The World Needs (a lot) More Thermal Infrared Data from Space
Our Planet is On Fire
OK… maybe that’s a bit hyperbolic, but it shouldn’t be controversial to say that over the last 10 years things have been getting hotter and drier. There has been a significant increase in the frequency and severity of drought conditions in certain areas of the world. As I write this, much of the Western United States is in a severe drought. A recent article in the New York Times illustrates how extreme and widespread drought conditions are this year as well as the high likelihood that conditions will worsen through the summer. Climate change continues to increase the volatility and severity of drought cycles.
While we’ve been hearing a lot about US drought in the past month, these conditions are not unique to North America. Brazil’s government agencies have recently been sounding the alarm that the country faces it’s worst drought in 91 years. In India, the period from 2016–2018 included the worst drought conditions in 150 years due to deficits in monsoon rainfall. In 2019, 42% of India’s land area was in drought conditions. In Europe, there have been drought years in 2007, 2011, 2012, 2015, and 2018.
These drought events have an array of negative impacts, including increasing risk of wildfires, decreasing agricultural crop yield or quality, and enlarging the need for an increasingly scarce resource: freshwater. There are also significant socioeconomic knock-on effects from these phenomena. Wildfires can obviously destroy houses, businesses, and critical civil infrastructure leaving people without homes, income, or basic utilities like power. Low crop yields or crop failures can lead to volatility in commodities pricing, famine, migration, and conflict. Lack of freshwater exacerbates the above issues and is increasingly the source of both global and regional conflict.
To manage these challenges, we need good tools for monitoring and mitigating the impacts of water stress and drought.
Thermal Infrared Remote Sensing
Of the various remote sensing datasets available from government agencies and commercial providers thermal infrared and particularly long-wave infrared remote sensing may be the most promising in its potential to support mitigation strategies for drought-related hardship and contribute to related UN Sustainable Development Goals like the goal to “End hunger, achieve food security and improved nutrition and promote sustainable agriculture”. Long-wave infrared earth observation allows us to measure energy radiated by the surface of the earth between 8 to 12 micron wavelengths and derive land or sea surface temperature through the application of a few fundamental laws of physics discovered by Plank, Stefan-Boltzmann, Wien, and Kirchhoff. If we can measure the temperature of a vegetated land surface at the same time each day, we will see trends in temperature over time that will tell us whether the vegetation is staying relatively cool because the plant has good access to soil water content, stomata are open and the plants are transpiring at a nominal rate, or whether the vegetation is increasing in temperature because the plant is stressed, and the stomata have closed to preserve water.
Land Surface Temperature (LST) is a powerful rapid response indicator, showing increases in soil and vegetation canopy temperatures as soil moisture deficits and vegetation stress develop, and in some cases prior to measurable reductions in shortwave vegetation indices. Thermal infrared and surface temperature derivatives can detect stress weeks earlier than commonly used vegetation indices, like the Normalized Difference Vegetation Index (NDVI). This is due to the fact that the changes in chlorophyll and cellular structure from reduced photosynthesis that would lead to a reduction in vegetation index values are lagging indicators when compared to the initial stress response of closing stomata or increased warming trends in dry soil. Further, for some crop types it is easier to see temperature anomalies than anomalies in NDVI. Using a combination of visible and NIR vegetation indices with LST can provide both an early and robust indicator of vegetation stress.
Global Food Security and Irrigation Management
Agriculture currently consumes 80–90% of freshwater used by humans, with two thirds of this used for crop irrigation. Even if our access to fresh water remained constant, to simply maintain global food security by increasing crop production as the global population grows, we would need to reduce the the amount of water used per unit of crop yield (AKA “more crop per drop”). The reality is that climate change will lead to an overall rise in evapotranspiration rates, and we will need even more freshwater to support irrigation. Increased frequency of drought years will put additional pressure on our freshwater resources and require us to use the water we have very efficiently. Frequent observations of long-wave infrared derived LST can indicate early water stress and support decisions on where and when to apply irrigation so that crop yield is sustained, and water is not wasted.
There are a variety of approaches that use LST data to make inferences about the level of water stress in vegetation. The simplest approach is to look at the cumulative difference between canopy temperature and air temperature over a period of time (CATD and Stress Degree Days). An unstressed plant canopy will be 2–5 Kelvin or degrees Celsius below the ambient air temperature. When the difference between the air temperature and the plant canopy temperature is 0, water stress is assumed. The Crop Water Stress Index (CWSI) is similar but normalizes the value with the boundaries of the potential leaf temperature under maximum and minimum transpiration rates, implicitly accounting for other influencing factors like wind and vapor pressure deficit. There are other approaches that use NDVI or other vegetation indices as a proxy for the relative contributions of vegetation canopy and soil within a pixel (TVDI, WDI, IG) with the added advantage of being able to be derived solely from satellite earth observation data.
While water stress indices are easy to derive and quite useful on their own, it’s possible to use LST to go a step further in modeling the Surface Energy Balance and Evapotranspiration (ET) as a component. This modeling approach essentially tells you how much evaporative cooling is occurring to get to the temperature observed from remote sensing given a known or estimated level of incident solar radiation. It typically requires a number of inputs, including weather, landcover, and leaf area. There are a few approaches to modeling ET using LST as an input, like ALEXI and STIC. Once ET is derived, it can also be input to an Evaporative Stress Index (ESI) that provides an indicator of when estimated ET is below nominal ET for healthy vegetation and the vegetation is stressed.
If we can measure when and where crops are stressed, then we can also apply precision or variable rate irrigation to ensure that the crops are getting just the right amount of water when it’s needed. Traditionally, the application of irrigation has been based on the land managers intuition or simple approximations of current ET from a single weather station and adjustment coefficients (Kc) for entire fields. With daily high resolution indicators of water stress, like those described above, irrigation managers can make better sub-field decisions about where to apply water. With daily high-resolution modeled ET, irrigation managers can apply an amount of water that optimizes yield without wasting water. There are a variety of published studies and articles (for example) that show significant reductions in water consumption (20 — 40% reduction is not uncommon), water costs, and lost yield. The added benefit of using earth observation data is that you can monitor all agricultural areas without the need for expensive in-field sensors. In other words, deploying more long-wave infrared earth observation systems, data, and analysis techniques will have a substantial positive water use efficiency impact and have the side benefit of increasing agriculture profitability.
Wildfire Risk and Monitoring
All of the dynamics between LST and agriculture water stress are also relevant for forests, scrubland, or any other vegetated area. As an area dries up, the likelihood of ignition and fire propagation increase. Combinations of LST and vegetation indices can be used to estimate live fuel moisture content for fire danger ratings. It’s also possible to forecast potential burned area and duration of wildfires using LST data. Once a fire has started, it is possible to detect and track the fire with long wave thermal infrared data, even through significant smoke that would obscure the fire from visible and NIR systems. When a fire has been controlled, it’s possible to map the burned area and even characterize the contribution to global warming.
But Wait… There’s More!
While the vegetation monitoring applications above are incredibly valuable, thermal infrared remote sensing is not a one-trick-pony. There are a variety of other application areas that can be addressed with surface temperature from long-wave infrared:
- Monitoring volcanic events
- Early prediction of earthquakes
- Improved land use and land cover classification
- Oil spill detection and monitoring
- Mapping malaria, cholera, and other communicable disease potential
- Vessel and vehicle detection
This is just to name a few other land-based applications. The list grows much longer when we include ocean-based applications.
The Need for Data
While there is so much potential in thermal infrared remote sensing, there’s not enough available data to realize that potential. The data from MODIS, Landsat, Sentinel 3, ECOSTRESS, and other government missions are vitally important and have enabled essential research into the applications above but they lack in resolution or revisit. Many articles in the literature express the need for resolution better than 100 m and daily or sub-daily revisit. New systems like Sentinel 8 (LSTM), NASA missions like Landsat Next, and those in support of the Surface Biology and Geology Designated Observable will address resolution to some extent but will still lack high frequency revisit and won’t be available for many years.
In the not so distant past, the situation was similar in the world of visible and NIR remote sensing before companies like RapidEye and Planet stepped in with commercial offerings. New commercial providers aim to do the same now for long-wave thermal infrared by developing constellations of earth observation satellites that will provide daily, high resolution LST. Not all commercial thermal infrared data will be the same, though. To address the applications described above, there is a requirement for high radiometric calibration accuracy and global coverage. Hydrosat, for example, aims to meet all of these requirements.
Get Involved
Not only does the world need more thermal infrared data, it also needs more data scientists and software engineers that know how to use LST to address drought-related issues at scale. If you’re interested in learning more, there are great resources online or excellent textbooks. If you’d like to learn more about what Hydrosat is doing to bring daily, high quality, high resolution LST to the world or collaborate with us, please get in touch.
