Notes on Remote Sensing - Weather Observation Analysis | ATMO 251, Study notes of Meteorology

Download Notes on Remote Sensing - Weather Observation Analysis | ATMO 251 and more Study notes Meteorology in PDF only on Docsity! Weather Observation and Analysis John Nielsen-Gammon Course Notes These course notes are copyrighted. If you are presently registered for ATMO 251 at Texas A&M University, permission is hereby granted to download and print these course notes for your personal use. If you are not registered for ATMO 251, you may view these course notes, but you may not download or print them without the permission of the author. Redistribution of these course notes, whether done freely or for profit, is explicitly prohibited without the written permission of the author. Chapter 6. REMOTE SENSING 6.1 Satellite Thermal Sounders The Earth is a planet covered by a blanket of warm gas. Except for when there are clouds, that blanket is nearly transparent to sunlight, which is how the ground heats up during the day. The Earth and its atmosphere are not hot enough to glow themselves, at least not that we can see. They do emit radiation depending on their temperature; it turns out to be mostly in the range of wavelengths we call infrared. One neat thing about the atmosphere is that, in the infrared range, different gases absorb and emit radiation at different wavelengths. (If a gas can absorb at a particular wavelength, it can emit there too.) So a satellite sensor out in space that detects infrared radiation might see a glowing ball of gas, if it’s tuned to a wavelength that a gas in the atmosphere absorbs and emits, or it might see radiation coming all the way from the Earth’s surface, if it’s tuned to a wavelength that no gas in the atmosphere touches. ATMO 251 Chapter 8 page 1 of 23 Think about those two wavelengths for a moment. Suppose you could adjust the satellite to change its detection wavelength gradually from the atmosphere-opaque one to the atmosphere-transparent one. As the wavelength changes, most of the radiation detected by the satellite will be coming from farther down in the atmosphere, until eventually it starts seeing some of the Earth’s surface and, still later, all of it. At this point, we’re still in the “who cares” portion of this exposition, as in “Who cares what the radiation is coming from?” The answer to that question involves two very key facts about electromagnetic radiation. The first fact is that the intensity of emitted radiation depends directly on the temperature of whatever object (or gas) emitted the radiation. I bet you can see why that might be useful, since the temperature of the Earth and its atmosphere is kinda relevant to the ATMO 251 Chapter 8 page 2 of 23 At all three wavelengths, clouds are opaque and thus emit well too. An extensive comma-shaped cloud mass over the eastern United States shows up in all three images. The highest cloud tops appear to be emitting at a temperature of about -50 C. The clouds show up as slightly colder in the opaque image, because the atmosphere above the high clouds is absorbing some of the emitted radiation. Where the atmosphere is cloud-free, the three images tell different stories. The opaque image shows mostly north-to-south temperature variations in the upper troposphere, but the pattern is kind of wavy. Temperatures are much warmer over Northern California, for example, compared to over Colorado and Kansas. This is because there is a jet stream ridge over the West Coast and a trough over the central United States. The trough is a bit to the west of the cloud mass that marks the low-level cyclone; this westward displacement is typical of active midlatitude systems. The image in the transparent, or “atmospheric window” channel, shows surface temperatures beneath cloud-free air. Temperatures are warmest in the desert Southwest, although the ground is starting to heat up a bit too in eastern Montana and Wyoming. There is a sharp color contrast along the West Coast in California and Baja California where the ground is warm and the ocean is cool. The even cooler temperatures farther offshore and over the Iowa-Missouri-Kansas area are low clouds rather than ground temperatures. Because the clouds are low in the troposphere, they are not much cooler than the ground beneath or beside them, so the contrast in temperatures is not all that great, and if you don’t look carefully (or look at a visible satellite image) you might be fooled into thinking that there are no clouds there at all. The image in the semi-transparent window only shows a blurry view of the ground. The Sierra Nevada Mountains show up in California as a line of cooler temperatures, but the contrast between low cloud and bare ground offshore and in the central Plains is gone. The warmest temperatures in this image are off the coast of Baja California, indicating that the middle and lower troposphere is quite warm there. Suppose the satellite detects radiation from a gas with variable concentrations, such as water vapor. In that case, even a single channel might see some widely varying radiation intensities. Where there’s lots of water vapor, the radiation would be coming from near the top of the atmosphere, where it’s cool. Where the top of the atmosphere is dry, the radiation would be coming from lower down in the atmosphere, where it’s warmer. If you have some way of estimating the vertical distribution of temperature (perhaps from the same satellite), you can use these measurements to estimate the concentration of water vapor in the upper atmosphere. And with several channels sensitive to water vapor in ATMO 251 Chapter 8 page 5 of 23 varying degrees, you can extract information on the vertical distribution of water vapor and total amount of water vapor in the atmosphere. One such retrieval of total column water vapor is shown below. The water vapor is expressed as a quantity called “precipitable water”, which is the amount of rain that would fall if all the water vapor were instantly condensed out of the atmosphere and brought to the ground. The retrieval is not possible where it is cloudy, since the liquid water droplets and ice crystals that form clouds are strongly opaque and block the view of the lower troposphere. (Clouds, or otherwise impossible retrievals, are shown in gray or white in the image.) Remember, water vapor is the gaseous form of water and is everywhere; clouds form when the gas condenses into liquid or solid particles. The greatest amount of water vapor is over the Gulf of Mexico in this image. If all the water vapor were condensed and fell as rain, it would cover the Gulf of Mexico to a depth of over 50 mm, or 2 in. By contrast, the atmosphere is so dry over Maine that all the rain that could possibly fall would barely wet the surface. There is little water vapor in the atmospheric column over the Rocky Mountains, too, partly because the relative humidity is low and partly because there’s less atmosphere there because the mountains are so high! 6.2 Images and Gradients The laws of physics make it very difficult to extract detailed temperature and humidity information from satellite observations. On the other hand, there’s no such limitation to horizontal resolution. The horizontal resolution is limited by the size of the satellite receiver and the strength of the radiation. Even the coarsest satellite sounders have a ATMO 251 Chapter 8 page 6 of 23 horizontal resolution of a few tens of kilometers. Satellites designed for imaging are even better. The coarsest meteorological satellites have a pixel size of 4 km, and many are down to 1 km or better. With a lot of satellite data, especially at wavelengths designed to measure water vapor or other constituents, the data is not coming from a single level. Yet one can plot the image on a two-dimensional map. What vocabulary do we then use for variations within the image? Can we talk about gradients, even though we’re not looking at a horizontal plane in the atmosphere? The answer is, yes you can. While the physical quantity may not lie on a plane in the atmosphere, the image lies on a plane on the computer screen or piece of paper. And normally, the variations (gradients) that show up in the images have their counterparts on horizontal surfaces in the atmosphere, even if we don’t know exactly which horizontal surfaces are involved. You can compute a gradient in a satellite image in pretty much the usual way. You generally don’t have contours, but you have the value of the field at every point so you don’t need to do any interpolation. You can compare two values of temperature, and divide by the distance between them. The only thing you might have to be careful about is your coordinate system. Many satellite images are not remapped to a normal projection. So distances near the center of the image might be quite different from distances near the edge of the image. ATMO 251 Chapter 8 page 7 of 23 The water vapor images discussed in the previous section use radiation in the infrared range. Other infrared wavelengths are absorbed by other gases, or not at all. Despite this generality, the term “infrared image” is reserved for images at a wavelength that has almost no absorption and emission by the atmosphere, such as 11-12 microns. In other words, an infrared image looks right through the atmosphere. Although there are slight variations in the ability of the Earth’s surface to emit infrared radiation, for the most part the Earth is a very efficient emitter. Thus, variations in surface temperature are easy to detect as variations in infrared radiation intensity, and they show up well on an infrared satellite image. On a clear day or night, you can tell just how warm or cold the surface of the Earth is from an infrared image. On a cloudy day, it’s a different story. Clouds absorb and emit infrared radiation too. With the water vapor image the clouds weren’t much of a problem because most of the clouds were hidden by the water vapor. In a standard infrared image, there’s nothing to hide the clouds, and in fact the clouds are hiding the surface of the Earth from the satellite. This is a problem if you care about surface temperature, but it’s a good thing if you care about the clouds themselves. As long as the clouds are a different temperature from the underlying surface, it’s possible to easily identify the edges of clouds in an infrared satellite image. Not only can we distinguish clouds from Earth, we can distinguish high clouds from low clouds. The higher the cloud top, generally the colder the cloud top temperature. So an infrared image is useful for looking at different layers of clouds, although it can only see the topmost layer at any given point. The same grayscale plotting convention is often used as with water vapor images. The weakest radiation is coming from the highest clouds, and it is tagged as white. Radiation consistent with warm surface temperatures is tagged black. So even though you’re really looking at different temperatures, your brain tells you you’re looking at clouds. In the example infrared satellite image, both high clouds and low clouds are visible over the Pacific Ocean. The high clouds stand out because they are colored bright white in the image. Low clouds are harder to see, but there is quite extensive low cloud cover over the Pacific too. The telltale sign is the rapid spatial variations from medium gray to dark gray: the temperatures in the Pacific Ocean are much smoother and more uniform than that. ATMO 251 Chapter 8 page 10 of 23 6.5: Visible Images Satellite images from the visible spectrum are unlike the other images discussed so far. As noted before, the Earth doesn’t “glow” in the visible wavelength range, so all the visible radiation detected by a satellite is scattered or reflected from the surface of Earth or the surface of clouds. Since clouds are a very bright white compared to most surfaces, they are easy to detect. An advantage of visible images over infrared images is that visible images easily pick up the contrast between low clouds and the surrounding ground, as long as the ground is not covered with snow or is white for some other reason. If the two surfaces are nearly the same temperature, you may not be able to tell them apart with just an infrared image. An obvious disadvantage of visible images is that, because they rely on a radiation source external to the Earth, they only function during daytime (or during a full moon if you stare hard at the image). There are times of year when visible satellite images are simply not available in polar regions. ATMO 251 Chapter 8 page 11 of 23 Of all the images, it makes the least sense to talk about gradients in the context of visible satellite images. Most features are true discontinuities: here you have cloud, here you don’t. In the example visible satellite image, it’s easy to see the low clouds, because they are just as bright as the high clouds. This ease is both an advantage and a disadvantage, since it’s harder to tell low clouds and high clouds apart in a visible satellite image than in an infrared satellite image. You really need to look at both types of images at the same time to properly diagnose the cloud cover. 6.6 Wind Profilers and Radars Rawinsondes are one source of observations of wind above the surface, but it’s a limited source. The launches are typically twelve hours apart, and there’s only one data point at each level from each sonde. By contrast, radar wind profilers provide data with a frequency of about an hour, and scanning Doppler radars provide data over a broad volume every five minutes. Plus, unlike a rawinsonde, you can use a radar more than once. ATMO 251 Chapter 8 page 12 of 23 the air because they are being pulled downward by gravity. However, they are drifting along with the horizontal wind like a balloon, so the horizontal velocity of the scatterers is an excellent estimate of the horizontal velocity of the air. If the wind is blowing sideways rather than toward or away from the radar, there won’t be any change in the phase of the signal received by the radar, so sideways velocities are undetectable. Indeed, if you think about velocity as a three-dimensional vector and orient a coordinate system so that one component is parallel to the radar beam, that’s the only component of velocity that’s measurable by a Doppler radar. The other two components are perpendicular and invisible. Despite this limitation, there are many ways of determining all three wind components. One simple-sounding approach is to scan along the east-west, north-south, and vertical axes to measure all three components. The catch is that the three scans, being in three different directions, measure winds in three different places. Typically the wind varies by location, so only close to the radar do you get an accurate wind estimate. And if you wanted the wind close to the radar, you could use an anemometer. Winds aloft are much more interesting, because no anemometer measures them. Suppose the scanning radar points up at some angle and then scans a complete 360 degree circle. Typically upward motion is much weaker than horizontal motion, so if the scanning elevation angle is fairly shallow, ATMO 251 Chapter 8 page 15 of 23 the vertical motion will hardly contribute to the radial velocity at all and can be neglected. As long as the elevation angle is not zero, each distance along the scan will correspond to a different height above ground. All that must be done is determine the average horizontal wind components at each distance. This is easily done: assuming there’s not much wind variation horizontally across the radar scan, the wind will be blowing from the direction with the largest component toward the radar and will be blowing toward the direction with the largest component away from the radar. The wind speed is given by those peak magnitudes. ATMO 251 Chapter 8 page 16 of 23 One can estimate winds at different heights from a single tilted scan of the radar. To keep the measurements close to the radar, a higher- angle scan can be used for higher altitudes. The result is data describing the vertical distribution of winds. Since this information is available for each volume scan of the radar, a new set of observations comes in every five minutes or so. A wind profiler takes a different approach. Rather than scanning, it uses three fixed angles for its beams: straight up, fifteen degrees to the east of straight up, and fifteen degrees to the north of straight up. The beam pointing slightly to the east will measure a radial velocity that has contributions from the eastward wind component and the upward wind component. Similarly, the beam pointing slightly north of straight up will receive contributions from the northward and upward components. But with the third beam pointing straight up, the upward component is known exactly, and vector arithmetic can be used to determine the other two components. 6.7 Time-Height Sections of Wind The display of wind observations from a wind profiler or VAD is usually done by means of a time-height section. The time section method takes advantage of the frequent profiles to display information about the time evolution of the wind. Along each vertical column of the chart, wind ATMO 251 Chapter 8 page 17 of 23 6.8 Other Time-Height Sections Time sections can be used to display any meteorological information for which both vertical distribution and time evolution are important. As with horizontal maps, scalar variables are depicted on vertical time sections using contours or isopleths. One such variable is temperature. Temperature changes at different altitudes describe changes in temperature inversions and stratification. If the air column is close to convective instability, temperature drops aloft imply decreasing stability. All of these can be diagnosed from a vertical time section with temperature plotted. Another aspect of the temperature distribution that is relevant for forecasting is the dry adiabatic lapse rate. When the lapse rate is dry adiabatic, the atmosphere is well mixed and neutral. At low levels, knowing where the atmosphere is neutral provides information about pollution distribution and maximum temperature; aloft, it provides information about turbulence. In the example of a temperature time-height section above, isotherms are every 2 degrees C and winds are also shown. The vertical axis is pressure. The image is from a model forecast and covers three days, with time running from left to right. Notice how the warm ATMO 251 Chapter 8 page 20 of 23 temperatures every afternoon are confined to the lowest 100 mb or so; above that level, the temperature responds more to large-scale changes than to the daily heating/cooling cycle. The cooler temperatures on the last day are caused by a cold front passage; note how the winds were from the south during the first part of the forecast but have shifted to NE by the end of the forecast. More useful than temperature for diagnosing the dry adiabatic lapse rate is potential temperature. Rather than looking for a particular vertical rate of variation of temperature, one can look for where the potential temperature is constant with height. And a time section of potential temperature shows all the other important information contained in temperature too. Since potential temperature is proportional to temperature, rises and falls of potential temperature say the same about temperature. Variations in stratification are even easier with potential temperature. The greater the stratification, the larger the rate of increase of potential temperature with height. The more tightly packed the isentropes (potential temperature contours) in the vertical, the greater the stratification. In the example for a potential temperature time-height section, for the same location as the temperature time-height section, potential temperature (in K) can be seen to increase upward, compared to temperature, which decreased upward. Also, the warmest times of the day ATMO 251 Chapter 8 page 21 of 23 correspond to periods in which the potential temperature is constant in the lowest 100 mb or more. This means that the lapse rate is dry adiabatic, exactly what would be expected for the daytime planetary boundary layer. In the model forecast, a squall line is forecasted to pass the station about midway between 12 UTC on the 17th and 00 UTC on the 18th. The gust front of the squall line shows up as lump of low potential temperatures at and near the ground. Fronts and squall lines are usually easier to see in a potential temperature plot than a temperature plot because the high stratification at the top of the pool of cold air corresponds to a cluster of potential temperature isopleths, grabbing the viewer’s attention. 6.9 Vertical Sections The same plotting styles and conventions for vertical time sections apply to vertical sections in space. Vertical sections can be constructed from gridded model output or from rawinsondes. With gridded output the creation of a vertical section is straightforward. However, one will rarely find rawinsondes lined up perfectly. To create the section, one determines the straight line representing the section and then assigns nearby rawinsondes to the points on the line closest to the rawinsonde stations. ATMO 251 Chapter 8 page 22 of 23