(What could possibly go wrong in beautiful California?!!)

I have started another small (16.3 sq mi, 72 radar pixels) drainage simulation for Nexrad real-time runoff  to USGS gage comparison in the San Jouquin Valley foothills of the Sierra Nevadas.  I used a nearby Wunder Gage MAP893 to get a first guess calibration at the USGS flow gage 11284400 Big Creek nr Groveland, California (mistake #1).  The Wunder Gage is located outside the basin about 2 miles northwest of the USGS gage at 2500 ft elevation.

USGS flow station info is here.

The Wunderground weather station info is here.

This Wunderground tipping bucket is also being used in my ongoing radar rainfall comparisons here.

The NEXRAD radar site DAX is located 93 miles northwest of the USGS stream gage.

Runoff  Calibration with nearby Wunder Gage Tipping Bucket 12/11/2009 through 12/13/2009:

Lag Time = 8 hours

Basin Area =16.3 mi^2

Loss0 = 0.55 inches (initial was 1.20)

Loss = 0.10 in/hr (initial was 0.20)

%Impervious = 1.0 (initial was 0.5)

Time Step= 5 minutes

(I interpolated the accumulation of  erratic Wundergage time steps using my GetRegression.xls and I used GetRealtime.exe to calibrate this site instead of GetMapArea.exe because of  the 2 days of initial rainfall.)  For help in using GetMapArea and GetNexrad to set the basin boundary and point files go here for an example.

As you can see my Blue Line computed runoff isn’t that hot.  Lots of initial rainfall and initial loss being reset during no rainfall periods but this resetting of the initial loss at the constant loss rate in GetRealtime.exe seems to work. This initial loss condition based on previous rainfall may be an area worth further investigation.  A synthetic unit graph might help out my SCS triangular unit graph method’s recession but if the peak and volume can be met then pretty pictures is just dancing before mirrors (or my rationalization).  A little GetRealtimes.exe routings thrown in might have helped here.  But I could wade 50 cfs. I hold the record below Hoover Dam. Look it up! ;-)

I do not expect this initial calibration to hold, but let’s wait and see.  I got a feeling a lag time of 8 hours on a 6 mile stream is not going to work well when the next July hail storm hits.  (Does it hail in central California?)

The results below come from continuous rainfall runoff modeling in real-time so once the model parameters are set in the GetRealtime.exe setup the real-time results are on their own until/if parameters are changed (initial loss, costant loss, % impervious, recession ratio).  Only the initial loss is self adjusting as noted above.

Update 1/20/10

 Now that some precip events have occurred it looks like this site just may be a waste of time.  As you can see below on the nearby Nexrad vs Tipping Bucket graph, Nexrad is failing miserably in this area of the Sierra foothills with winter type stratus rainfall.

Old Groveland 5-minute tipping bucket comparison:

The daily peak flow computation graph below shows more miserable failure.

I hope to salvage something out of this effort for this basin by somehow adjusting the DAX N0R radar winter rainfall.  I have also found a new Wunderground tipping bucket located near the center of this basin (and I am not making this up).  As soon as I have some rainfall events perhaps a correlation at this new Wunder gauge may be useful.  And I will add real-time runoff computation using the new Wunder gauge itself (was missing 4 hours of peak rainfall on 2/24/10).  The old gauge at Groveland appears to have leaves and dead wasps in it as shown above and is only good for daily values.

If you followed my monolog on Sierrea Snow Fall you will see that the DAX radar is about off by a factor of 2.86 at Groveland.  Based on daily accumulations below at Buck Meadows is off by about a factor of 4.  Be that here nor there let's just use 2.86 and see if salvaging the DAX radar is even possible.

GetRealtime's setup file parameters were set like this to obtain the 2.86*DAX N0R real-time runoff record:

New Tipping Bucket vs N0R daily accumulations at Buck Meadow, CA:

USGS vs Computed Runoff 2/26/10 to 2/28/10:

Basin Area =16.3 mi^2

Lag = 6.0 hrs (was 8.0)

Loss0 = 0.55 inches (no change)

Loss = 0.10 in/hr (no change)

%Impervious = 1.0 (no change)

Time Step= 5 minutes  (no change)

Runoff was computed by GetRealtime.exe for the 5-minute DAX-NCR, 2.86*DAX-NCR, and rainfall for the tipping bucket record for Buck Meadows, CA.  The Buck Meadows record time step is really bizzar.  It uses something like a 23 min, 11 min, 9 min, 12 min, 5 min sequence which  varies a little also.  GetRealtime.exe used the time steps as they are by adjusting the runoff duration on each time step for good or evil and comming up with the same bizzar time steps for the computed runoff.  The basin lag time was reduced from 8 hours to 6 hours based on the faily good runoff computation using the Buck Meadows tipping bucket rainfall record.

Rainfall for the Feb 26-27, 2010  2 day period was:

Buck Meadows Tipping Bucket = 2.79 inches

2.86*DAX NCR = 2.31 inches

DAX-NCR = 0.81 inches

DAX-N0R = 0.77 inches

Is salvaging the DAX radar rainfall for this area by applying a predictive factor possible?  Based on peak runoff, yes it's not too bad.  Based on correcting the whole storm event rainfall 5-minute values... you're living in fantasy land!  Using the DAX radar rainfall as is... you really are California dreamin like this young hydrographer.

Looks to me like we will have to wait for the summer thunderstorms with some height to the clouds for Nexrad to report meaningfull rainfall this far out.

Update 8/6/2010... You are dreaming if you think it rains in California in the summer.  Man, how does this state last out these seasonal doldrums?  Interesting, I was hoping to verify some summer Nexrad data... so far it has been perfect... zip!

Water Year 2011...

Part 2

With New Higher Resolution N0Q Radar

The historical radar N0Q images for Jan 20-23, 2012 were obtained using NOAA's Weather Toolkit program.  These historical images were processed by my GetNexrad and stored in GetAccess HDB database.  Below is the adjustments made to the convective and cool west Z-R to match the observed Wundergage.  There is also now a portable Wundergage somebody installed right next to our Wundergage and is giving excellant confirmation of rainfall amounts.  Got to love that rain gages actually work!

GetRealtime 5-Minute Snowmelt Setup:

SNOWMELT-Unit; -10420,-23420; Melt, SWC; Big Creek, Ca 1.99*N0Qcool Input; 0; 0,1,0.75,0.8,1,1.99,1

Runoff was calibrated with GetMapArea and GetRealtime for Jan 20-26, 2012.  Snowpack computations showed no snowpack for this period so melt equaled precip.  Because of the long recessions observed I will be using 3 trianular unit-graphs for simulating the peak, interflow, and recession and just to show off.

GetRealtime Runoff Setup:

COMPUTE-Unit; -30420; Runoff; Big Cr nr Groveland, Ca N0q; 0;.... and....

 Runoff Coefficients:  3,4.45,0.30,2.0,16.3,2.2,0.15,0.3,4.5,0.3,12;

3 hr lag, 4.45 initial loss, 0.30 constant loss, 2.0 % impervious, 16.3 sqmi drainage area, ...

2.2 for the peak's tail/peak ratio, 0.15 times constant loss as initial loss recovery, 0.3 fraction of runoff as interflow, 4.5*lag as it's recession time, 0.3 fraction of runoff in recession flow, 12 times lag for this tail recession time.  For a Youtube how-to video on this multi-triangular unit-graph method go here.

What I found interesting about the above calibration was adjusting the 2nd peaks initial loss by playing with the 0.15 times constant loss as initial loss recovery factor from the 1st peak.  That factor was what produced the 2nd peak after setting all the others to match the 1st peak.  It's really a big deal and probably needs some exponential time thing added to my simple linear reverse abstraction on zero precip.  This basin was  bone dry comming into this event as you can see.

Ongoing Study

Based on the above Januarly radar precip and runoff calibration only, the ongoing snowmelt and pack and runoff follows and will probably begin in earnest next winter when precip returns to California:

Realtime N0Q radar precip:

Daily Temperature:

Daily Peak Flows:

April 11, 2012:

April 11-13 showed snow accumulation below a Temperature factor +3.0 F but only reduced the peak runoff.  The radar unit rainfall values that produced the observed peak runoff was way under the observed Wundergage rainfall and so there was not much sense in calibrating a royal radar fubar.  The Wudergage and radar hourly values did not match up at all but the daily totals were still pretty good as shown above.

April 25, 2012:

With a 4.5" initial loss, the percent impervious is the only factor in play here except for the rainfall was a SNAFU again.  Temperatures were above 40F and N0Q Convective did no better.