Multiple Atmospheric Rivers Impact California in Early 2017

The state of California has been suffering from a multi-year drought that has severely depleted water resources and reservoir levels. Recent winters have failed to produce precipitation and mountain snows to replenish the losses during the dry summers. However, the situation has rapidly changed this winter, particularly in the past week when multiple atmospheric rivers have impacted the state.

An atmospheric river is a concentrated channel of deep moisture that is transported from the tropical Pacific Oceanic regions to the West Coast of the United States.  These events are often associated with prodigious amounts of rainfall and mountain snows that lead to flooding, mudslides, and avalanches.  We have seen such events this past week impact California, especially the central and northern parts of the state.  CIRA’s total precipitable water product in Figures 1a and 1b depict two separate atmospheric rivers impinging on central California from 8 and 10 January 2017, respectively. The first wave transported a plume of tropical moisture from the south-southwest, which led to massive rainfall and high snow levels.  The second atmospheric river on the 10th was less directly connected to the tropics (coming in from the west-southwest), but nonetheless exhibited a well-focused transport of high moisture content.  Widespread flooding and mountain avalanches have resulted from these moisture plumes as the impacted California, as well as dramatic replenishment of reservoirs.


Figure 1.  CIRA total precipitable water product (inches) valid at (a) 2100 UTC 8 Jan 2017, and (b) 2100 UTC 10 Jan 2017.


SPoRT’s real-time instantiation of the Land Information System (aka “SPoRT-LIS”) has nicely depicted the substantial replenishment of the moisture content in the soils over California.  The SPoRT-LIS is an observations-driven, ~3-km resolution run of the Noah land surface model that consists of a 33-year climatology spanning 1981-2013, and real-time output at hourly intervals sent to select NOAA/NWS partnering forecast offices.  The one-year change in the SPoRT-LIS total column soil moisture at 1200 UTC 11 January (Fig. 2) shows large increases over most of California, particularly in the higher terrain (given by blue and purple shading).  [At the same time, annual degradation in soil moisture can be seen across the central and eastern U.S.]  Interestingly, a substantial portion of California’s annual soil moisture increases has occurred in just the past week (Fig. 3; SPoRT-LIS total column soil moisture change over the past week).  One can certainly see the important role that atmospheric rivers play in being “drought busters”!


Figure 2.  One-year change in the SPoRT-LIS total column relative soil moisture, valid 1200 UTC 11 January 2017.



Figure 3.  One-week change in the SPoRT-LIS total column relative soil moisture, valid 1200 UTC 11 January 2017.


A map of the SPoRT-LIS daily soil moisture percentiles from 11 January highlight the very wet anomaly over California relative to the 33-year soil moisture climatology (Fig. 4; similar to the pattern of annual soil moisture change from Fig. 2).  Blue shading denotes greater than or equal to the 98th percentile, thus indicating unusually wet soils on the tail end of the historical distribution.


Figure 4.  SPoRT-LIS total column relative soil moisture percentile from 11 January 2017.


Finally, SPoRT is acquiring and assimilating in real time the Soil Moisture Active Passive (SMAP) Level 2 (L2) retrievals produced by NASA/JPL into an experimental version of the SPoRT-LIS.  SPoRT is a SMAP Early Adopter and has a funded project to conduct soil moisture data assimilation experiments with LIS and evaluate impacts on land surface and numerical weather prediction models.  Figure 5 shows SMAP L2 retrievals of the evening overpasses from ~0000 UTC 11 January.  Panel (a) is the 36-km resolution radiometer product, while panel (b) shows the enhanced-resolution product, obtained from the SMAP radiometer by using Backus-Gilbert optimal interpolation techniques to provide data on a finer (9 km) grid.  The enhanced-resolution product provides much more detail of the wet soils in California, while retaining the same overall regional patterns as the original 36-km retrieval.  Given the loss of the active radar component of the SMAP mission, SPoRT plans to assimilate both the 36-km and 9-km products separately, and compare results on model accuracy.


Figure 5. SMAP Level 2 soil moisture retrievals for the evening overpasses from ~0000 UTC 11 January 2017; (a) 36-km resolution product; (b) enhanced 9-km resolution product.

Update from Tucson

Just a quick initial note to show a few things we have been looking at with NASA SPoRT LIS and some of the satellite imagery so far.

1) We have had a limited sample size so far this year, but we have been looking at integrating LIS volumetric soil moisture and relative soil moisture into the Dust Storm decision making process (both short fused warnings and our longer term new “Watch”-like product).  Here is the 0-10cm Volumetric soil moisture image from Wednesday,  coincident with a moderate outflow (up to 20 kts) that pushed northward from Pima county into Pinal County. Not a strong outflow, but  one that can generate enough dust for reduced visibilities (likely not to warning levels).   soilmoisture2

In this case, with percentages in the 18 to 21 percent range in the origin area and 12 to 14 percent in the path of the outflow, there were no indications whatsoever of reduced visibilities.  We will be watching to see if the (relatively) elevated soil moisture in the origin area is any indication of limited potential and strength of blowing dust issues as the season progresses.

We are also trying to incorporate the more shallow volumetric and relative soil moisture levels into heightened awareness for flash flood threat areas daily.  More on this later.

2) We have been impressed with the superior accuracy and versatility of the CIRA LPW products.  Here is a recent comparison with the AMSU and SSM/I Blended Total Precipitable Water product: tpw compare

About 1.3″ from CIRA versus 22mm (0.86″) for the Tucson area.  12Z sounding showed 1.76″.  Add afternoon mixing in there and 1.3″ worked much better.  With our typically deep boundary layer and elevated subcloud layer, the individual lower layers of the LPW have also been useful to help determine the threat of dry vs. wet microburst activity with initial convective development.

We continue to evaluate the NESDIS QPE and  Passive Microwave Rain Rate imagery.  More on this later as well.

Tropical wave brings historic rains to San Juan PR – July 18, 2013

A tropical wave moved across Puerto Rico on Thu July 18, 2013 and brought historic rains to the San Juan metro area dumping between 8 to 10 inches of rain in a 24-hr period with most of these rains (8.91 inches) falling in a 6-hr period between 11AM to 430PM. These rains had significant impacts across the San Juan area causing significant flash flooding in area streams and rivers and greatly disrupting ground and air traffic. At one time, over 12 flights scheduled to arrive at the San Juan Luis Muñoz Marin Int’l Airport (SJU LMM) were diverted to the Borinquen airport in northwest PR and the Charlotte Amalie airport in St. Thomas. The 9.23 inches of rain that fell at the SJU LMM Int’l Airport was the second highest 24-hr total ever observed in any single day since records began in 1898.  To read more about other climate records broken with this event, click here.


This 36-hr animation loop (click above for animation) of total water vapor content product from CIRA starting from 18Z Wed July 17 through 03Z Fri July 19 shows the progression of the tropical wave from east of the Lesser Antilles across PR. Values of 60 mm (2.36 inches) were observed over PR with this wave. These values were near the 99th percentile for mid-July according to a precipitable water climatology study done by the Rapid City SD Forecast Office.


Composite Reflectivity animation loop (click above for animation)  from the TDWR from 1239Z through 1609Z shows several clusters of showers and thunderstorms moving over the San Juan area and northeast PR. Just after 1609Z, the TDWR failed likely due to intense lightning activity cutting out power at the radar site.


GOES 4-km infrared imagery animation (click above for animation) shows very cold and intense convection with a large area of tops colder than -70C with tops as cold as -82C at 1710 UTC just north of San Juan.

As part of an evaluation project between WFO SJU and NASA SPoRT, the WFO SJU is evaluating the NESDIS Quantitative Precipitation Estimate (QPE), a satellite-based precipitation estimation product that is derived from both Infrared and microwave data. The image below shows a maximum of 3.14 inches right over the San Juan area during the 24-hr period ending 8AM Fri July 19. Because of the coarse resolution of GOES-IR channel (currently 4km), the satellite is going to miss some of the smaller-scale extremes that will show up in individual gauge reports since the satellite represents an average value over the entire pixel. Even taking that into consideration, satellite based estimates were greatly underestimated for this event. Another factor that may have contributed to the underestimation of precipitation by GOES was the temporal resolution. Rain rates are currently derived every 15 minutes and that didn’t seem to keep up with the rapid evolution and explosiveness of the convection. At one time during the event, 15-minute rain rates were 0.75 inches with 1-hr rain rates of 3.10 inches. However, in the future GOES-R era, both the spatial and temporal resolution will be improved by a factor of 2 with the resolution of the infrared channel at 2km and routine imagery available every 5 minutes.


Image Credit: Bob Kuligowski – NESDIS Center for Satellite Applications and Research (STAR)

The 24-hr rainfall total (image below) precipitation ending for the same period described above shows widespread rainfall amounts of 1-3 inches of rain across all of PR. Over the western half of PR, the satellite-based estimates appeared to have overestimated the amount of rain. This is because the satellite can’t see through thick cirrus clouds and incorrectly assigns a rain rate because the cirrus appear to be a cold cloud. This is a well-known limitation of the current algorithm to derive rain rates from current GOES.


Image Credit: Anita LeRoy – NASA SPoRT

Tropical Moisture Plume Spells Heavy Rainfall for the Southeastern U.S.

The Southeastern U.S. is poised for substantial amounts of rainfall in the coming days due to a deep moisture plume with a connection all the way to Central America.  Figure 1 shows the SPoRT MODIS air mass RGB product centered on the Southeastern U.S.  An upper low pressure area is slowly retrograding to the west, and is situated over western Missouri overnight on 3 July.  The dry air wrapping around the southern/eastern edge of the upper low is evident by the red/orange shading in the air mass RGB over Arkansas and Missouri.  At the same time, a Bermuda high pressure is becoming entrenched over the western Atlantic, with dry air wrapping around its west side, given by the orange plume to the east of Florida in Fig. 1.  In between these two systems is a large plume of very moist air (given by the green shading in Fig. 1), accompanied by extensive cloud cover and precipitation.  The deep southerly flow between these systems is advecting rich tropical moisture northward from the Gulf of Mexico into the eastern U.S.

Figure 1. MODIS air mass RGB (valid 0715 UTC 3 July 2013), depicting the deep moisture plume as denoted by the green colors, extending from the Gulf of Mexico to the U.S. East Coast.  The moisture is being transported northward between a strong Bermuda high over the Atlantic and an upper-low over Kansas City, MO.

Figure 1. MODIS air mass RGB (valid 0715 UTC 3 July 2013), depicting the deep moisture plume as denoted by the green colors, extending from the Gulf of Mexico to the U.S. East Coast. The moisture is being transported northward between a strong Bermuda high over the Atlantic and an upper-low over Kansas City, MO.

The moisture plume can be traced back to Central America by examining CIRA’s total precipitable water (TPW) product.  Figure 2 shows the CIRA TPW from the early morning hours of 3 July.  Very high TPW values exceeding ~48 mm (~2+ inches) extend from the Yucatan peninsula of Mexico and western Caribbean Sea, across  the eastern Gulf of Mexico and all the way north along the U.S. East Coast.  This persistent moisture plume has already been responsible for substantial flooding in parts of the Northeastern states.  The CIRA TPW anomaly (i.e., percent of normal TPW in Fig. 3) shows the very high anomalies in the Northeastern U.S., nearing 200% of normal values.  TPW values are not nearly as anomalous in the eastern Gulf, as deep tropical moisture is more typical there for this time of year.  However, the TPW values are still marginally above normal in a distinct corridor from the Gulf/Caribbean Sea into the Southeastern U.S.  The persistence of this feature will continue to bring a prolonged period of heavy rainfall into the southern/eastern U.S. for the next several days.

Figure 2.  CIRA total precipitable water (TPW, in mm), valid 1126 UTC 3 July 2013.

Figure 2. CIRA total precipitable water (TPW, in mm) product, valid 1126 UTC 3 July 2013.

Figure 3.  CIRA TPW anomaly product, valid 1126 UTC 3 July 2013.

Figure 3. CIRA TPW percent of normal product, valid 1126 UTC 3 July 2013.

NESDIS GOES-R Quantitative Precipitation Estimate and CIRA Layered Precipitable Water Evaluations…

Evaluations of the NESDIS Quantitative Precipitation Estimate (QPE) and CIRA Layered Precipitable Water (LPW) products have been ongoing for a portion of March at three of SPoRT’s collaborative West Coast NWS offices (Monterey, Eureka, and Medford).  The GOES-R QPE is an experimental satellite-based precipitation estimation algorithm using both IR and microwave data from the GOES-East and West satellites.  A suite of products are currently being generated and ported into AWIPS I, including a 15-minute rainrate product and several accumulation products on timescales of 1-hour, 3-hours, 6-hours, 1-day, 3-days and 7-days.  The IR and microwave data used to generate the QPE have different timescale resolutions, about 15 minutes and 1+ hours, respectively.  IR data are used to infer cloud top temperatures and heights, and rainfall rates are derived from this information.  Further calibration with microwave data from the TRMM and NOAA 18/19 helps to fine-time the precipitation accumulations.  However, calibration with tropical environments does not always necessarily translate well in mid-latitude environments, even coastal marine environments.  As a result, precipitation in areas with relatively think cirrus clouds can be overestimated, while the QPE product is thought to perform better in convective situations for operational purposes.

Unfortunately, during the evaluation period, precipitation has generally been below normal, but a couple of events have allowed for a few evaluations of these products so far.

First, the good…

On March 16th, one forecaster noted that the QPE and LPW data were useful for helping to “comfirm that heavy precipitation was not going to be a factor with the incoming front.  QPE was small offshore, and TPW matched model TPW pretty well.”  Below are images of the QPE (Image 1) and LPW (Image 2) as viewed through AWIPS at that time.

Figure 1.  NESDIS QPE 6-hour accumulation product valid 0600-1200 UTC 16 March

Figure 1. NESDIS QPE 6-hour accumulation (inches) product valid 0600-1200 UTC 16 March

Image 2.  CIRA Layered TPW product valid 1500 UTC 16 March.  From upper left clockwise: Total precipitable water, surface-850mb LPW, 700-500mb LPW, 850-700mb LPW

Image 2. CIRA Layered TPW product valid 1500 UTC 16 March. From upper left, clockwise: Total precipitable water, surface-850mb LPW, 700-500mb LPW, 850-700mb LPW

Although relating that the impacts of the LPW product to the forecast process were small, the forecaster did express high confidence in the values and asserted that the value of the layered PW product over a standard total column product was large.  With increased forecaster confidence, details about the incoming storm moisture were noted in the Area Forecast Discussion that morning.

I am eager to evaluate this product over the upcoming weeks, especially during strong moisture advection events concurrent with low or mid-level jets.  It will be interesting to see if having satellite-based information about the vertical distribution of moisture and determining intersecting regions of isentropic lift within these vertical layers will provide better assessments of precipitation.

Now, for the not so good…

The QPE product has tended to overestimate precipitation where cold, thick cirrus clouds are present, making it difficult or impossible to infer reasonable precipitation amounts from the QPE product alone.  On March 20th, a survey respondent noted that “with so much high-level cold cloud tops, the QPE product was virtually unusable.”  In this particuar case, of mostly stratiform precipitation ahead of a frontal boundary, the QPE product far overestimated precipitation east of the Cascades in southern Oregon and underestimated the rainfall near and along the coast.  Notice the swaths of heavy rainfall across northern CA and SE Oregon in the QPE product in Image 3.  However, notice the general lack of precipitation in these areas per the Stage-III product in Image 4 below.

Image 3.  NESDIS 24-Hr QPE product for Oregon and sourrounding areas, valid period ending 12 UTC 20 March

Image 3. NESDIS 24-Hr QPE product for Oregon and sourrounding areas, valid for the period ending 1200 UTC 20 March

Stage-III 24-Hour precipitation for the period ending 12 UTC 20 March

Image 4.  Stage-III 24-Hour precipitation totals for the period ending 1200 UTC 20 March

The forecaster went on to state that “perhaps [the QPE] will be better in the summer with convection…or perhaps in the next few days with the colder NW flow showery regime.”  Indeed, this product is expected to perform better in such situations.

Despite some of the limitations discussed so far, these products do show potential for assessing atmospheric moisture content and precipitation.  The SPoRT team and our collaborative offices will be conducting further evaluations of both stratiform and convective precipitation cases on the West Coast and elsewhere to provide input about these potentially valuable products.

Thanks to Anita LeRoy, Geoffrey Stano and Kevin Fuell for help in providing imagery for this post.

Moisture Plume in Layers

Moisture plumes from the tropical Pacific can extend into the mid-latitudes, and the CIRA Layer Precipitable Water (LPW) product based on microwave (AMSU, MHS) and infrared (AIRS) sounding instruments from NASA and NOAA polar-orbiting satelliltes provides information on the amount of moisture in each layer.  Traditional total precipitable water (TPW) data only give part of the picture and Water Vapor (WV) imagery only captures the upper tropospheric moisture.  Note here how the GOES WV imagery from the NASA GHCC site agrees well with the values of 2-4 mm in the 500-300 mb layer between Hawaii and the west CONUS.


500-300 mb Layer Precipitable Water by CIRA, 13 March 2013, 2100Z


GOES Water Vapor imagery from NASA GHCC site, 13 March 2013, 2130Z

In the images below, the surface to 850 mb layer shows a wide plume of 0.5 to 0.75 inches of PW extending from Hawaii to Washington and Oregon. Moving upwards, the 850 to 700 mb layer continues to show a wide swath of moisture (~0.25 to 0.5 inches) in this same area, with a sharp gradient to the east.  Lastly, the 700 to 500 mb layer shows a more narrow moisture swath, but still with values ranging from ~0.25 to 0.33 inches, and extending into the northwest CONUS. The observations of vertical distribution of moisture in data void regions can be compared to NWP models as well as applied to estimating the available moisture at low levels for potential precipitation and flooding events.


Surface to 850 mb Layer Precipitable Water by CIRA, 13 March 2013, 2100Z


850 to 700 mb Layer Precipitable Water by CIRA, 13 March 2013, 2100Z


700 to 500 mb Layer Precipitable Water by CIRA, 13 March 2013, 2100Z

A Tale of Three Northern Hemisphere Storms

The fact that yesterday was a leap day would seem to make it stand out, but nature had a say in just how much this day would stand out. From devastating tornadoes in the Midwest to a rapidly intensifying North Atlantic storm that eventually became hurricane force to a strong storm in the Eastern Mediterranean that caused many problems in Turkey, Lebanon, Israel, and Egypt. One of the perks to being a GOES-R “Satellite Champion” at a National Center is that I get to observe many different features on multiple satellite products. I thought these three storms were significant enough to write a blog post on and show just how the new GOES-R era and even current satellite products could be used to enhance our understanding of extratropical cyclones.

Midwest High Impact Storm

GOES-Sounder RGB Airmass product valid at 21z on 02/29/12 with METARs overlaid.

This GOES-Sounder RGB Airmass image shows the dry, stratospheric air (pink/orange shading in yellow circle) that is advecting towards the Potential Vorticity anomaly over WI/IA/MN area. There is an intersection between this stratospheric intrusion and the low to mid-level dry air associated with the dry conveyor belt that originates in the desert Southwest. This is where some higher wind gusts were reported around 21z yesterday.

Zoomed in GOES-Sounder RGB Airmass image with METARs overlaid valid at 21z on 02/29/12.

This zoomed in image highlights the Mid-Mississippi Valley where some of the higher non-thunderstorm wind gusts were recorded. Note the pink shading rotating around the upper-low and co-located with the 30-35 knot wind barbs over the area. Wind gusts were in the 50-60 mph range through MO and IL with Lambert-St. Louis International reporting a wind gust to 58 mph and St. Charles, MO reporting at gust to 64 mph. Most likely, the stratospheric intrusion and ample sunshine allowed excellent mixing and thus, excellent momentum transport over the area at this time.

MODIS RGB Airmass image valid at 1923z on 02/29/12 with the stratospheric intrusion highlighted by the yellow circle.

The above image shows a 1923z MODIS RGB Airmass product with the yellow circle highlighting the stratospheric intrusion. Note the two red streaks that feed into the highlighted area, one from the northwest and another from the southwest. I am not sure if the southern one is truly stratospheric in origin, but the northern stream is most likely associated with stratospheric air and PV “roping” as higher PV values rotate around the parent low.

North Atlantic Hurricane-Force Storm

The SEVIRI RGB Airmass product valid at 19z on 02/29/12 showing a rapidly intensifying North Atlantic extratropical cyclone.

The SEVIRI RGB Airmass product valid at 2345z on 02/29/12 with an OSCAT high resolution wind pass overlaid (courtesy of Jim Kells (OPC)).

The above images show the strong North Atlantic storm that was producing hurricane force winds last evening. Note the very evident stratospheric intrusion (dark pink) advecting towards the PV anomaly center. The first image shows the storm with a central pressure of 976 mb at 19z. The second image shows the storm at 2345z (963 mb) with an OSCAT high resolution wind pass overlaid to show the surface wind direction and strength. The red flags in the southwest quadrant indicate winds over 65 knots and they are co-located with this stratospheric intrusion, further supporting the notion that downward momentum transport is helping to produce this strong wind near the bent-back front in the comma-head region.

Strong Eastern Mediterranean Storm

The SEVIRI RGB Airmass product valid at 1330z with the ASCAT high resolution wind pass overlaid.

The SEVIRI RGB Airmass product valid at 14z on 02/29/12 with yellow arrows identifying the separate advection jets (stratospheric intrusions).

These images show the strong Eastern Mediterranean storm that affected Greece, Turkey, Lebanon, Israel, and Egypt with some very adverse weather. The first image shows the six separate advection jets rotating in towards the strengthening PV anomaly near southern Turkey. This is the most extreme case of stratospheric intrusion I have seen as most cases reveal one to maybe three separate streams. I am unsure of what is going on here to create this phenomena, but it is worth investigating further. The second image gives you some indication of wind speeds in the southeastern Mediterranean with ASCAT finding winds between 30 and 40 knots.

The SEVIRI RGB Dust product valid at 14z on 02/29/12 with blowing dust indicated by the yellow circles.

This final image is the SEVIRI RGB Dust product which highlights the blowing dust occurring from northeast Egypt and Sinai, to parts of Jordan, Syria, and Iraq. Also notice the dust streams over Saudi Arabia.