NCEP/Climate Prediction Center ATLAS No. 5

A Precipitation Climatology for Stations in the Tropical Basin; Effects of ENSO


4. Rainfall Anomaly Histograms, 1955-96

Figures 3a and 3b show the actual rainfall amounts (mm) and the corresponding percentiles, respectively, as companion figures on an individual station basis for running 3-month seasons spanning the 1955-96 period chronologically. The figures are presented as histograms. In Fig. 3a, the seasonal cycle of the quartile boundaries (25 %ile: lower light line; 50 %ile [i.e., median]: dark line; and 75 %ile: upper light line) are plotted as well as the actual rainfall for the given season/year as a vertical bar. The year labels shown on the horizonal axis are placed at the center of the calendar years rather than at Dec-Jan-Feb, with the latter denoted by tick marks. The same ordinate scale is used for most of the stations; however, for some dry stations (e.g. Honolulu) the scale is reduced. The ENSO status of each boreal winter (see below) is shown underneath the main panels of the histogram.

Figure 3b provides information similar to that of Fig. 3a , but from a somewhat different perspective. First, percentile values are shown instead of actual rainfall amounts. Rather than using solid vertical bars for the histogram, discrete characters (numerical symbols) are plotted to form the vertical column. These are plotted with respect to the median, in the middle of the vertical range, in contrast to zero as in Fig. 3a. An upward extension of the characters from the median indicates above-median precipitation, and blow-median precipitation is indicated by a downward extension from the median. The percentiles are rounded to the nearest 5 %ile: 50 %ile indicates 47.5-52.5 %ile, 85 indicates 82.5-87.5 %ile, etc. The 0 and 100 %iles refer to 0.0-2.5 and 97.5 to 100.0 %iles, respectively. The general up-and-down patterns formed by the character-created bars in Fig. 3b approximately correspond to the patterns of the solid bars of Fig. 3a , except that in Fig. 3a the median is shown as a seasonally varying actual rainfall quantity whereas in Fig. 3b it is defined as the midpoint of the vertical range of the plot for all seasons. In Fig. 3b, characters (numerical symbols) are used to form the histogram bars such that the choice of character shows the climatological median value of the rainfall for the season represented (as does the dark line in Fig. 3a). Just as the median in Fig. 3a goes through the same annual cycle within each of the 42 years plotted, the choice of character (which remains the same within any single histogram "bar") goes through the same cycle during each year. Thus, for example, when a bar for a given station uses the "3" symbol for Feb-Mar-Apr 1955, the bars for Feb-Mar-Apr of 1956, 1957, and all other years through 1996 also use the "3" symbol. The rainfall amount categories represented by each character choice are shown at the top of Fig. 3b. These increase somewhat nonlinearly from 0 (representing the driest category--less than 100 mm of rainfall received in a 3-month period) to 9 (representing the wettest category--1200 mm or more received in a 3-month period). The choice of rainfall cutoffs for each of the 10 categories was made on the basis of creating approximately equal frequencies of each category across all seasons and stations. The character choice enables a user of Fig. 3b to determine whether a period of rainfall deficit coincides with a rainy part of the year or a dry part of the year, without having to look at any other figure such as Fig. 2 or even Fig. 3a. To further facilitate this determination, shading appears in Fig. 3b for the six driest 3-month periods of the year. The shading appears identically for each year as a fixed cycle in similar fashion to the choice of symbol used to form the histogram bar.

Figures 3a and 3b are designed to provide information about the seasonal, interannual, and decadal timing of the dry periods. Intervening neutral and wet episodes are also shown. Figure 3a can be considered the most basic rainfall data presentation of this atlas. Together, the two parts of Fig.3a and Fig.3b can be used for individual case studies for one or more stations. Along the bottom of the panels of these figures are shown the ENSO status for each boreal winter: WARM (El Ni Niño), COLD (La Niña) or blank (neutral). These are provided to aid with rainfall studies that take ENSO into consideration, as it has been established that many of the regions to which the stations included here belong, are substantially influenced by ENSO (Ropelewski and Halpert 1987, 1996). The criteria for classification of each boreal winter's ENSO status, as well as the rainfall consequences of ENSO for the stations studied here, are discussed below in section 6. As discussed there and detailed further in the appendix, the ENSO classification criteria differ from those used in Ropelewski and Halpert (1987, 1996), but nonetheless result in a fairly similar classification.


back to the table of contents