It was fascinating to watch Shamika Ravi, Prime Minister’s Economic Advisory Council member, correlate satellite imagery of construction in the Indian subcontinent to increased economic activity. This happened during the India Today Business Conclave 2024, held in New Delhi.
Remote sensing, often called "eye in the sky", collects data on how visible and near-infrared bands of radiation interact with objects on Earth like water bodies, forests, meadows, buildings, etc. It can also record man-made light emissions and precipitation. It can collect data from large areas, especially inaccessible ones. This has led to the widespread use of remote sensing in a broad range of economically relevant activities ranging from urban planning to tracking the effects of climate change on crop production.
The multi-spectral satellite cameras can record radiation in microwave, infrared, ultraviolet, or visible regions. Sensors may record information from multispectral or hyperspectral bands.
The Indian Space Research Organization manages India's remote sensing programme. The programme began in 1988 with the IRS-1A, the first of a series of state-of-the-art satellites developed in India and launched in 1988.
Advantages of Remote Sensing Data
Remote sensing technologies collect recurrent low-cost data on a large scale. The data includes lights from dwelling and public places, rain, flooding, topography, forest cover, crops, urban development, building type, roads, and fish abundance.
Remote sensing data is available at a considerably finer spatial resolution than data from other sources. Much of the open satellite imagery used by economists has a 30x30 meter grid size. Decisions of economic relevance like zoning, building types, or crop choice have similar spatial scales. Since 1999, imagery at resolutions below 0.5 meters has been commercially available. Remote sensing data is unique in its wide geographic coverage. The data is captured from a single location repeatedly (every day or week) for long periods.
Light Emission as a measure of Economic Activity
Anthropogenic night-light data reflect economic activity at spatial and temporal scales. Similar data from conventional sources are often nonexistent.
A pixel in a satellite image maps a terrestrial area of less than a square kilometre. One can associate with each pixel a parameter that measures the luminosity at night. The data integrated over every pixel of one country is a measure of the activities of that country at night. The indicator reflects the economy and its variations when compared between countries and over time. The data aggregated to the city, district or state level makes it ideal for spatial analyses of economic activity. Satellite data being available at a much higher temporal rate adds to its usefulness. They allow us to determine development surges, the economic impact of civil strife, etc.
Nightlights broadly capture both economic surges over space and their spatial contraction. Growing economies light up more areas for longer periods, and more pixels capture that light. In poorer regions or those mired in conflict or economic depression, more patches of land become dark, and more pixels lose light.
Using spatial light intensity as a substitute for economic activity allows us to map and assess the actual extent of GDP losses as a consequence of conflicts or economic isolation. The dramatic difference in night light emission from the North and South Korean night lights is instructive.
Gross Domestic Product (GDP) is a comprehensive parameter in representing economic growth. Consistent, region-wise GDP data are not available from many countries. Thus, the interesting variation in economic growth taking place within countries is thus missed out. Conversely, data at the transnational level is needed to verify theories about geographical factors that affect growth from regions made up of parts of many countries.
We can use emitted light to reassess the GDP of a country undergoing war or conflict based on how it compares with data from other countries at different stages of economic progress. Since the ebb and flow of the informal economy controls this post-conflict environment, reliable data from conventional sources may not be available.
Buildings and Settlements
Using remote sensing data to identify individual buildings and to categorise them is an emerging research area (1). This demarcates urban and non-urban areas distinctly. In a path-breaking study of what causes urban sprawl, Burchfield et al. (1) use two datasets that classify US land cover in 1992 and 1976 to track both urban and nonurban land cover categories over time. The extracted database with 30-meter resolution allows them to record the remarkable level to which city development in the US in that period has taken place in grid cells proximate to previously developed cells. Urban growth seems to have happened over contiguous cells without skipping across undeveloped cells.
Applications to Agriculture
How remote sensing can assess the impact of climate change on agricultural yields can be seen in the work by Costinot, Donaldson, and Smith (2). They draw on an agro-economic model that is partially based on remote sensing data. When evaluated under both present and future (2070–99) climates, the results predict a change in productivity for any crop in any part of the globe. The fast development in the growth of algorithms for crop classification and yield measurement makes future applications of satellite data very likely to contribute a wealth of information related to agriculture.
The location-wise volume of crop output is an important economic parameter enabling us to determine a broad measure of agricultural productivity. Improvements in measuring remotely sensed yield data suggest that future applications in this area could be valuable.
Indices derived from satellite imagery, such as the Normalized Difference Vegetative Index (NDVI), provide a measure of vegetation health and the Normalized Difference Water Index (NDWI), a proxy for plant water stress, can be integrated in crop yield models to aid in prediction and forecasting. Previously, remotely sensed data was often supplemented with ground truth data on crop information. In recent years, higher-resolution imagery, such as that collected from Unmanned Aerial vehicles (UAVs) equipped with sensors, has been applied to precision agriculture. To work with big data from remote sensing, much recent work has focussed on utilising deep learning to analyse large datasets and learn the relationships between various variables.
Satellite data have been used to quantify large-scale deforestation. Foster and Rosenzweig (3) used time-extensive data from Landsat satellites to explain the increase in forest cover in India starting from the late 1950s. This study combines satellite and village-level survey data to show a correlation between forest growth and growth in the demand for forest products (such as firewood) due to local prosperity. Satellite data with a 30-meter resolution is also used to monitor illegal lumbering activity. Burgess, Costa, and Olken (4) demonstrate how Brazil’s 2006 anti-deforestation policy affected land use choices there.
Further Reading
1. Burchfield, Marcy, Henry G. Overman, Diego Puga, and Matthew A. Turner. 2006. “Causes of Sprawl: A Portrait from Space.” Quarterly Journal of Economics 121(2): 587
2. Costinot, Arnaud, Dave Donaldson, and Cory Smith. 2016. “Evolving Comparative Advantage and the Impact of Climate Change in Agricultural Markets: Evidence from 1.7 Million Fields Around the World.” Journal of Political Economy 124(1):205–248.
3. Foster, Andrew D., and Mark R. Rosenzweig. 2003. “Economic Growth and the Rise of Forests.” Quarterly Journal of Economics 118(2): 601–637.
4. The Power of the State: National Borders and the Deforestation of the Amazon, Robin Burgess, Francisco J. M. Costa, Benjamin A. Olken, March 28, 2016, http://conference.nber.org/confer/2016/EEEs16/Burgess_Costa_Olken.pdf
Comments