I wanted to create an easy way for government agencies to analyze impacts on their climate action plans after natural disasters. For this use case, I examined the effect of the Camp Fire in Paradise, CA, and some of its climate-related consequences. I hope to develop this application further so that any government agency can use this to make data-driven decisions on impacts to their climate action plans.
About the map
The map shows vegetation health change from April 2018 (pre-burn) and April 2019 (post-burn) in Paradise, CA. This is displayed using the Normalized Difference Vegetation Index (NDVI) where darker green areas are healthier while lighter green areas are less healthy. I then created a difference image between the two NDVI data sets to show the areas of change. When looking at the map, pay particular attention to how well the change matches up with the official burn area. NDVI is used in part to calculate the damage a wildfire causes. If you're interested in how satellite imagery can be used to monitor earth phenomenon go check out what the awesome folks at Development Seed are doing.
After having an output image, I converted it into a polygon whose area could be estimated. Using this estimate in acres, I determined the change in healthy vegetation, CO2 uptake loss, and cars that used to be taken off the road. Referring to the image above there was a 75% loss in vegetation. CO2 uptake measures how much of the molecule is sequestered out of the atmosphere. After the fire, the forested area went from about 140,000 metric tons/yr to 36,000 metric tons/yr. This is the equivalent of putting 23,000 cars back on the road. If you're interested, go check out my open source version of this. My goal is to create an online tool that any local government agency can use to estimate the impact of natural disasters to their climate action plans.
Here's the list of technologies and data I used to create this map and the data it displays. You can click on each element and it will take you to a page for more info:
- GDAL - Python implementation: GDAL was used to convert the Landsat-8 data into numpy arrays in order to calculate NDVI.
- ArcGIS Online Great for deploying maps this hosts the NDVI, burn perimeter, and Paradise town boundary.
- Landsat-8: A massive archive of past and current remotely senses data that is at the global scale. Amazing this is made available for free for anyone to use!
- Cal Fire: From what I've read NDVI is used to approximate the burn area of fires. I wanted to compare my results with the official data set and show you how closely the two match.
There are four layers in the map. The first is the Paradise town boundary, which is in yellow. The second is the fire perimeter and shows the area the fire burned. The third is the 2018 NDVI, which shows healthy vegetation in April 2018. Fourth is the 2019 NDVI which shows healthy vegetation in April 2019. Last the NDVI difference image which shows the change between 2018 and 2019 NDVI. It's a more natural way to see the difference between the two years with areas red experiencing the highest decrease in vegetation health.
The Landsat-8 satellite captures the entire earth in multi-spectral imagery every 16 days. With this frequency and the stack above the possibilities are nearly endless. From mapping desertification to predicting famine in Africa remotely sensed data plays a critical role in the management of climate and humanitarian disasters around the world.
I wanted to make a few comments on the map above. There's a lot more that goes into creating useful data that decision-makers can use to make good choices on policy. Some of the factors that can affect an analysis like the one above include cloud cover, time of year, and context on the ground that isn't readily identifiable from space. This was a basic look to illustrate what's possible with the current technology and methodologies. If you want to learn more read The use of the Normalized Difference Vegetation Index (NDVI) to assess land degradation at multiple scales: a review of the current status, future trends, and practical considerations.
Here's a list of some the resources that I used to help me make this map:
NDVI Tutorial - Hatari Labs
Thanks to Hatari Labs for helping me with their easy to follow tutorial on creating NDVI. It showcased how to do this with QGIS but I was able to learn it there and then apply to GDAL.
NDVI Calculation Using Python GDAL - José Gómez-Dans
Thanks, Jose, for getting me started on the structure of my code. It was a good starting to point to start testing the functionality of GDAL!