Sovereign economic activity monitoring from night-light radiance
VIIRS Day/Night Band radiance time series offer a politically independent, near-real-time proxy for economic output, electrification and urban activity, filling the gap where official statistics are delayed, suppressed or simply wrong.
Sensors
- VIIRS Day/Night Band (DNB): Suomi NPP and NOAA-20 each provide daily global coverage at approximately 750 m ground sample distance, with a panchromatic band sensitive from 0.5 to 0.9 µm. The DNB's on-board gain stages give a dynamic range of roughly seven orders of magnitude, allowing it to detect radiances from a single fishing vessel up to a fully lit city without saturation in most cases. Monthly composites from NOAA/NGDC are publicly archived from 2012.
- DMSP-OLS (legacy archive): Operational from 1992 to 2013 across multiple satellite generations, OLS provides the only continuous night-light record long enough to study multi-decade economic trajectories. Its 2.7 km resolution and 6-bit radiometric depth cause saturation in dense urban cores and prevent inter-satellite calibration without cross-sensor regression. Essential for long time-series work; must be treated as a separate instrument family from VIIRS.
- Luojia-1: China's dedicated night-light satellite, launched 2018, offers 130 m resolution nighttime imagery, roughly six times finer than VIIRS DNB. Revisit is approximately 15 days. It can resolve individual road segments and industrial facility boundaries that VIIRS conflates into a single bright blob, making it useful for district-level attribution within cities. Archive depth and continuity are limited compared to VIIRS.
- VIIRS Nightfire (VNF product): A sub-product of the VIIRS suite processed by the Colorado School of Mines Earth Observation Group, Nightfire isolates high-temperature combustion sources (gas flares, biomass burning, industrial furnaces) by fitting a Planck curve to the M-band shortwave infrared channels. Separating flare radiance from settlement radiance is essential before using night-light data for economic inference in hydrocarbon-producing regions.
Why radiance is a better witness than a statistical office
GDP figures arrive quarterly, are revised repeatedly, and in a non-trivial number of countries are compiled by the same ministry whose budget depends on the result. Night-light radiance is collected every night by a sensor that has no opinion about the answer. The correlation between VIIRS DNB radiance and independently estimated GDP is well established in the academic literature, with studies across Sub-Saharan Africa, South Asia and conflict-affected states consistently finding that radiance explains 70 to 90 percent of subnational income variation where ground-truth data are available for calibration.
For sovereign-risk analysts, the value is not in replacing GDP but in detecting divergence. When official figures report growth and radiance is flat or falling, that gap is itself the signal. The same logic applies to electrification claims, urban population estimates and post-conflict recovery narratives. Radiance does not care what the press release says.
What the sensor actually measures, and what it does not
VIIRS DNB measures top-of-atmosphere upwelling radiance in a broad visible-to-near-infrared band. It does not distinguish the source of that radiance. A fishing fleet, a gas flare, a burning field and a city district can all produce similar digital numbers. Treating raw composites as economic proxies without source separation is the most common analytical error in this field.
Gas-flare contamination is the most quantitatively significant confound in hydrocarbon-producing regions. Nigeria's Niger Delta, the Permian Basin, and the Zagros fields all produce radiance signals that can swamp nearby settlement light if not masked using a product such as VIIRS Nightfire. Agricultural burning creates transient bright events that corrupt monthly composites if cloud-screening and fire-flag masking are not applied. Moonlight and seasonal snow cover alter background radiance in ways that require careful temporal normalisation. None of these problems are insurmountable, but ignoring them produces nonsense.
Saturation in cities, darkness in the data
VIIRS DNB largely solved the saturation problem that made DMSP-OLS useless for measuring economic intensity in dense urban cores. Tokyo, London and New York no longer appear as featureless white blobs. However, saturation can still occur in the brightest industrial zones, and at 750 m resolution the sensor cannot distinguish a prosperous district from an adjacent deprived one separated by a few hundred metres. Luojia-1's 130 m imagery is the current best option for intra-urban economic differentiation from night light alone, though its revisit and archive depth remain limited.
The opposite problem matters equally. Rural electrification programmes in low-income countries often involve solar lanterns, small battery systems and low-wattage LED installations that emit radiance below VIIRS DNB's practical detection threshold under standard monthly compositing. A village that gains electricity access may remain invisible to the sensor for months. Analysts should treat absence of signal as ambiguous, not as confirmed darkness.
Stitching DMSP to VIIRS without breaking the time series
The most requested analytical product in sovereign economic monitoring is a consistent radiance time series from 1992 to the present. The problem is that DMSP-OLS and VIIRS DNB are physically different instruments with different spectral responses, spatial resolutions, radiometric depths and calibration philosophies. DMSP has no on-board calibration source; its inter-annual comparability depends on empirical cross-satellite regression using invariant targets such as desert surfaces and stable urban areas.
Several published intercalibration methods exist, including the polynomial regression approach applied to pseudo-invariant calibration sites and the more recent machine-learning harmonisation frameworks. All of them introduce uncertainty that widens as you move further from the calibration period. A credible long time-series product should carry explicit uncertainty bounds, and any trend detected at less than roughly 10 percent radiance change per year should be treated with scepticism across the DMSP-VIIRS join. This is an honest limit of the method, not a reason to abandon it.
Conflict, displacement and infrastructure collapse as radiance events
Some of the clearest demonstrations of night-light analysis as an intelligence tool come from conflict zones. Syria's radiance declined by approximately 83 percent between 2011 and 2013 according to published analysis of DMSP-OLS data, a figure that preceded and then corroborated population displacement estimates from other sources. Ukraine's grid damage from 2022 onwards has been tracked in near-real-time using VIIRS monthly and nightly composites, with radiance drops in specific oblasts correlating with reported infrastructure strikes.
For insurance underwriters and sovereign-bond investors, these events matter in two ways. First, they provide an independent check on loss estimates submitted by counterparties in affected territories. Second, they give an early-warning signal when radiance begins declining in a region before official acknowledgement of deteriorating conditions. A sovereign that is quietly losing economic output will show it in its lights before it shows it in its filings.
From radiance to a risk-assessment product
Radiance time series become useful for financial clients when they are converted into interpretable economic indicators with explicit confidence intervals, contextual overlays (administrative boundaries, infrastructure networks, conflict event databases) and regular update cadence. A monthly radiance anomaly report for a specific country or province, benchmarked against a pre-event baseline and flagged when deviation exceeds a defined threshold, is more actionable than a raw raster stack.
Satellize runs this class of analysis on open VIIRS and DMSP archives, applying flare masking, fire flagging and seasonal decomposition before deriving radiance indices at the administrative-unit level. The methodology is the same family of approaches used in published academic and World Bank research on night-light economics. For clients assessing sovereign credit risk or pricing political-risk insurance on infrastructure assets, the Overhead column publishes periodic worked examples of radiance-based country assessments. The next concrete step is a scoped methodology review for your specific country or portfolio.
Typical figures
| Spatial resolution (VIIRS DNB) | ~750 m ground sample distance at nadir |
| Spatial resolution (Luojia-1) | ~130 m (nighttime panchromatic) |
| Spatial resolution (DMSP-OLS legacy) | ~2.7 km (fine mode); 6-bit radiometric depth, saturates in urban cores |
| Revisit (VIIRS DNB) | Daily global coverage; monthly cloud-free composites typically available within 2 weeks of month end |
| Spectral band (VIIRS DNB) | 0.5–0.9 µm panchromatic; dynamic range ~7 orders of magnitude |
| Archive depth | DMSP-OLS: 1992–2013; VIIRS DNB: 2012–present; Luojia-1: 2018–present (limited) |
| Minimum detectable radiance (VIIRS DNB) | ~2 × 10⁻¹¹ W cm⁻² sr⁻¹ under favourable conditions; small isolated light sources detectable but sub-village installations often below threshold |
| Latency (monthly composites) | NOAA monthly VIIRS composites typically published 2–4 weeks after month end |
| Coverage | Global, including polar regions; cloud screening reduces effective temporal resolution in persistently cloudy tropics |
| Delivery formats | GeoTIFF radiance rasters, CSV radiance-index time series by administrative unit, anomaly alert feeds |
Analytics Satellize can run
| Monthly radiance anomaly index by administrative unit | Temporal decomposition of VIIRS DNB monthly composites against rolling baseline; flare and fire masking applied using VIIRS Nightfire and MODIS active-fire flags | CSV time-series table and choropleth GIS layer, updated monthly |
| GDP proxy nowcast for data-sparse sovereigns | Regression of radiance index against available GDP observations to calibrate a radiance-to-output elasticity; applied forward to generate quarterly economic activity estimates | Quarterly country report with confidence intervals and stated calibration assumptions |
| Conflict-impact radiance assessment | Pre/post event radiance comparison at oblast or district level, cross-referenced with conflict event databases; percentage radiance change mapped against administrative boundaries | Event report with before/after raster pair and tabular radiance-change statistics |
| Electrification progress monitoring | Lit-area expansion tracking using binary thresholding of VIIRS DNB composites; change in lit-pixel count within defined administrative polygons over time | Annual electrification progress GIS layer with pixel-count time series |
| Gas-flare-separated settlement radiance | Subtraction of VIIRS Nightfire flare radiance estimates from total DNB composite to isolate settlement and industrial light sources | Flare-corrected radiance raster stack for hydrocarbon-producing regions |
| Harmonised DMSP-VIIRS long time series | Intercalibration using pseudo-invariant desert and stable-urban calibration sites; polynomial cross-sensor regression with stated uncertainty bounds at the instrument join | Annual radiance index from 1992 to present with per-year uncertainty estimates, delivered as GeoTIFF archive |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.