Main objectives of 2A23 are (1) to detect bright band (BB) , (2) to Classify rain type, and (3) to detect warm rain.

Algorithm Overview

2A23 uses two different methods for classifying rain type; one is vertical profile method (V-method), and the other is horizontal pattern method (H-method). Both methods classify rain into three categories; (1) stratiform, (2) convective, and (3) other. To make the result user-friendly, 2A23 outputs a unified rain type. Further information about 2A23 can be found in Awaka et al., (1998).

The V-method starts with a detection of BB. The detection of BB is made by a spatial filter which is based on the second derivative of Z with respect to range and by imposing several conditions on BB, such as Z above BB peak should decrease appreciably, the height of BB should appear almost at the same height, and so on. One of the major condition imposed on BB is that the height of BB must be located in a BB window, whose range is:

from freezH - 1.5 km to freezH + 1.5 km

where freezH is the height of freezing level estimated from a climatological surface temperature at sea level, Tsurface, by the following formula:

freezH = Tsurface / Tlapse

and where Tlapse is the lapse rate of temperature (2A23 assumes that Tlapse = 6.0 [deg/km]).

After the BB detection, the V-method goes on to classify rain type. The outline of rain type classification by the V-method is as follows:

(a) When BB exists, rain is basically classified as stratiform,

(b) When BB is not detected, and maximum value of Z at a given angle bin exceeds a convective threshold, rain type for this angle bin is classified as convective,

(c) Other type is defined as not-stratiform and not-convective.

It should be noted that other type of rain by the V-method is defined as not convective and not-stratiform: this means that:

(1) there exists appreciable radar echo but it is not strong enough to be convective, and

(2) BB is not detected.

The H-method also classifies rain into 3 categories: stratiform, convective, and other, but with the definitions of these being different from those of the V-method.

The H-method is based on the University of Washington convective/stratiform separation method (Steiner et al., 1995), which examines the horizontal pattern of Z at a given height; where Z has a 2 km horizontal resolution. In 2A23, the following modifications are made:

(a) Instead of examining a horizontal pattern of Z at a given height, a horizontal pattern of Zmax is examined; here, Zmax is the maximum of Z along the range for each antenna scan angle below freezH (minus 1 km margin).

(b) Parameters are changed so that they may be suitable for the TRMM data with 4.3 km horizontal resolution. Choice of parameters was made before the launch of TRMM using a test Ground Validation (GV) data in such a way that a 4.3 km resolution data produces almost the same result as that with a 2 km resolution data.

(c) Other type of rain is introduced to handle noise.

In the H-method, detection of convective rain is made first. If one of the following condition is satisfied at a pixel, which correspond to the angle bin data being considered, it is judged that the pixel is a convective center:

(a) Zmax exceeds a convective threshold, or
(b) Zmax stands out against the background area.

Rain type for a convective center is convective, and rain type for the pixels nearest to the convective center is also convective.

In the H-method, if rain type is not convective and if the rain echo is certain to exist, rain type is stratiform.

In the H-method, rain type is 'other' if the radar echo below freezH (minus 1 km margin) at a given angle bin is possibly noise. This means that the other type by the H-method includes the case of (i) noise, and (ii) cloud.

2A23 outputs a unified rain type. The unified rain type is expressed by 2 digits: the first digit indicates the rain type (1: stratiform, 2: convective, 3:), and the last digit indicates a level of confidence, which decreases as the number increases. Unification of rain type is made in such a way that the rain types by V-method and H-method can be reconstructed from the unified rain type if we use a suitable table.

Detection of warm rain is also made in 2A23. When the following two conditions are satisfied, it is concluded that there exists warm rain:

(a) Storm top of warm rain is much lower than the height of freezing level. (This condition implies that when BB is detected, it is not warm rain because the existence of BB means that the storm top is higher than the height of freezing level).

(b) Warm rain must be isolated from the other rain certain areas.

File Format

The file content description for 2A23 can be obtained from the Volume 4 - Levels 2 and 3 File Specifications provided by the TRMM Data and Information System (TSDIS). It is available at: http://tsdis02.nascom.nasa.gov/tsdis/Documents/ICSVol4.pdf.

Caveats

(1) Bright band (BB) detection has angle bin dependence because of smearing of BB near the antenna scan edges. Sample test indicates the rate of BB detection is about 80% near nadir directions (at antenna scan angle ranging from about -7 to about +7 degrees), but the rate decreases as low as about 20% at the antenna scan edge.

(2) When a large numbers of orbit data are processed, statistics of rain type shows angle bin dependence because the rain/no-rain judgment in 1B21 also seems to have angle bin dependence.

(3) Bright band detection and rain type classification are carried out for rain-certain case only. For rain possible, rain type is automatically classified as other type.

(4) When rain is 'certain' and the unified rain type is 'other', it most probably indicates that there exists cloud only.

(5) Side lobe clutter may still affect the detection of BB, hence affects rain type classification as well, though a side lobe clutter rejection is tried in the algorithm.

References

Awaka, J., T. Iguchi, and K. Okamoto, 1998: "Early results on rain type classification by the Tropical Rainfall Measuring Mission (TRMM) precipitation radar", Proc. 8th URSI Commission F Open Symp., Aveiro, Portugal, pp.143-146.

Steiner, M., R.A. Houze, Jr., and S.E. Yuter, 1995: "Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data", J. Appl. Meteor., 34, pp.1978-2007.