice Beach sidewalk, 1401 Ocean Front Walk, Venice, CA 90291, United States, 33.988150293650065, -118.47427467441031

USA

Texas

Houston

https://worldcams.tv/united-states/houston/city-views

Downtown Houston, 3310 Bissonnet St, Houston, TX 77025, United States, 29.725834403165607, -95.42901026106763

https://www.webcamtaxi.com/en/usa/texas/houston-downtown.html

Downtown Houston

AIG life building, 2727 Allen Pkwy, Houston, TX 77019, United States, 29.759503034716975, -95.39472730339293

Cameras with coordinates were integrated into the code by AK and IK. Tanzania and Colombia were completely removed from the code. IK debugged the new code and AK developed associated files. Issues were experienced by AK and IK with computers overheating when running the code, this took substantial time and effort. At 20.31 GMT+1 (1am local time) IK finished the code with only a few slight issues remaining. The code is running and is performing as expected.

VM and GF met at 1854 on 29/06/2021 to discuss the video production. It was decided that the video would not have narrative, and show the step by step use of the system. VM provided GF with a video manually recorded with her mobile of google earth zooming in to Rome. GF was sent the TIF files of Rome, and VM took screenshots of the Rome webcams and sent those. VM prepared a video example of the Rome data from the population movement data and sent it to GF. GF continued with preparation of the video.

IK proofread the project documents VM has prepared. KM signed off the project at 2100 GMT+1.

VM prepared project documents until 2300 GMT +1.

VM and GF worked on the video until 23.41 GMT+1, VM proofread and after corrections uploaded it onto the team google drive with the following link https://drive.google.com/file/d/1n1hCyv3_vPhA_U06lz9xuWPGcJhtqpX8/view?usp=sharing. VM tested all links and URLs in the project documentation.

Project was submitted to EO Dashboard Hackathon at 23.47 GMT +1.

The SMP/RMP data collected by the EO Covid 19 team was the foundation for the work done in this project. The 120 countries monitored by Nasa and its partner agencies were used to sort the WHO/CDC data and the alert system applied to these countries. The Radar images collected by Nasa and its partner agencies were used for side by side comparison with ground truth data in the form of webcams and images. Facebook Population Location data, available as an open source data set, was used as a further comparison to the Radar images, providing a digital check-in component to the comparison process. This data provides the record of where people are and when they are there. This is useful when considering the subtle changes visible in the NASA and Partner agencies Radar images. 

https://drive.google.com/file/d/1n1hCyv3_vPhA_U06lz9xuWPGcJhtqpX8/view?usp=sharing

Code for the webcam capture is located on Github and can be found here https://github.com/adityakm24/NasaSpaceApps-EO. 

The program could be integrated through an interactive map option which will bring up an alert of yellow, amber or red, or no alert for each of the 120 countries. When clicking on the alert, the dashboard will take you to a map of the cities monitored by the SMP program. When a city is clicked on, the three data options for that location will be brought up side by side; 1. The RADAR images, 2. the webcams with corresponding lat/long coordinates and locations within the city, and 3. the population location data. The three data options will have the capability to play back as required. While there are several cameras located with 365 days recording, the majority of cameras are livestream and have recorded images available since the project began running the script and image capture started on 27/06/2021. Following the analysis of the three data options, an interrogation field will allow entry and recording of analysis data. The questions asked could include

• Were there cars visible in car parks?
• Was new construction evident? 
• What is the population count in the image? 
• Are there groups of 3 or more people in close proximity?
• Did you observe loitering?

These questions would be adaptable to the interests of Nasa and the partner agencies utilizing the program. 

The SMP/RMP data collected by the EO Covid 19 team was the foundation of this project. The 120 countries, specifically the cities, monitored by Nasa and its partner agencies were the criteria for sourcing relevant webcams and images, as well as the corresponding population movement data. The question was asked as to how this data could be adapted or verified to increase its value to users and partner agencies.

Below is a list of the data used in our project:

https://data.humdata.org/dataset/movement-range-maps

https://covid.cdc.gov/covid-data-tracker/#global-counts-rates

https://covid19.who.int/table

Webcams:

https://www.webcamtaxi.com/en/usa/texas/houston-downtown.html

https://worldcams.tv/united-states/houston/city-views

https://www.westland.net/beachcam/

https://www.earthcam.com/usa/california/losangeles/hollywoodblvd/?cam=hollywoodblvd

https://www.earthcam.com/usa/california/losangeles/hollywoodandvine/?cam=hollywoodandvine

https://www.facebook.com/197761626911992/videos/230567388684465

https://www.hotelhasslerroma.com/en/il-palazzetto-footer-pages/webcam

https://www.skylinewebcams.com/webcam/italia/lazio/roma/piazza-di-spagna.html

https://www.skylinewebcams.com/en/webcam/italia/lazio/roma/fontana-di-trevi.html

http://www.albergodelsenato.it/webcam.php

KMZ and Tiffs from the RMP program:

Rome_S1_TA117_SPM_20200103-20200127_20200315…0502_th-0.175.tif

Rome_S1_TA117_SPM_20200103-20200127_20200315…502_th-0.175.kmz

Rome_S1_TA117_RPM_20200315-20200502_20200719…806_th0.2.tif.kmz

Rome_S1_TA117_RPM_20200315-20200502_20200719…0200806_th0.2.tif

PanamaCity_S1_TA121_SPMraw_20200103-20200127…_th-0.275.tif.kmz

PanamaCity_S1_TA121_SPMraw_20200103-20200127…0502_th-0.275.tif

PanamaCity_S1_TA121_RPM_20200408-20200502_202…06_th0.25.tif.kmz

PanamaCity_S1_TA121_RPM_20200408-20200502_202…200806_th0.25.tif

LosAngeles_S1_TA137_RPM_20200404-20200428_202…1_th0.225.tif.kmz

LosAngeles_S1_TA137_RPM_20200404-20200428_202…00621_th0.225.tif

LosAngeles_S1_TA64_RPM_20200411-20200429_2020…2_th0.175.tif.kmz

LosAngeles_S1_TA64_RPM_20200411-20200429_2020...0062_th0.175.tif

Houston_S1_TD143_SPM_20200101-20200215_20200…1_th-0.15.tif.kmz

Houston_S1_TA034_SPM_20200101-20200215_20200…200431_th-0.2.tif

Houston_S1_TD143_SPM_20200101-20200215_20200…00431_th-0.15.tif

Houston_S1_TA034_SPM_20200101-20200215_20200…31_th-0.2.tif.kmz

#Covid19 #EOdata #Covid4Radar #humanbehaviour #socialimpact

This project has been submitted for consideration during the Judging process.