INSIGHTS ON POWER DELIVERY USING SPATIAL-TEMPORAL ANALYSIS OF NIGHTLIGHTS
Using night light imagery acquired by NASA (VIIRS) as a proxy, one can estimate the trends in electric power delivery. Below is a nine-month analysis of nightlights (NL) at monthly intervals, from August 2017 to May 2018. Hurricane Maria hit the island of Puerto Rico on September 20, 2017. Below are three maps.
1) changes in the light intensity before versus after the hurricane,
2) Percent reduction of lights measured in October 2019, mapped at 1km spatial scale,
3) impact rating measured by percent reduction multiplied by population density at the census tract level.
Watch the introduction demo by Masoud Ghandehari
CHANGES IN NIGHTLIGHT INTENSITY BEFORE VERSUS AFTER MARIA
Move the slider back and forth to see the changes in Night Light radiance at 500-meter resolution, comparing before versus after Maria.
RESILIENCY METRICS (percent reduction and percent recovery of lights)
Below maps show the percent loss of capacity immediately after the hurricane, followed by recovered capacity after 6 months, each at 500-meter resolution. Follow the below steps to work with the interactive map.
Click any point on the map to see the time series data on night light for nine months August 2017 to May 2018. There is a total of 43,200 plots where the vertical axis is the actual light intensity (nanoWatts/cm2/sr).
The corresponding histogram shows the distribution of the information shown on the map, where the horizontal axis represents % values and the vertical axis represents relative probability density of occurrence. You can also select a desired range by moving bars to lower bound and upper bound) and see the map corresponding to the selected range of percent loss or percent recovery.
Percent loss of capacity
Recovered capacity (capacity measured in February 2018 vs August 2017)
People-Day without power
To establish a quantity that refers to the number of people impacted by the power loss the population density of each pixel was calculated using census population density data at the block group level and multiplied by the resilience triangle area (which accounts for the total number of days without power at that pixel).
People-day without power = (0.5 X % loss of power X days to reach full capacity) X (population density)
The infrastructure systems explicitly modeled in this research are power and water networks (hydrological to capture flooding, water distribution, and waste water to capture limits to removal rates) along with delays caused by disruptions in the cellular communication networks (cell phone records). Both power and water are conveniently modeled as distribution networks that move or transform materials to meet demand. A network model of the electric grid will be developed based on existing electrical power generation plants in Western PR and of the actual layout of interconnecting high voltage and distribution lines.
Water Distribution and Watershed Modeling
In modeling the hydrological processes and connection to the water supply system, two components will be developed and integrated for all the 30 nodes in the western region of the island. The first component is the “Watershed model” and the second element is the “Water supply and distribution model” There are two important uses for these models: one being to inform the network model, and the other for future scenarios impact assessment.
Watershed model: A watershed model will be used for the study area to reproduce the hydrologic response during and after Hurricane Maria. The model will produce useful information because the monitoring systems (stream flow and weather) were knocked out by the Hurricane. In fact, Puerto Rico’s Doppler Radar (NEXRAD) was severely damaged during the hurricane. The National Weather Service (NWS) has provided rainfall data during and after the hurricane with operational satellite algorithm called the Hydro-Estimator. During Maria, NWS estimated a rainfall distribution of 5 inches to more than 35 inches on the island.
Water supply and distribution model: A water distribution system simulation will be performed using a simplified network model for the Western region of Puerto Rico. In the region, the local water authority, AAA, has 196,155 clients, 16 filtration plants, 177 drinking water sampling stations, 238 water storage tanks, 42 deep wells, 9 sewer plants and 121 sewer sampling stations. The sanitary sewer system is limited to the urbanized areas. The vast majority of the study area uses septic systems to treat wastewater.