Regional and Mesoscale Meteorology Branch

How is this product created now?

Three data sources are used to create the product: Blended TPW from CIRA, which indicates the strength and location of atmospheric rivers impinging on the U.S. West Coast, GFS 850 mb winds (V), which are used to advect the water vapor to a forecast time (every 3 hours), and USGS Global 30 Arc-Second Elevation Data (GTOPO30) terrain elevations (H) (horizontal resolution of approximately 1 km).

CIRA developed the Blended TPW product by combining several data sources. These include Special Sensor Microwave / Imager (SSM/I) passes from the Defense Meteorological Satellites Program (DMSP) and the Advanced Microwave Sounding Unit (AMSU) from Polar Operational Environmental Satellites (POES), and information from Global Positional Satellite (GPS) meteorological data. GOES-West and GOES-East Sounders are used to fill in missing land locations over the continental United States. By combining all of these observations, a smooth product is created.

The blended product provides information over the continental United States primarily from GPS data, and over observation poor regions, such as the oceans, from the SSM/I and AMSU data. The Blended TPW eliminates the bias between data sets and provides an easier and faster product to analyze. This product has been used by forecasters and satellite analysts for years, and they have found it of value in helping to analyze and predict heavy rain and flooding, and also the transfer of moisture from ocean to land. The Blended TPW product is available on AWIPS. The ORI product uses this Blended TPW to calculate an index that represents the amount of moisture blowing up higher terrain. The basic calculation is:

ORI = TPW * V•∇H

TPW has units of kg m^-2, V has units of m s^-1, and ∇H is unitless (m/m). ORI, then, has units of kg s^-1 m^-1. Based roughly on the work of Nieman et al. 2008 (J. Hydrometeorology, 9, 22-47) we set 50 kg s^-1 m^-1 as a threshold below which no rain is likely, to 250 kg s^-1 m^-1, which definitely deserves the forecaster’s attention. (Maximum values are probably in the 500 kg s^-1 m^-1 range.)

Product Examples and Interpretation.

West Coast example from 5 May 2009

V•∇H is an indicator of vertical motion caused by wind blowing up hill. (If the wind is blowing down hill, ORI is set to zero.) In other words, V•∇H is an indicator of terrain-induced “lift.” TPW is an index of atmospheric moisture. ORI, then, is the product of moisture and terrain-induced lift.

Figure 2 shows the TPW field, the GFS 850 mb winds, and the GTOPO30 terrain used to calculate the ORI product shown in Fig. 1. Figure 3 shows the atmospheric river which was impacting the West coast. The ORI product in Fig. 1 is quite detailed, since the topography used has a horizontal grid resolution of approximately 1 km.

Fig. 2. Winds, TPW, and topography (0300 UTC 5 May 2009) used to produce the ORI in Fig. 1. (Click on figures for full resolution.)

Fig. 3. CIRA Blended TPW (ocean only) showing the tongue of tropical moisture impacting the U.S. West Coast.

Example from displays on AWIPS from near Seattle on 14 February 2010

In this an example we will show displays of ORI as they would appear on AWIPS on the WFO scale for a rain event in mid-February 2010. For this example we will focus on the Seattle WFO forecast area. After a brief discussion to give an overview of the event, some static images of ORI and other fields are shown for a single time, 0300 UTC on 14 February to explain the ORI product and its characteristics. A number of the images are also available as loops.

An overview for a portion of the event is shown in Figure 4 using IR images and 700 mb data for 0000 and 1200 UTC on 14 February. A large Pacific frontal zone is moving onshore at 0000 UTC, then passing over the WFO Seattle area by 1200 UTC. Shown in the images is a band of cloudiness trailing off to the southwest. Larger scale imagery shows this to be a fairly extensive plume with substantial total precipitable water (TPW), as seen by the TPW imagery in Figure 5.

Fig. 4. IR image with 700 mb plot for 0000 UTC (left) and 1200 UTC (right) on 14 February 2010.

Fig 5. CIRA Total Precipitable Water (TPW) image at 0000 UTC on 14 February, showing the plume of moisture from Washington all the way back to Hawaii. Click on the image for a loop from 0000 through 1800 UTC.

The precipitable water values used in the ORI calculations are determined from the imagery in Figure 5, as explained earlier in the section “How is the product created now?” In this case the imagery reveals an extensive moisture plume stretching all the way back to near Hawaii, also referred to as an “atmospheric river”. Such moisture plumes are often responsible for the flooding events that occur along the West Coast in the winter season, especially if they remain quasi-stationary. In this case, the entire trough continued to move slowly inland, so the rain event lasted approximately 24 to 48 hours. Clicking on the image in Fig. 5 shows a short loop that illustrates how the plume progressed across the area on 14 February. The other meteorological factor that is used to calculate ORI is the 850 mb wind from the GFS model, usually a short-term forecast. For this case we show the 850 mb analysis for 0000 UTC/14 February in Figure 6. Note the very strong SSW flow ahead of the approaching Pacific front. The 0000 UTC sounding from Quillayute (UIL) had an 850 mb wind of 54 knots from 185 degrees.

Fig. 6. Analysis of height (dm) and temperature (oC) at 850 mb along with observations, for 0000 UTC on 14 February.

The final factor in the ORI calculation is the topography, which of course is rather extreme in western Washington. An image of the “High Resolution” topography from AWIPS is shown in Figure 7 to illustrate the nature of the dramatic elevation changes in this area. In the actual ORI calculation the topography used is derived from the USGS Global 30 Arc-Second Elevation Data (GTOPO30) terrain elevations, with a horizontal resolution of approximately 1 km. An image of this topography is also shown in Figure 7. Clearly the USGS topography at high resolution is very detailed, and by using such detailed topography the ORI product is able to depict the potential for precipitation at a very high resolution with an obvious strong terrain-forced component.

Fig. 7. AWIPS “high-resolution topography” image (left) of the western portion of the Seattle WFO forecast area, compared to an image of the USGS terrain (right).

Next we will look at the actual ORI imagery as it would appear on AWIPS, and compare the imagery to conventional radar from the Seattle KATX 88-D. The time of 0300 UTC on 14 February is chosen for the discussion below, but many of the images can be looped. At 0300 UTC some of the heaviest rain was moving through the higher terrain area between Seattle and the coastline. The ORI imagery as it would appear on the WFO scale is shown for 0300 UTC in Figure 8.

Fig. 8. ORI image at 0300 UTC on 14 February 2010. The scale is shown at the top. Click on the image for a loop from 0000 to 1800 UTC.

A comparison of the ORI imagery in Figure 8 with the topography imagery in Figure 7 shows the strong influence of topography in the ORI product. Recall from the 850 mb analysis in Figure 6 that the 850 mb flow was strong out of the south. This flow interacting with the topography should produce the pattern that we see in the ORI image, that is, elongated west to east bands of higher values of ORI that match the west to east ridges of higher terrain. This is very apparent across the Olympic Peninsula, and we focus on a portion of this area in the zoomed in image shown in Figure 9.

Fig. 8b. ORI image at 0300 UTC zoomed in on a portion of the Olympic Peninsula. The scale is shown at the top.

The values of ORI in the closeup image shown in Figure 8 range up to about 250 kg s^-1 m^-1, which is the value where forecasters should take note for the potential for significant precipitation. In this case the precipitation event was relatively short-lived as the storm system kept moving, and these high values were not maintained for very long. Of course, at this point experience among most forecasters with the ORI product is somewhat limited, and additional real-time use by forecasters under varying conditions will help to fine tune the likely ORI values of importance. The units for the product are kg s^-1 m^-1, which are not very meaningful to forecasters. The point is that the ORI product yields values that cannot be strictly translated to a particular amount of rainfall, but should be treated more as an index, much in the way that we consider CAPE for instability. Because of this, however, it will take some experience using the product to determine values of importance, and the 250 value noted earlier is an estimation based on earlier work with some California events and other experience by satellite analysts. Another point to note is that the ORI product is a short-term forecast for a particular time, using the 3-hour and 6-hour forecasted 850 mb winds to advect the analyzed TPW from the satellite imagery. Changes in the moisture that may occur in the forecast time will, therefore, not be taken into account. Generally these changes are not likely to be that great in a 3 to 6 hour period. Probably more important to remember is that the ORI product represents the orographic component of lift only, and does not consider any larger scale forcing (either up or down) that may be occurring with the associated synoptic system.

Because the ORI product is radar independent, there is no limitation from terrain blocking as there can be for radar data. The blocking problem is of course most prevalent in areas of higher and complex terrain. This point is nicely made using the combined image in Figure 9, which combines the ORI image in Figure 8 with a low level (0.5o elevation angle) scan from the Seattle (KATX) radar. In this combined image we see that there is a significant area where the radar fails to scan the higher terrain area, at least at lower elevation angles. Another point to be made from the combined image in Figure 9 is that the scale of the ORI output is considerably finer than what the radar can resolve, except perhaps if the precipitation is very close to the radar site. For example, the elongated west to east ORI band at the top of the figure is resolved by the radar as a fairly broad area of reflectivity.

A comparison of the ORI product with a higher elevation angle (the 1.5o elevation angle) radar scan is shown for 0300 UTC in Figure 10 (you can click on the image to get a loop through much of the event). The same point can be made about the absence of detail in the reflectivity image compared to this higher elevation angle. One final point to be made about the ORI product is that in the absence of much varying terrain, the calculated ORI value will be small or even zero. This of course does not necessarily mean there is no precipitation in this area, just no orographically forced precipitation. You can see this effect by looping Figure 10; note how there is radar echo but no calculated ORI over the ocean area north of the Olympic Peninsula (the Strait of Juan de Fuca). This same effect is seen over land that is relatively flat, as in the area near Seattle, and extending northward into British Columbia.

Fig. 9. ORI image at 0300 UTC, as in Fig. 8, combined with an image of radar reflectivity from the 0.5o elevation scan of the KATX radar.

Fig. 10. Combined ORI and radar image, as in Fig. 9, except for the WFO scale and with a radar elevation angle of 1.5o elevation scan. Click the image for a loop from 0000 to 1800 UTC on 14 February.

Just how well do the ORI values relate to rainfall for this case? Unfortunately, it is usually impossible to get a precipitation analysis with the same high resolution as the ORI product, or get measurements in areas where the ORI values are highest, since these will tend to be in rough terrain. A look at precipitation totals for the 24-hour period ending at 1200 UTC on 14 February from the NWS Precipitation Analysis is shown in Figure 11. There are some high amounts within the Olympic Peninsula indicating this was a respectable although not unusual precipitation event. But the analysis does not reveal the kind of detail seen in the ORI product. The same can be said for the 6-h precipitation analysis shown in Figure 12. There are hints of some west to east banding in the Olympic Peninsula area, but the main point to be made here is that the higher ORI values for this case correspond to a period of significant rainfall that moved through the area. More experience with different rain events will allow forecasters to better associate the ORI values with potential rainfall.