Observing the Caribbean from Outer Space: Light Intensity
Traditionally the indicators for Economic Activity have been in some form or fashion, GDP, GDP per capita and other National Output/Income Variables. However, I am quite sure any economist or analyst can tell you they have at least faced one of the following issues.
- Data Availability- You are looking at GDP data for a particular country and then in the middle of the series there is a sizeable gap or maybe there is a large lag between data publications. (i.e every 5 years)
- Data Consistency- When you do get all the desired data, you begin to scratch your head. Are these countries even measuring GDP effectively or in a comparable manner?
- Lack of Grid-level data- Finally you decided upon a workaround for the data, at least on a country-level. You go to compare economic activity in the capitals of each country and guess what, there is no subnational level data!
Light Intensity Approach
In simple terms, light intensity is basically brightness which is measured as the rate at which light energy is delivered to a unit of surface. Great now that that’s out of the way, lets dig in. Light Intensity in countries have been shown to be useful for measuring urbanization, densities and economic activity, among other things. Although there are many other established approaches to measure the previously mentioned, in developing countries such as those in the Caribbean they are either unreliable or non-existent. In this regard, light intensity data brings to the table some very useful properties, such as:
- The data collection process for light intensity is consistent and centralized, making it globally consistent and comparable across countries.
- The light intensity data has a Spatial Resolution of ( with a few modifications)
These two properties alone allow for endless possibilities in both the Micro and Macro realms of analysis.
Collection of data:
The US Air Force’s Defense Meteorological Satellite Program Operational Linescan System (DMSP-OLS) is responsible for collecting the night-time lights data. Their satellites are equipped with a Visible and Near infrared system with photomultipliers attached which orbits the earth recording and collecting the artificial light data at night. The raw data is then cleaned by an algorithm to remove noise such as gas flares, fires and unstable lights within a threshold time.
Extraction of data:
Here comes the tedious part, my part. DMSP-OLS provides the data in several different formats, however I chose Average Visible, Stable Lights & Cloud Free Coverages (1992- 2013) as it was the most suitable type for my analysis. To extract the data into a useable format, I used a Geographical Information System (GIS) which I must say was quite user unfriendly. Googling was not of much assistance due to the nature of what I was trying to achieve. Eventually, I was able to teach myself how to extract the data by creating several custom shapefiles and zonal statistical manipulations.
Measuring Economic Activity: CARIBBEAN
In general, several studies have shed positive light on using light intensity as a measure for economic activity, especially in developing countries. Radiance Calibrated Lights were shown to be better for developed countries due to light saturation, while stable lights were better for developing countries. It should be noted that the range of light intensity goes from 0 to 63 with 0 being no light intensity and 63 being the highest level of light intensity.
What does light intensity look like in the Caribbean?
The goodness of fit map between light intensity and economic growth below, clearly shows that the stable lights data is a better measure for developing countries vs developed. [Scale: Yellow (low) to Deep Blue (high)].
A quick look at the correlation between light intensity and real gdp shows really significant correlations with over half of the countries having a correlation of over 0.80; only Jamaica revealed an extremely low positive correlation. With regards to the relationship with real gdp per capita, correlations were a bit lower than compared to real gdp but still more than half the countries were observed to have a correlation of over 0.70. Surprisingly, Jamaica was shown to have a moderately negative correlation which may be due to the structure of Jamaica’s economy or it may just simply be the data distribution.
One of the interesting things about light intensity is that from 0 to 62, light intensity progresses in a linear fashion. For example, if one country has a light intensity of 10 and another has 20 then the latter country is said to have twice as much light intensity. Given its linearity and the general data consistency, light intensity can be useful for comparing development across countries.
Over the 12 year period, all countries were shown to have developed to some degree with Barbados maintaining the lead in terms of light intensity, moving from a value of 35.4 in 1992 to 44 in 2013, followed by Trinidad & Tobago and Antigua & Barbuda with a value of 37.2 and 37 respectively. Surprisingly, in 2013 many countries still have relatively low light intensity which may be due to large areas of land mass with little civilization.
Grid-Level Tracking in Barbados
Similarly to country level comparisons of development, light intensity can be used to examine subnational levels of development.
Over the 12 year period as expected the capital’s parish, Saint Michael, maintained the highest level of light intensity with a remarkable value of 61 followed by Barbados’ two major tourist hot spots Christ Church (47.5) and Saint James (47.2). Notably, Saint Peter has seen a upward shift in its position by 2.
These primary results show great promise for using light intensity data as a measure of economic activity for most countries examined. However, night-light intensity data does have some limitations that should be noted.
- The photomultiplier on the satellites have a gain filter on them which sets the limit to 63, leading to saturation and information loss in cases where light intensity is higher. Luckily in the case of the Caribbean, it is not as prevalent when compared to advanced countries. Various radiance recalibration methods have been established to combat this issue if it ever does arise.
- The satellites are unable to detect interior lights such as office lights within a building at night which may be useful in very technologically advanced economies.
- There is a lack of inter-satellite calibration and as such, results in some information loss when attempting to merge the individual satellite data.
Nevertheless, its a great indicator to add to your arsenal especially for micro/grid level analyses.