Interpretation of WV Imagery

Water vapor imagery is derived from radiation at wavelengths around 6-7 ^m. Though this is not an atmospheric window, it is part of the spectrum where water vapor is the dominant absorbing gas. It has a strong absorption band centered on 6.7 ^m. In regions of strong absorption, most of the radiation reaching the satellite originates high in the atmosphere. The stronger the absorption, the higher is the originating level of the emission that ultimately reaches the satellite. As the Relative Humidity (RH) decreases the main contribution of the radiance received by the satellite comes from lower in the troposphere.

WV imagery is usually displayed with the emitted radiation converted to temperature like IR imagery. Regions of high upper tropospheric humidity appear cold (bright) and regions of low humidity appear warm (dark) i.e. when the upper troposphere is dry, the radiation reaching the satellite originates from farther down in the atmosphere where it is warmer and appears darker on the image. In a normally moist atmosphere most of the WV radiation received by the satellite originates in the 300-600 hPa layer. But when the air is dry some radiation may come from layers as low as 800 hPa. Due to the general poleward decrease of water vapor content, the height of the contributing layer gets lower and lower towards the poles.

Since clouds do emit radiation in this wave band, high clouds may be seen in this type of imagery. Thick high clouds as Cb anvils stand out prominently in both WV and IR imageries. Broad-scale flow patterns are particularly striking in WV imagery. This is because WV acts as passive tracer of atmospheric motions. WV imagery is therefore useful for displaying the mid tropospheric flow (for example), upper tropospheric cyclones are defined clearly by moist spirals or comma-shaped patterns. Subsidence areas appear dark. Jet streams are delineated by sharp gradients in moisture with dry air on the poleward side. Even when a WV image indicates a very dry upper troposphere, there may well be moist air near the surface. Moist air or cloud in the lower half of the troposphere is not depicted well in WV imagery.

Figure 1: Karman cloud vortices in Pacific Ocean as they are seen in visible images Rainfall

Rainfall is not only an important parameter for agro-meteorology, it is also a vital component of earth's hydrological cycle. Oceans receive heat from the sun, and the evaporation takes place. When the same evaporation travels vertically, it gets condensed to form clouds and eventually precipitation. During this process, enormous amount of energy is released which is called the latent heat of condensation. This is one of the major sources of energy that drives the circulation in tropical atmosphere. Hence, the knowledge of rainfall and its distribution at current time is also important for its future prediction.

There are several techniques to derive rainfall from satellite observations. The earliest developed methods that are useful even today are based on the visible/infrared observations from satellites. Techniques based on visible sensors rely on the identification of cloud types. Each cloud type is assumed to have different rain intensity, and then the rain is derived based upon the extent of each cloud type. In infrared-based methods, the most common approach is to find cold clouds (say, colder than 250oK) within an overcast area. The raining potential of the clouds is proportional to the fractional area covered by cold clouds, and thus the rainfall derived. More complex techniques use both visible and infrared observations to create a bi-spectral histogram of the cloud images. Bi-spectral histogram method is a simple technique in which the clouds can be classified based on the combination of cloud signatures in visible and infrared frequencies. Then the rainfall is derived by estimating the extent of each type of cloud and multiplying it by the a-priori rain potential of respective classes. However, all the rain-retrieval techniques based on visible/ IR observations are basically "inferential" in nature, because these sensors can sense the clouds (that too, the top surfaces of the clouds) but not the actual rain, that occurs at several layers below the clouds. Visible/IR techniques make a "guess" about the rainfall based on the cloud features. Due to this shortcoming, the estimates of rainfall based on visible/IR technique are not very accurate on instantaneous time scale. However, long time averages (e.g. daily, weekly, and monthly) of rainfall are better and usable for practical purposes.

On the other hand, rainfall estimation techniques based on microwave frequencies (0.1 cm to 100 cm wave length) are more direct in nature. Due to their large wavelengths, these frequencies can easily penetrate clouds. However, these frequencies interact effectively with rainfall. Let us consider the case of passive microwave methods. In these techniques, the microwave instrument onboard satellite does not have any source of microwave illumination. It has just a receiver that can gather the microwave emission coming from earth, ocean or atmosphere. Due to small emissivity in microwave region, ocean surface emits small amount of microwave radiation. When the rainfall occurs over a layer in the atmosphere, two different processes take place.

The atmospheric rain layer itself emits microwave radiation and thus the radiation received at satellite is greater than the radiation received in no-rain situation. This process is predominant at lower frequencies (e.g. 19 GHz, or about 1.5 cm wavelength). On the other hand, the rain drops, ice and snow particles, scatter the microwave radiation (particularly at higher frequencies, e.g. 85 GHz, or about 0.3 cm wavelength) that is coming up from the ground. In this case the radiation received at satellite will be smaller than that in norain situation. In both the cases, the change in the microwave radiation (measured in terms of brightness temperature) can be related to the intensity of rainfall. Various algorithms have been developed in the past that use either low frequency or high frequency, or a combination of both. It is to be noted that for emission based algorithms (using lower frequencies of microwave e.g. 19 GHz), it is important that the emission from the background should be uniform and also as little as possible, so that the emission from rainfall can be detected clearly. So emission based algorithms are effective only over the ocean surfaces, while the scattering based algorithms using higher frequencies of microwave, can be used over the ocean as well as over the land. Special Sensor Microwave Imager (SSM/I), and TRMM Microwave Imager (TMI) are good examples of passive microwave sensors that are quite effective in the determination of global rainfall (Figure 2).

Figure 2: Global annual rainfall observed by Special Sensor Microwave/Imager (SSM/I) (Picture courtesy : rst.gsfc.nasa.gov)
0 0

Post a comment