Land climate
What processes determine past, present and future climates over land?
- Why is relative humidity high in some places, and low in others? Surprisingly, there is no simple answer to this question, at least for the part of the atmosphere that we live in (near the land surface). We provided a simple theory that explains the large-scale spatial structure of near-surface relative humidity over land. In its simplest form, the theory argues that near-surface relative humidity is primarily determined by soil moisture (2024).
- A common view is that drylands will expand as the planet warms. We showed that this result is an artifact, and that there is no clear evidence that drylands will expand with warming (2021, 2022).
- Are plants more sensitive to dry air or dry soils? The difference matters because climate models project relatively drier air as the planet warms, but are unclear on what will happen to soil moisture. Many recent studies have argued plants are more sensitive to dry air. We used a natural mechanism-denial experiment — unvegetated salt flats in the desert — to show that plants are probably more sensitive to dry soils (2023). The Harvard Gazette wrote an article about this work.
- Soil moisture is a key control on land climate. We used global satellite observations and a simple theory to understand its seasonal cycle (2022).
- Wet soils can cause rain, leading to feedbacks between the land and atmosphere; but this is impossible if the land rapidly ‘forgets’ soil moisture anomalies caused by rain. We mapped land ‘memory’ globally using soil moisture observations (2017a, 2017b, 2019), allowing us to identify regions where land-atmosphere feedbacks can occur.
Storms and rainfall over land
Under what conditions do land surfaces enhance or suppress rain?
- We discovered two new mechanisms by which land surfaces can control rainfall. First, we showed that dry soils cause dry atmospheres, leading to more raindrops evaporating as they fall, and less rain reaching the surface (2021). Second, we showed that “rough” surfaces (such as cities, forests and wind farms) heat up more during the day, generating mesoscale circulations that cause rainfall (2023).
- Painting a landscape white increases its albedo and cools it. Similar “land radiative management” (LRM) schemes have been proposed as a way to locally reduce temperatures as the planet warms. We showed that conducting LRM on a region causes it to rain more in the area surrounding that region. More rain and wetter soils lead to lower temperatures, meaning surrounding regions should be expected to benefit from LRM, too (2023).
- We showed that irrigating Australia’s inland desert would not increase rainfall, contrary to a century of claims otherwise (2023). See this article we wrote for The Conversation for more background on the historical and political debate on this.
Evaporation
How do landscapes dry out?
- Landscapes mainly dry due to evaporation and transpiration (“evapotranspiration”, or ET). Most models of ET require land surface information as inputs, which limits their use. We developed a very simple theory (2019, 2020, 2021) to estimate ET from weather data alone, without any land surface inputs required. The theory’s ET predictions are about as accurate as state-of-the-art measurements (2020) and are more accurate than a sophisticated model product (2021), even though the theory is radically simpler.
- Long-term records of evaporation don’t exist, making it difficult to understand the origins of past droughts. We combined our simple theory with long-term records of standard weather data to create an observational drought record extending back to the 1940s (2023).
- We developed an alternative to the famous Penman-Monteith (PM) evaporation equation (2020). The alternative fixes a major error in the PM equation, and is consistently more accurate. There is no cost to using the alternative: no extra assumptions, parameters, or data inputs are required.
- Evaporation is driven, in part, by the turbulent lower atmosphere. We developed a parsimonious theory — based on Kolmogorov’s scaling of the turbulent energy spectrum, one of the most empirically well-supported results in turbulence — that explains observed features of the mean velocity (2016) and temperature (2018) profiles in the turbulent lower atmosphere (and in other wall-bounded turbulence settings, more generally).
Remote sensing of the biosphere
How can we accurately measure the state of the land surface from space?
- Using radar observations from a satellite designed for something completely different, we estimated soil moisture and a proxy for vegetation cover globally (2014).
- Soil moisture can be measured globally with satellites: all else being equal, a wet soil will emit less microwave radiation compared to a dry soil. By measuring microwave radiation emitted from the land surface, we can work backwards to infer soil moisture. We showed that soil salinity can cause larger-than-expected biases in soil moisture retrievals at satellite scales (2012). Caution is required in highly saline areas.
- How accurate are satellite observations? There is often a massive scale mismatch between satellite observations and ground truth measurements. Naive comparisons will overestimate errors in satellite observations. We extended a statistical technique for handling this problem — known as ‘triple collocation’ — to estimate the correlation coefficient of the satellite observations with respect to the unknown true value (2014). We also derived variants that work for validating satellite precipitation (2015) and landscape freeze/thaw state (2016, 2018) retrievals.
- We performed the first global assessment of satellite retrievals of near-surface air temperature and humidity from the AIRS mission. We found the retrievals were less sensitive in the tropics, consistent with known physics of the tropical atmosphere (2021).