Microwaves measure soil moisture from space satellites

Remote sensing is being viewed increasingly as a critical tool for supplying badly needed data to advance the field of hydrology. Only from space is it possible to efficiently and repeatedly monitor the earth's surface for soil moisture and vegetation, and get consistent pictures of the distribution of water. Prof. Eni Njoku ('76 PhD, Electrical Engineering & Computer Science) just finished an appointment at MIT on a Martin Luther King Visiting Faculty Fellowship, where he worked with Profs. Dara Entekhabi and Rafael Bras, "using remote sensing to understand more about the global water cycle, and in particular that part related to the land surface. We want to understand the distribution of soil moisture at the surface, and how it affects weather and climate.

"It's been very helpful for me to spend some time at MIT, working my remote sensing background more into the coursework and research, and also learning more about hydrological modeling," Njoku continues. During spring semester Dara Entekhabi and Njoku co-taught 1.713J Land Atmosphere Interaction, and introduced remote sensing. Some of the students completed term projects using remote sensing data which Njoku brought to MIT from his post at the Jet Propulsion Lab (JPL) in California.

During his fellowship year here, Njoku says, "I could mentor the students and introduce them to methods for working with these data. We want to get them enthused about these new ways of looking at the earth. In the past, most hydrologic data have been gathered by doing in situ sampling in the field. That method gives very precise localized measurements, but it takes a satellite to get the big view."

Every few years NASA solicits new ides for space missions. As a collaboration between MIT, the Goddard Space Flight Center and JPL, Entekhabi, Njoku and colleagues have compiled a proposal for a new satellite mission called Hydros to measure global soil moisture and its frozen/thawed state. Essentially the satellite will have an instrument to measure soil moisture across a wide swath as it circles the earth at 670 km (418 miles) above the surface. Every three days this swath will completely cover the surface of the earth, and the data will be transmitted to the ground in raw bits.

Measurements from the satellite will be a combination of the soil moisture signal modified by the surface vegetation, roughness and topography, altered by the atmosphere, and received by the instrument. Researchers will have to "uncouple all these effects by having a very good physical model of how the atmosphere, surface, and instrument effects influence the microwave radiation being received," says Njoku. "We will put that physical model together with the data and estimate the moisture at the surface, and then provide maps of the surface moisture."

One of the key advances in recent years has been in data assimilation, "which is a method for combining measured data with a hydrological soil model. The soil model expresses how the surface moisture and the deeper layer moisture are related. The satellite will give measurements for the surface layer, and we couple that with a model of the deeper soil layer. The surface measurement will allow the model to better predict the soil moisture down through the root zone, which is where plants draw up moisture and transpire. The flux of water from the soil into the atmosphere is very much controlled not just by the surface evaporation, but by the root uptake and transpiration through vegetation. Satellite measurements of soil moisture need to be coupled with hydrological models and put into a data assimilation framework to determine those fluxes of water that modify the hydrologic cycle. The hydrologic cycle is the big picture that we really want to understand better," says Njoku.

The researchers are looking at changes on scales of tens to thousands of kilometers as the earth goes through its natural seasonal cycles of precipitation and evaporation. Njoku explains, "Examining the long-term record of precipitation, we see annual cycles globally and some perturbations on top of that, which may be natural fluctuations or may be induced by human activities. The challenges are to understand what changes we're imposing on the environment."

Soil moisture distribution supplies another piece of that puzzle. The footprint (what the antenna sees) covers an area of 10 to 40 km (6.2 to 25 mi). At that scale, the data are more relevant to weather forecasting and climate than to basin-scale hydrology, which normally requires data and modeling at or below the 1 km (0.6 mi) scale, says Njoku. Bigger antennas in the future will allow smaller scale measurements.

With this large scale monitoring, researchers could examine widespread changes such as desertification and major flooding. Njoku notes, "It would have helped to have measured precursor conditions to the huge Mississippi River floods of 1993, and understood in advance how saturated the soil was. Knowing whether precipitation will soak in or run off would be very useful in predicting when floods will occur."

Researchers are trying to predict floods with existing satellite instruments which were not actually created for this purpose. The instrument which Njoku and Entekhabi envision will be specifically designed with very long wave lengths (i.e., low microwave frequencies), to better penetrate the vegetation which masks the radiation emanating from the soil. Since the existing instruments in orbit cannot see very well past vegetation down to the soil, they are very poor at estimating moisture even for fairly open areas such as grasslands and savannahs. With its long wavelength, the new instrument will see through a significant amount of vegetation and should be able to detect soil moisture in most places aside from dense forests.

Njoku is on the science team of an instrument called AMSR (Advanced Microwave Scanning Radiometer), which was launched in May 2002 along with five other instruments on a NASA earth sciences satellite called Aqua. As the name indicates, Aqua will focus on the water cycle in its frozen, liquid, and vapor states.

The AMSR instrument includes a low frequency radiometer with a wavelength of about five cm (two in.), although the researchers would prefer a wavelength of about 21 cm to better penetrate the vegetation. Njoku states, "This instrument was not designed for soil moisture, but was optimized for other measurements which use higher frequencies, such as sea surface temperature, precipitation, sea ice and snow. We'll have to see how well we can estimate the soil moisture with it. It's too early to tell just yet since the initial instrument calibration and engineering assessments have not yet been completed."

At JPL, Njoku has been involved in developing the models and software to analyze data from this instrument. As the calibrated data become available during the coming year, they will be turned into maps of soil moisture. In areas with bare soil and minimal vegetation, researchers expect to observe the soil moisture seasonal cycles quite well. "To detect the soil moisture over more of the globe, we'll have to go to the next generation instrument. The usual sequence is to make do with the non-ideal instruments available at the moment, and try to push forward the instrument you would really like."

The long wavelengths necessary for seeing through the vegetation require big antennas, which are much more expensive and difficult to launch into space. Designs of future large antennas to be used on remote sensing satellites include umbrella-shaped objects up to 12 m (37 ft) across. Made of a mesh of tiny gold-plated wires, they fold up compactly for the launching process.

Njoku plans initially to use a smaller, 6-m (19.5 ft) antenna for soil moisture remote sensing purposes. Radiation given off from the earth's surface will be reflected by the antenna into a receiver on the spacecraft. Another smaller communications antenna will transmit the data down to ground. As the spacecraft moves in orbit, the antenna rotates around a vertical axis and maps out a constantly changing swath of surface. From each of the footprints viewed by the antenna, it receives a signal and transmits it back to earth, giving researchers measurements to convert into soil moisture maps. This design is the basis for the Hydros mission, which could be launched by NASA later this decade if its continued development is successful.

Errors can creep into remote sensing measurements in many ways, making a validation program necessary to check for accuracy. Njoku and colleagues are planning an extensive validation program over large areas where they can make physical measurements to compare with the satellite data. Because the satellite views land in 10 to 40 km chunks, each pixel or footprint provides an average of the landscape. Researchers will know where the points are, so they can discard soil moisture data in dense urban sprawls where all the soil is buried under concrete. In the countryside, the data reflects an estimate of the average soil. "In some areas the averages will be meaningless. But the measurements will be extremely useful in a high percentage of the global area, where no other technique can make those measurements."

"We did two validation experiments in Oklahoma, called Southern Great Plains '97 and '99," describes Njoku. "The southern Great Plains have huge areas of pasture and farmland, and the average soil moisture is a relatively true, good measurement. This summer we made measurements on huge fields of corn and soybeans in Iowa. These field experiments provide validation for some of the existing satellites, and are a key part of determining how accurate AMSR will be for soil moisture now that it's launched."

During the validation experiments, students and scientists make field measurements of soil moisture, vegetation biomass, and surface temperature. They gather data on the meteorology to measure the fluxes of energy from the surface to the atmosphere. All these measurements are collected over large areas by the ground teams for several weeks.

The validation group encountered dry conditions followed by a series of thunderstorms while working in Iowa this summer, "so we saw big changes in the soil moisture, as well as significant crop growth." Other researchers flew in aircraft over the sample area with the same type of instruments to be installed on the satellite. "We collected three sets of data, all at different scales. The ground data consist of multiple points. The aircraft measurements can see areas of about a few hundred meters, and the satellite measurements cover tens of kilometers. We will examine how the soil moisture data look at different spatial scales, so we can understand how measurements on an average of 10 to 40 km relate to a large number of aircraft samples at a few hundred meters."

It is often assumed that an average of all the aircraft samples should yield the same result as the much larger satellite footprint. "If the surface is very heterogeneous with a mixture of trees, other vegetation, and bare fields, some theoretical analysis has shown that the moisture average of those different places may differ from what the satellite detects. We need to better understand the heterogeneity effects," says Njoku. "All these questions are being studied in the field validation experiments so we can better interpret future satellite data."

Usually the first time a satellite instrument is launched for a given scientific purpose, "the information it sends back is quite different than what you expected," acknowledges Njoku. "We anticipate that it will take a few years to fully understand how best to use these data, and how to fold them into the type of hydrologic models they will eventually be used for."

The ultimate goal of much of this work is to improve weather forecasting. Right now weather forecasts can look ahead for three to seven days. Long range climate forecasts try to predict the precipitation for the coming year. One disadvantage is that these models don't have much information about what's happening on the surface.

"If the models run for a few days, they start giving rather erroneous predictions," says Njoku. "The predictions would significantly improve if we had more information about the surface moisture. The surface moisture really controls the evaporation, which carries off the latent heat energy that has a major influence on winds and precipitation. Eventually an enhanced ability to monitor soil moisture could help improve weather and climate predictions and have huge economic impacts. It could help improve agricultural forecasting, and help to plan irrigation needs and crop yields."

This satellite instrument will help take a snapshot of moisture levels and indicate where conditions are changing. "If we know how things are changing, we can better understand why they change. What natural cycles are involved? Is a drought caused by something which happened a few months ago, or by conditions in a different part of the world? What are the connections between what's going on and the physical processes? We don't know enough right now to interpret what's happening and make predictions. These measurements will allow us to build better models that help us understand what's going on."

When El Niño episodes occurred decades ago, people noticed that it led to increased precipitation and drought in seemingly unrelated regions of the world. "After a great deal of attention in the last 10 to 20 years, we now understand more about El Niño's far-reaching effects. In the same way, we want to understand what's happening on the land surface. Are there El Niño like signals? Are there times of the year or cycles of the decade when certain areas are much wetter than others? Does that wetness relate to drought somewhere else? With more information from space, we could get a better picture," states Njoku.

International cooperation in space is particularly important given how expensive it is to build and launch instrument-laden satellites. Now that Europe, Japan, India, and countries of the former Soviet Union are becoming very active in space, more governments and national agencies are discussing how to plan new space missions, pool resources and coordinate measurements. The AMSR instrument that Njoku is working with was built by Japan, for example, but launched by NASA on its Aqua spacecraft. The Hydros mission includes components that will be provided through Canadian and Italian partnerships.

With the proliferation of remote sensing, "we have to figure out how best to understand, process, and utilize all the data to bear on our problems. Universities need to train students in the potential of these data. Hopefully I'll take away from my stay here more interaction with faculty and grad students, and a closer link between the university and the government lab," says Njoku. "JPL has always put an emphasis on fostering closer ties with universities and developing joint proposals, and that's what I'm trying to do here."

Undergraduates in England specialize quite early, as opposed to American students who might spend two or three years narrowing down a major and are required to choose electives outside that subject. As a student at Cambridge University, Njoku "took science and engineering as an undergrad. One big exam at the end of the year pretty much determined the course grade, so there was less pressure to work hard during the year."

Arriving at MIT in 1972 to study electrical engineering, Njoku was immediately confronted by the constant pressure of quizzes and graded homework assignments. "I couldn't afford to relax for a few weeks or months because it affected my grade. Since the university courses are more specialized in England, I was initially ahead. However, the pace is very fast at the graduate level here, and soon I was presented with extremely challenging work. I found that first year to be quite an adjustment in the pace of work and the courses I was taking."

As Njoku studied communications satellites, antennas and propagation, "I took a very interesting course from Prof. Jin Kong on electromagnetic theory, where we did a term project in remote sensing. Remote sensing is an application of electromagnetic wave theory, and I got interested in the subject toward the end of my SM thesis work."

In 1974, Njoku heard a seminar by Joe Waters, a recent EE/CS grad who had moved out to the Jet Propulsion Lab in California. "After his seminar I asked if he had any summer jobs in remote sensing. A week later he invited me to join a project. I worked on remote sensing of water vapor and cloud water to see whether we could measure the cloud liquid water droplets and water vapor at different wavelengths and detect the total amount of liquid water and water vapor in the atmosphere, as well as getting information on the distribution of the cloud water droplets. I did some theoretical modeling work and really enjoyed it."

The following summer Njoku returned to JPL for experiments on surface changes as measured from microwaves. The group constructed a large sandbox, irrigated the sand, and studied how the microwaves emitted from the sand depended on the amount of water and the temperature. "The idea was to detect changes in water and temperature in the sand using microwaves. We fit the experimental data to the fairly simple theoretical models of the time. Everyone programmed on punch cards. You'd take your stack of cards to the computer operator to put them in, pick up the output the next day, learn that you had a problem, then try to fix it. It was so slow and cumbersome that looking back, you wonder how we ever did good work that way!"

After finishing his PhD in 1976 under Prof. Kong, Njoku joined the earth remote sensing program at JPL, and has remained there for most of his career.