In the Spring of last year Dr. Camilo Mora of the University of Hawaii Manoa contacted our team. He wanted to know if we would be interested in developing an interactive web map to display the results of a research project he was leading into the effect of rising global temperatures on climatic conditions resulting in human mortality due to heat stress. We were glad to hear from Dr. Mora again. The year before we had developed a web map to display the results of his research into how climate change could affect the length of plant growing seasons. For this new study, Dr. Mora’s team analyzed 1763 documented lethal heat events and identified a global threshold beyond which daily mean surface air temperature and relative humidity became deadly. Using Earth Systems Models to predict daily values for temperature and humidity across the globe up to the year 2100, they estimated the likely number of lethal heat days annually under low, moderate, and high carbon emissions scenarios.
Although his new research project was still in it's early stages, we found the initial results to be very compelling and we agreed to move forward with the project. Using preliminary data from their research, we explored some ideas for how to present the data and developed a couple of prototype applications. Several months later, we heard from Dr. Mora again. His team had completed their research, and he was ready to share his finalized data with us and to collaborate on the design of the final application. The time-frame was short. Dr. Mora and his team were writing the final drafts of a paper for publication in the journal . So we rolled up our sleeves, reviewed our initial prototypes, explored the finalized data, and then got straight to work.
The application leverages the robust capabilities of the ArcGIS platform to distill complex scientific data into intuitive maps that enable users to interact with and understand the data. This was an interesting development project, not only for it's subject matter, but also on a technical level. So we thought it would be worthwhile to share some details about how we built the application.
The data for the years 2006 - 2100 are based on the average of 20 Earth System Models developed for the Coupled Model Intercomparison Project Phase 5, under low, moderate, and high (i.e. "business as usual") carbon emissions scenarios (i.e. Representative Concentration Pathways, RCPs 2.6, 4.5, and 8.5 respectively). By selecting from a drop-down list of RCPs you can view the modeled results for the different carbon emissions scenarios.
When you click on a location over land, a window appears with a line chart and a scatter plot that reveal further insights into the study results for that location. The line chart displays the trend in the annual number of lethal heat days at the location for each year of the study period. The scatter plot displays the temperature and humidity for each day of the selected year over a curve which represents the lethal heat threshold.
Now let's take a look at some of the deeper technical details of this application. On the back-end of the application are two web services that deliver the data from the study results to the web application for display. These services are hosted on the Esri Applications Prototype Lab's GIS Portal.
An image service provides the web application with the data for the annual number of lethal heat days for each year of the study period. The data source of the service is a mosaic dataset that defines a single point of access to a collection of single-band raster datasets of global extent. Each raster dataset contains the number of lethal heat days across the globe for a given year. For the historical period 1950–2005, the data for each year are stored in a single raster dataset. For the future period 2006–2100, the data for each year are stored in 3 raster datasets – one for each of the carbon emissions scenarios.
The image service has two roles: 1). to provide the images showing the annual number of lethal heat days for display in the web map 2). to provide the data for the graph of the trend in time of the annual number of lethal heat days. To generate the images for the map layer, the mosaic dataset applies a raster function chain that dynamically clips the source raster datasets to the coastlines and applies a color ramp to convert the single-band source data into three-band color RGB output images. To provide the data for the trend graph, the service delivers the pixel values at a given location from each of the historic rasters and from the future rasters for the selected carbon emissions scenario.
A geoprocessing service provides the data for the chart that plots the temperature and relative humidity for each day of a given selected year. The source data for this service are a collection of 36 NetCDF files that contain the daily values for temperature and relative humidity for the study period and for each carbon emissions scenario. Each file contains data for a twenty year period for either temperature or relative humidity for the historic period, or for one of the three carbon emissions scenarios. In total, the files use 17 GB of storage and contain 12,570,624 unique points of data. To build this service, we started by developing a Python script with input parameters for the selected year, the selected carbon emissions scenario, and the coordinates of the location where the user clicked on the map. The script obtains the requested data from those files in four steps:
- The NetCDF files containing the relevant temperature and humidity data are identified from the first two input parameters.
- In-memory tables are created from the files using the Make NetCDF Table View geoprocessing tool.
- Queries are crafted to obtain the temperature and humidity values from the tables for each day of the selected year at the specified location.
- The results of the queries are sorted by day of the year and returned the to client application.
The python script was then wrapped into a Python script tool and published as a geoprocessing service.
The application also includes links three video animations showing the increase in lethal heat days over time for each of the carbon emissions scenarios. These videos were created using the animation tools in ArcGIS Pro. The videos representing rcp2.6, rcp4.5, and rcp8.5 can be viewed here, here, and here. Links to the videos and the source code of the application are also available from the application when you click the "About" button at the top right corner.
In conclusion, we'd like to thank Dr. Mora and his team for their very important research and for the opportunity to contribute in our way towards helping to extend the reach of their findings. We enjoyed working with Dr. Mora and hope to collaborate with him on his future projects.
In the news