October '07 Newsletter

by Gordy Slack

Compared to developing a human/machine neural interface, say, getting a human being to Mars, or mitigating a global climate crisis, some of the twenty-first-century's most pressing engineering problems seem simple. How, for instance, to provide families living in Indian slums—or villagers in rural communities—with safe, clean sources of drinking water. Two UC projects are attempting to do just that.

Although the technologies employed must be straightforward, the problems themselves are anything but simple, says Ashok Gadgil, Senior Staff Scientist in the Environmental Energy Technologies Division of Lawrence Berkeley National Laboratory. Not only must the systems be safe, absolutely reliable, and very inexpensive, if they are not culturally appropriate, no one will employ them, Gadgil says. "They can end up doing more harm than good."

Kara Nelson, Assistant Professor of Environmental Engineering at UC Berkeley concurs. Getting clean water to families in the slums of Mumbai is "partly an engineering problem, but it is also an education problem."

"Most of the slum's very poor residents lack formal education and do not understand germ theory of disease transmission," says Erin Inglish. "They do not necessarily know how important it is to wash their hands carefully with soap after using the toilet and before putting their hands into the family's water source."

Inglish is a senior engineering student and the co-director of the Haath Mein Sehat (Health in Your Hands) Project, a student-based effort to aid Behrampada, a slum community of 175,000 people in Mumbai, on India's west coast. Since 2004, Berkeley students have been working with a group of local women from Behrampada in a two-pronged effort to reduce the transmission of water-borne diseases.

The first prong, the engineering component, is to design a reliable on-site filtration system that could be purchased by residents for ten dollars or less. The second aspect of the project is creating a sustainable educational system, run by local university students, that will teach the people of Behrampada hygiene and sanitation practices that will help keep their water clean.

Water is delivered into the slum by the Mumbai municipal water agency. There are spigots throughout the neighborhood where families draw water in large containers and carry it home. "Ironically, it is some of the cleanest water in Asia," says Inglish. "But old pipes carry it through storm drains and open sewers in the slum's narrow alleyways. By the time it gets to the spigots where the residents draw it, it is contaminated by E. coli, coliforms, and other pathogens," says Inglish.

Because there is only water pressure in the pipes for a few hours a day in these neighborhoods, any breaks or small holes in the pipe will let sewage and other contaminants into the system. "We know the pipes leak because when the pressure is on, we can see water spurting out of cracks and small holes. When the water pressure is low enough, it can create a vacuum in the pipes that will draw sewage right in," says Inglish.

Contamination also occurs when the water is put into dirty containers. It gets further contaminated by residents once it is in their homes, Inglish says.

Behrampada is densely developed (175,000 people live in one square kilometer) with makeshift five-story dwellings placed so close together that no vehicles can pass and no direct light can penetrate. Fixing the plumbing itself in these conditions is far beyond the reach of the of the student group, says Nelson. "For both political and engineering reasons, the plumbing situation in Mumbai is not going to change anytime soon. So the students decided early on to focus on making a point-of-use filtration system."

Elsewhere in India, in remote rural areas, the problems of clean water are just as pressing, but of a slightly different nature. It is here that the Ashuk Gadgil's engineering expertise is making a difference.

The water sources themselves are usually untreated and host various contaminants. So he has devised an inexpensive, UV-based water processing plant that can be purchased and operated affordably by villages. The garage-sized system includes a number of innovations that make it longer lasting, safer, and less expensive to buy and operate than other available UV systems, says Gadgil. The UV unit itself, which neutralizes pathogens, is suspended above the water, for instance, so that the quartz sleeve that houses it does not get fouled with algae or potassium and calcium carbonate.

Gadgil's system is also engineered to a 300-percent margin of safety. "It is failsafe," he says. "The system shuts down at the intake if the power goes off or there is a voltage sag or the lamp is not replaced properly. All the water already in the plant remains safe to drink," he says.

Providing clean water through a village-based micro-utility not only helps villagers steer clear of the pathogens in contaminated water; it also helps alleviate the health and environmental costs of boiling water in homes to purify it.

"There is already a fuel shortage in many rural areas and women already spend a lot of time collecting fire wood," Gadgil says. "Boiling uses three times as much fuel as cooking, and smoke from fires also causes lung cancer and cardiovascular diseases. So it is much better to purify water in other ways."

A private, Irvine, CA-based firm, Waterhealth International, has licensed Gadgil's design from UC, and in partnership with one of India's biggest banks, helps village councils set up affordable financing to purchase the micro-utilities. Waterhealth set up two of the garage-sized treatment centers in India in 2005; it set up fifty of them in 2006; and by the end of this year it will have totaled 200. Waterhealth is also starting programs using Gadgil's design in the Philippines and Ghana as well.

The Haath Mein Sehat project's treatment unit is still in prototype phase. The latest model is a two-compartment plastic unit. Water is poured into the upper section along with chlorine or some other virus- and bacteria-killing purifier. After chemical decontamination, the water passes through a filter to remove parasites as it enters the lower section where the clean water is stored until use. The lower unit is fitted with a spigot.

The student engineers are experimenting with different kinds of filters and materials for decontamination. Right now they are experimenting with carbon block filters with a very small pore size and halogenated resin beads that could go right into the filter units to chemically treat the water.

"The resin beads would not require the users to add anything extra to the water, a step that is often overlooked for financial or other reasons," says Mike Fisher, an environmental engineering graduate student and the project's technical coordinator.

All experimentation is undertaken with a sharp eye on cost. "Effective commercially available systems cost about 30 dollars," says Fisher. "That is about three times too much for most of the families we are working with to afford. We are trying to get the cost down below ten dollars without compromising effectiveness."

Gadgil, Inglish, Fisher, and Nelson agree that such real-world engineering projects can be frustrating.

"You have to look far beyond the engineering boundary conditions of a problem," says Gadgil. "You need a much deeper understanding of the social and human complexity of the whole situation to make it work."

Difficult and evasive as solutions may be, the engineers agree that the work is extremely gratifying, paying off in the ultimate reward: life.

"It is very satisfying to know," says Gadgil, "that more than half-a-million people are already getting safe, affordable drinking water through Waterhealth's UV systems every day."

by Gordy Slack

Roger Bales is Professor of Engineering at UC Merced, where he is also acting director of the Sierra Nevada Research Institute. Bales’s research focuses on tracking the snow provided to the state of California free of charge by the three-hundred-mile-long Sierra Nevada mountain range. The availability of new, low-energy, wireless monitoring devices, and the maturing of satellite imaging technology plus advances in cyber-infrastructure, are enabling large strides in our ability to track and predict the storage and flow of water.

CITRIS writer Gordy Slack interviewed Bales on September 5, 2007, before Bales gave a CITRIS Research Exchange Talk at the Hearst Memorial Mining Building on the UC Berkeley campus.

CITRIS: Why is it important to know how much water is stored in Sierra Nevada snow?

Roger Bales: The Sierra Nevada, of course, is the source of much of California’s water supply. Being a mountain region and having snow packs that are near zero degrees Celsius, it is susceptible to small temperature increases that will change water runoff patterns and the ecology of the Sierra Nevada. In addition, the urban and exurban development and population growth in the Sierra Nevada and San Joaquin Valley will make major changes in the natural systems and put stress on water and other human systems.

More specifically, though, every spring, the first question water managers ask is, “How much water is out there in the snow pack?” A great deal depends on the accuracy of the answer. If more water is predicted, a Valley farm may plan to plant two crops. With less water, farmers will only plant one. Across the entire valley, we are talking about hundreds of millions of dollars at stake. Even if the uncertainty in our predictions is only ten percent, that can still add up to a huge amount of revenue or loss.

Then, there is flood control. The amount of water stored in the snow pack can overwhelm the reservoir system if it is not timed and released properly. The release is all based on predictions of the amount of snow up there and when it is likely to melt. We are pretty good at predicting the amount of water stored in the snow in normal years, but not very good at predicting extremely wet or extremely dry years. It is those extreme years that pose the greatest challenge and the greatest danger.

Also, there is forest and watershed management. The conditions that determine the likelihood of forest fires, for example, also depend on how much snow has fallen and how much of it is likely to melt and when.

So that question, How much water is stored in the mountains this year, cannot yet be answered as accurately as it should be.

CITRIS: How does climate change play into this equation?

RB: About two-thirds of the Sierra Nevada’s total precipitation is snow. That is even higher in the southern part of the range. Much of that now falls when temperatures are right around zero degrees Celsius. A three-degree increase in temperature would turn a third of that snow into rain. A three-degree average increase would also shorten the winter by about two months and turn some of our biggest snowstorms into rainstorms.

On average, fourteen million acre-feet of water are stored in the snow pack each winter. That is larger than the reservoir storage capacity of the dams on the Sacramento or the San Joaquin. With warming, we are facing a potential loss of all of that storage, and there is no obvious way to make that up.

My work does not measure climate change. But I can say that as climate becomes less predictable, the reliability of the kinds of estimates we can make now will go down. When climate shifts, you find yourself outside the range that the statistical relationships that our current prediction methods are built on. And they are not so good at extrapolations on extremes. We would need to build up another 30 years of history before our statistical relationships would give us comparable results. Even the results they give now with the stationary climate are not as good as water managers would like them to be.

As the pressures of water use increase with population growth, the stakes go up. For all these reasons, it is key that we understand the systems involved and can make accurate predictions about them.

CITRIS: How do you monitor the snow levels now?

RB: We do two main things. First, we use daily satellite images to tell us what areas are covered with snow. Second, we take ground-based measurements that record the snow depth and snow density at various sites around the Sierra Nevada. One of the problems is that there are not enough of those sites and they are not all that representative of the terrain as a whole.

CITRIS: How does the satellite imaging work?

RB: Sunlight hits the earth and reflects differently depending on what it hits. The most difficult problem for detecting snow is separating it from clouds. In the visible part of the spectrum, they look about the same. Fortunately, in the infrared it is possible to tell the difference. So we have good data every day about how much of the surface of the Sierra Nevada is covered with snow.

CITRIS: But from a satellite, how does a lot of snow look different from a little?

RB: It doesn’t. That is a problem. All you can tell is the extent of snow cover. In order to estimate how much snow is out there, you still need good ground-based measurements and that is where the need for new measurement sites, and the new sensor technology, comes in.

Right now, at the end of a season I can always go back and calculate how much snow fell across the basin. I know how much energy it takes for an area to become snow-free, I can convert energy into frozen water, and from that I know how much snow had to be there in order for it to take the number of days it did for it to melt out. But by that method, I have to wait until the snow has all melted to figure out how much was there. That may be fine for me as a researcher, but it does not help the water manager who wants to know how much snow there is right now. Forecasters want to look ahead, not back.

CITRIS: What about the ground measurements? How are they done?

RB: There are now several index sites where snow pack is measured. They tend to be on flat ground, however, and near highways—not necessarily very representative sites. Those are statistically correlated with total seasonal runoff in streams. They are not designed to give quantitative estimates of actual snow pack. But as long as you have a good historical record and no changes in climate, your statistical relationships may be pretty good, at least toward the mean. As I said, they are not very good at the extremes, so significant climate shifts would throw off all bets.

CITRIS: But you’re employing new technology and shifting to a new model, right?

RB: Yes. We are starting to move toward more of a mass-balance model, rather than statistical-base approaches. A mass-balance model means that you know how much snow is out there and you estimate then how much will melt, how much goes into the atmosphere, into the soil, and into the streams and runs off. So rather than a statistical relationship, you are tracking the mass of water from the snowpack all the way to the stream and down to the reservoirs.

CITRIS: So you are getting an absolute number of acre-feet?

RB: Yes, you are actually tracking how much water is there and where it all is in the system. The measurement network for snow pack that is out there in the mountains now was designed many decades ago. We need to get a greater number of representative measurements than we get today. And more different kinds of measurements, too.

Today, for soil moisture and other components like evapotranspiration there are basically no measurements. It is not just the snow pack. You need the soil moisture and other measurements in order to make accurate forecasts. We have some stream-flow measurements on some of the large tributaries and rivers, but not on all of them. Not enough of them. There are 20 some main rivers draining the Sierra Nevada, each with 1-4 active USGS gages. The utilities have additional gages. Putting a gage on each tributary would mean at least doubling the number.

The next generation of hydrologic models will start with the amount of snow and the amount of rainfall and then carry that amount all the way through the watershed hydrologic cycle and estimate how much goes for evapotranspiration and how much goes into groundwater recharge how much goes into stream flow runoff. So they actually track the mass balance of the water through the mountain catchments into the reservoirs. But to do that we need starting estimating how much snow is actually out there. To do that, we need to have many more monitoring sites that gather and transmit more kinds of information.

CITRIS: What new technologies make the new models possible?

RB: Three things need to come together. One is advances in cyber-infrastructure including data management and computing to officially run more computationally intensive models. The second is that the satellite technology has matured so we can get whole watershed information, spatially over the watershed, or over the mountain range. The third is that, with the advent of low-cost sensors and telemetry, we can instrument a greater number of areas more affordably than we could in the past.

CITRIS: Has the price of measurement sites been a problem in the past?

RB: Even though a good prediction is very valuable, there is no good institutional mechanism to recover those costs from the users. It is hard to charge people for a weather forecast. Well, you can charge a few people who want it specifically for their farm, but it takes a long time to build up that infrastructure. Weather forecasts, and climate forecasts, are a public good, so they need to be supported by public investment.

CITRIS: Given how important the issues are, I would expect the state to want to invest in good prediction infrastructure.

RB: There are champions at the state level that are trying to make that happen, but, for now, a reduction in the cost of installing and checking monitors will make a big difference.

CITRIS: What will the new monitoring technology and models allow you to do that you could not have done a decade ago?

RB: It will allow accurate estimates of how much snow is out there in the mountains on any given week. That enables a whole a raft of improved forecasts.

CITRIS: What kinds of things still pose obstacles to accurate monitoring, other than the number and types of sensors?

RB: In terms of the ground-based system there are still some engineering challenges with the cyber-infrastructure. To get the data from the sensor into a usable form for a decision support system is still tricky. It includes telemetry and data processing and archiving and retrieval. In collaboration with the Department of Water Resources and various water agencies, we will soon be developing that system. We are still refining techniques to make remote sensing data more accurate despite the fact that there are trees and clouds that get in the way and complicate the analysis.

But the main problem now is that the ground-based snow measurements are not representative of the terrain and so fail to demonstrate basin-wide snow depth or water equivalent. And there just are not enough of them.

CITRIS: Where are the frontiers in this field now? As you try to increase accuracy in decades to come, where is the greatest hope for progress?

RB: We still need better process understanding so that we can take the measurements and general insights we’ve gleaned in one basin and transfer them to another basin that is less intensively gauged and monitored. There is always going to be somewhat limited data. We cannot do a research project in every catchment. But the better we understand how the systems work, the better estimates we can make even in areas that aren’t heavily monitored.

CITRIS: What new technologies are you employing to set up this new monitoring network.

RB: We are deploying a new generation of instruments. We have put in five new measurement sites along the west side of the Sierra. In addition to snow depth and density, we have also dug pits down a meter to measure soil moisture and temperature. We are measuring stream flow and sap flow in trees. We have put in meteorological stations.

The idea is to get all the main components at representative sites throughout the watershed and to use Lidar and remote sensing data to scale it out.

The satellite remote-sensing technology is really maturing. It is reliable, and we have access to a stream of satellites that provide continuous high-resolution products.

Then there is the advent of low-cost sensors that you can easily deploy lots of places. They use a lot less power than old sensors, so we do not have to have huge solar panels to power each one. And we are still implementing wireless technology to get those communicating with each other and with us. And then the third thing is the improved cyber-infrastructure or data processing and management systems that can actually translate those measurements into valuable information.

We have tested a few different ways of connecting the sensors, anywhere from wires in conduit to state-of-the-art high-tech wireless nodes that communicate with each other and self organize. But none of them are completely satisfactory yet. We were promised a couple more prototypes from colleagues at a couple of other UC campuses that we will test, hopefully this year.

We use NOAA’s GOES (Geostationary Operational Environmental Satellite) satellites for communications in remote areas where you can’t get out a radio signal or a cell phone signal. But GOES has a fairly low data rate. If possible, we try to use radio signals to get data out instead. It is a lot faster.

CITRIS: What is the take-home message from all of this?

RE: We need better methods of estimating water fluxes, especially in the Sierra Nevada, where climate change and other pressures have created a big demand for new knowledge. The better ground-based systems we are employing combined with remote sensing devices that allow us to get and coordinate data from less accessible sites are providing the way forward.