07/02/2012  Important news for all GetRealtime users!

I had an appalling realization about Wunderground rain gage reporting.  I had assumed from prior year studies that the Wundergage rainfall parameter 'HourlyPrecipIn' for a 5-minute value was the incremental rainfall over the 5 minute period and reported as inches/hour. To see what Blue Ridge Mtn folk think this means skip to the bottom of the page here.

Summary Table Beech Mountain Area Daily Rainfall

% Difference is the error in total rainfall, a measure of bias.

MAE % is the mean of the individual daily or hourly unsigned differences, a measure of error variablility, and has been limited to values >0.1 inches.

Live shot from elevation 3030 feet:

This new comparison of NEXRAD radar rainfall is with the new Ridge 2 Testbed N0Q Level 3 product with high dBz resolution of 0.5 dbz versus the old 5 dbz.  Rainfall is computed from the 5-minute radar images using the standard WRD-88D Convective Z-R relationship and East Cool in winter using GetRealtime.exe in real-time.  For additional information on NEXRAD radar rainfall using GetRealtime.exe go here.  For info on the great new N0Q radar reflectivity product go here.

GetRealtime dBz to Rainfall Conversion

Wunderground Gage Loactions:

#1 Beech Mtn, NC Wunder Gage KNCBOONE12:

Wunderground Gage: KNCBOONE12, Lat 36.227, Lon -81.720, gage elev=3400 ft.

MRX Pixel 664,494, radar distance=94.1 miles, mid scan elev=10,890 ft

GSP Pixel 549,367, radar distance=97.0 miles, mid scan elev=10,890 ft

FCX Pixel 361,577, radar distance=97.7 miles, mid scan elev=12,870 ft

Radar image pixel size 0.57 x 0.70 miles, radar scan bin size 0.62 range x 1.5 width miles

GetRealtime_setup.txt: KNCBOONE12; 10811; Rainfall; Beech Mtn. #1, NC

GetAccess rsite table: 811; 10811; Beech Mtn. #1, NC ; 10; rainfall; inches; inches; dailyrainin; KNCBOONE12; 8; NC;-3

GetRealtime_setup.txt: NEXRAD-MRX; -10811; Rainfall; Beech Mtn #1, NC; 0

GetAccess rsite table: 811; -10811; Beech Mtn. #1, NC ; 10; rainfall; inches; inches; N0Q-Ridge2; NEXRAD-MRX; 8; NC; 0

(note: Gage daily values computed from unit value rates based on a Pacific time -3 hr shift to match radar converted to my local Pacific time.)  First gage KNCELKPA2 died and  replaced 6/23/2012.

#2 Beech Mtn, NC Wunder Gage KNCBEECH5:

Wunderground Gage: KNCBEECH5, Lat 36.202, Lon -81.886,  gage elev=4,849 ft

MRX Pixel  647,497, radar distance=84.3 miles, mid scan elev=9,450 ft

GSP Pixel 533,370, radar distance=92.8 miles, mid scan elev=10,320 ft

FCX Pixel 345,579, radar distance=105.9 miles, mid scan elev=14,220 ft

GetRealtime_setup.txt: KNCBEECH5; 10812; Rainfall; Beech Mtn. #2, NC

GetAccess rsite table: 812; 10812; Beech Mtn. #2, NC ; 10; rainfall; inches; inches; dailyrainin; KNCBEECH5; 8; NC;-3

GetRealtime_setup.txt: NEXRAD-MRX; -10812; Rainfall; Beech Mtn #2, NC; 0

GetAccess rsite table: 812; -10812; Beech Mtn. #2, NC ; 10; rainfall; inches; inches; N0Q-Ridge2; NEXRAD-MRX; 8; NC; 0

(note: Gage daily values computed from unit value rates based on a Pacific time -3 hr shift to match radar converted to my local Pacific time.)

#3 Beech Mtn, NC Wunder Gage MD8069:

Wunderground Gage: MD8069, Lat 36.225, Lon -81.897,  gage elev=3,998 ft.

MRX Pixel  646,494, radar distance=83.8 miles, mid scan elev=9,380 ft

GSP Pixel 532,367, radar distance=94.7 miles, mid scan elev=10,610 ft

FCX Pixel 344,577, radar distance=106.0 miles, mid scan elev=14,170 ft

GetRealtime_setup.txt: MD8069; 10813; Rainfall; Beech Mtn. #3, NC

GetAccess rsite table: 813; 10813; Beech Mtn. #3, NC ; 10; rainfall; inches; inches; dailyrainin; MD8069; 8; NC;-3

GetRealtime_setup.txt: NEXRAD-MRX; -10813; Rainfall; Beech Mtn #3, NC; 0

GetAccess rsite table: 813; -10813; Beech Mtn. #3, NC ; 10; rainfall; inches; inches; N0Q-Ridge2; NEXRAD-MRX; 8; NC; 0

(note: Gage daily values computed from unit value rates based on a Pacific time -3 hr shift to match radar converted to my local Pacific time.)

#4 Beech Mtn, NC Wunder Gage KNCBLOWI4:

Wunderground Gage: KNCBLOWI4, Lat 36.121, Lon -81.758, gage elev=3,800  ft.

MRX Pixel  660,505, radar distance=91.9 miles, mid scan elev=10,550 ft

GSP Pixel 546,378, radar distance=89.2 miles, mid scan elev=9,790 ft

FCX Pixel 357,587, radar distance=103.9 miles, mid scan elev=13,830 ft

GetRealtime_setup.txt: KNCBLOWI4; 10814; Rainfall; Beech Mtn. #4, NC

GetAccess rsite table: 814; 10814; Beech Mtn. #4, NC ; 10; rainfall; inches; inches; dailyrainin; KNCBLOWI4; 8; NC;-3

GetRealtime_setup.txt: NEXRAD-MRX; -10814; Rainfall; Beech Mtn #4, NC; 0

GetAccess rsite table: 814; -10814; Beech Mtn. #4, NC ; 10; rainfall; inches; inches; N0Q-Ridge2; NEXRAD-MRX; 8; NC; 0

(note: Gage daily values computed from unit value rates based on a Pacific time -3 hr shift to match radar converted to my local Pacific time.)  First gage KNCBEECH3  replaced 7/2/2012.

#5 Beech Mtn, NC Wunder Gage KNCBANNE7:

Wunderground Gage: KNCBANNE7, Lat 36.174, Lon -81.913, gage elev=3,458 ft.

MRX Pixel 645,499, radar distance=82.6 miles, mid scan elev=9,290 ft

GSP Pixel 530,372, radar distance=91.1 miles, mid scan elev=10,060 ft

FCX Pixel 342,582, radar distance=108.9 miles, mid scan elev=14,650 ft

GetRealtime_setup.txt: KNCBANNE7; 10815; Rainfall; Beech Mtn. #5, NC

GetAccess rsite table: 815; 10815; Beech Mtn. #5, NC ; 10; rainfall; inches; inches; dailyrainin; KNCBANNE7; 8; NC;-3

GetRealtime_setup.txt: NEXRAD-MRX; -10815; Rainfall; Beech Mtn #5, NC; 0

GetAccess rsite table: 815; -10815; Beech Mtn. #5, NC ; 10; rainfall; inches; inches; N0Q-Ridge2; NEXRAD-MRX; 8; NC; 0

(note: Gage daily values computed from unit value rates based on a Pacific time -3 hr shift to match radar converted to my local Pacific time.)

#6 Beech Mtn, NC Wunder Gage KNCVILAS2:

Wunderground Gage: KNCVILAS2, Lat 36.246, Lon -81.765, gage elev=2,850 ft.

MRX Pixel 659,492, radar distance=91.3 miles, mid scan elev=10,480 ft

GSP Pixel 545,365, radar distance=97.8 miles, mid scan elev=11,090 ft

FCX Pixel 357,575, radar distance=98.8 miles, mid scan elev=13,040 ft

GetRealtime_setup.txt: KNCVILAS2; 10816; Rainfall; Beech Mtn. #6, NC

GetAccess rsite table: 816; 10816; Beech Mtn. #6, NC ; 10; rainfall; inches; inches; dailyrainin; KNCVILAS2; 8; NC;-3

GetRealtime_setup.txt: NEXRAD-MRX; -10816; Rainfall; Beech Mtn #6, NC; 0

GetAccess rsite table: 816; -10816; Beech Mtn. #6, NC ; 10; rainfall; inches; inches; N0Q-Ridge2; NEXRAD-MRX; 8; NC; 0

(note: Gage daily values computed from unit value rates based on a Pacific time -3 hr shift to match radar converted to my local Pacific time.)

Just Wunder Gage Accumulations:

When it comes to recording precipitation, all Wunderground stations are not alike.  This comparison is made to check for any large deviations over time.  It's not that easy calibrating and maintaining a good precip record so a huge thank you to those of you that really try.

USGS 03479000 WATAUGA RIVER NEAR SUGAR GROVE, NC (90.8 sqmi) Runoff Computations:

The Watauga River upper most gaged catchment was planimetered for basin area and basin radar image boundary pixels using my free Windows program GetMapArea.  For an example using GetMapArea go here or even better right below is anothr example.

USGS Gage:03479000, basin centroid Lat 36.209, Lon -81.757, gage datum elev=2,607.84 ft.

MRX Centroid Pixel 645,499, radar distance=91.2 miles, mid scan elev=10,540 ft, scan width=1.5 miles, radar antenna elev=1,434 ft.

GSP Centroid Pixel 545, 369, radar distance=95.1 miles, mid scan elev=10,670 ft, scan width=1.5 miles, radar antenna elev=1,050 ft.

FCX Centroid Pixel 357, 578, radar distance=99.4 miles, mid scan elev=13,230 ft, scan width=1.6 miles, radar antenna elev=2,963 ft.

Radar image pixel size 0.57 x 0.70 miles

GetRealtime_setup.txt: NEXRAD-MRX; -10817; Rainfall; Watauga R nr Sugar Grove, NC; 0

GetAccess rsite table:  817; -10817; Watauga R nr Sugar Grove, NC ; 10; rainfall; inches; inches; N0Q-Ridge2; NEXRAD-MRX; 8; NC; 0

Watauga basin area with radar sites MRX, GSP, and FCX as shown on GetNexrad radar viewer:

5-minute Snowpack and Melt Computations:

The temperature, humidity, windspeed at WunderGage #6 (or #1 if missing) was used to compute solar radiation and long wave radiation.  With these 5 parameters and adjusted rainfall, the snowpack and melt was computed at a 5-minute time step in real-time (or the next day).  The 3 radars are using the Cool East Z-R for converting radar reflectivity to rainfall/snowfall.  The Cool East Z-R provides precip down to 10dbz for adjusting (convective only 20dbz).  The graphs below shows what factors, if any, were used to adjust the 5 weather inputs and precip and albedo for the snowmelt computation.

It appears only Gage #4 of the 3 Wundergages in the basin is equipped with a heated tipping bucket and so radar ajustment will be based only on Gage #4 this winter.  Actual reported snow values of some sort (depth or new or swe???) can be found here for comparison but the reported values so far are not even in the ballpark with the Wunderground gages or with visual comparisons at Wunderground webcams in the area.  Snow will show up as white stuff on webcams and when reported values are off the chart and no white stuff, well I am not dumb enough to use those reported values... yet.

For more info on snowmelt computations see my past Sierra study, San Juaquin study, and Help.

USGS Gage vs Radar Rainfall (and melt) Runoff:

The radar unit values used have a 5-minute time step for runoff computations.  The Wudergages 3-gage average time step is 1-hour.

SCS Curve Number CN method used with my CNwet and CNdry.

GetRealtime_setup.txt: NEXRAD-GSP; -30827; Runoff; Watauga R nr Sugar Grove, NC;

50; 6.5,65,0.2,3,90.8,1.0,0.015,0.3,6,0.4,12,85,64,0.8; P1

GetAccess rsite table:  827; -30827; Watauga R nr Sugar Gr; 30;runoff;cfs,cubic feet per second; -27811; NEXRAD-GSP; 8; NC; 0

Where -27811 is tall ET computed from Beech Mtn Wundergage #1 above.

Quick CN method calibration tip: Always start with a yellow 0.2 constant loss and a yellow 0.8 GW loss.  Increase the red 65 CN to increase initial peak flow and then increase red 0.015 reset factor to reduce subsequent peak flows. Adjust the yellow 0.8 GW loss when finished.  Sometimes the recession slope  red 0.015 needs to be ballparked first for several dry days.  I have ignored the percent impervious so see more tips below.

Note that Nexrad Radars do not report the same rainfall.  Each radar has been adjusted using GetRealtimes hourly ratio method where the average of Gages 1, 4, 6 are ratioed against the average of the radar at each gage point.  The hourly ratio is then applied to basin average 5-minute radar rainfall values.  If a gage is missing, the average for the other 2 or even 1 will still be computed and ratioed against the average of the 3 radar points.  Probaly 1 good gage would work better than my 3 iffy gages, but I wanted to try this averaging of multible gages.

Calibration Period Jun 20 to  Jul 30, 2012:

USING THE ABOVE CALIBRATION AND JUNE 20, 2012 START DATE

(tip, for speed don't save unit values to the database until you got it all looking good with the hourly's.)

Summer hourly comparison with the least accurate radar MRX ajdusted:

Winter hourly comparison with the least accurate radar MRX ajdusted N0Q EastCool Z-R:

Hypothedical Basin Wunderground 6 Gage Average vs Radar Basin Area Average

Because radar rainfall should be better for areas rather than points, lets see how our 6 Wunderground tipping buckets stack up against our 3 all seeing radars.  GetMapArea.exe was used to determine the 3 radar's  basin 'Boundary File' used by GetRealtime.exe.  GetNexrad.exe was used to read the 3 boundary files and create the 3 'Point File's for the MRX, GSP, and FCX radar's N0Q images.

I will set you an example:

You may want to first watch these how to videos for conceptionaliztions:

NEXRAD Radar image basin area averaged rainfall to digitize a radar area.

Real-time Record of Nexrad Radar Area Averaged Rainfall to add the area to GetRealtime.
For a full list of more helpful videos click on Site Map at the very top.

Using the Gif image below, I loaded it into GetMapArea.exe.  To determine the Hypo Basin's sqmi area, I need to set the milage scale.  I selected points #3 and #4 and then loaded the radar MRX to find the distance between these gages as:

#3 Wunderground Gage: MD8069, Lat 36.225, Lon -81.897

#4 Wunderground Gage: KNCBLOWI4, Lat 36.121, Lon -81.758

Using GetNexrad I selected Boundary #3 and zoomed in.  I clicked menu 'Lat/Long' and entered gage #4's Lat and Long, and clicked 'Set Point'.  I clicked menu 'Miles', clicked on the set point and then the center of Gage #3 and got the distance as 10.7 miles.  Checking this value with Google Earth's distance I got 10.6 miles but was more tedious but highly preferred, so lets use 10.6 because N0Q pixels are 0.6x0.7 miles.

Using GetMapArea 2.7.0 and setting the scale form Gages #3 to #4 as 10.6 miles, I planimetered a 58.2 sqmi area for the gage centers in the Gif below. 

Next is to create the MRX radar pixel Boundary File. Above I have X,Y pixel values for each Gage and Radar.  For MRX  #3 pixel is 646,494 and MRX pixel for #4 is  660,505.  Using these pixels we can reset the GetMapArea image scaling using General Pixels.  The X1,Y1,X2,Y2 text to enter is 646,494,660,505.  We are now set to start the area.  Right click select Start Area and planimeter the area again.  Click End Area, Right click the coordinate text box and save as MRXHypoBoundary.txt.

Now we need to add a rainfall site to the GetAcces database table rsite.  I will use 819 as the site_id and so the rainfall datatype_site_id will be -10819.  I use a negative to remind me I calculated it and is not a real gage.  Having our datatype_site_id we have to rename our Boundary file MRXHypoBoundary.txt to our standard format as NexradBoundaryMRX-10819Q.txt.  The Q means the new Ridge2 N0Q product image is being used.... (and edit the the title line in the file.)

To create the Nexrad Point file, start up GetNexrad.exe, select Create Point File, and select Find our new file NexradBoundaryMRX-10819Q.txt in the GetMapArea folder.  Select No Conversion and then Create File.  You can zoom in on the red area to see our filled pixeled area.

Copy the 2 files NexradBoundaryMRX-10819Q.txt and NexradPointMRX-10819Q.txt from the GetMapArea folder to your GetRealtime.exe folder (and also to your GetNexrad images folder if you are using some other folder for GetNexrad images?).

All that is left is to add this Boundary to our GetRealtime_setup.txt file using Notepad as:

NEXRAD-MRX; -10819; Rainfall; MRX Hypo Basin, NC; 0

Test out a GetRealtime.exe radar retrieval for this site and if it works then repeat the above planimetering for the GSP and FCX radars with these scale points:

GSP 3 to 4 scale points: 532,367,546,378 >>> NexradBoundaryGSP-10829Q.txt

FCX 3 to 4 scale points: 344,577,357,587 >>> NexradBoundaryFCX-10839Q.txt

... and we are set to make some comparisons.

To compute the 6 gage Wunderground average add a 819 site to the GetAccess database table rsite with the 6 datatype_site_ids and you could add this text to your GetRealtime_setup.txt file as (I use a seperate folder for GetRealtime.exe for weather and flow data because I don't need it downloaded every 30 minutes like the radar):

COMPUTE-Hour; 10819; Rainfall; Hypo Basin Avg of 6 Rain Gages; 0; 0; (P1+P2+P3+P4+P5+P6)/N

where using N instead of 6 returns average of non-missing P's.

You should use COMPUTE-Hour because Wunderground gages use every kind of unit time-step imanginable.

  You could do the inverse distance to basin centroid squared weighting in my simple mean if it makes you feel better and each adjust weighting by 6/N for missing gages.

The GetAccess database table rsite would have parameter_code:

 10811,10812,10813,10814,10815,10816

and the station_id is COMPUTE

Wunderground Parameters 'HourlyPrecipIn'  versus 'dailyrainin'

I had an appalling realization about Wunderground rain gage reporting.  I had assumed from prior year studies that the Wundergage rainfall parameter 'HourlyPrecipIn' for a 5-minute value was the incremental rainfall over the 5 minute period and reported as inches/hour.  For a 0.01" 5-min increment the hourly rate would be 0.12" (0.01*60/5).  This would be consistant with meteorlogical convention.  I had noticed a few past cases of what I thought was a leafed clogged tipping bucket that delayed the unit values sometimes over several hours.

Noticing lately that Wunderground was now reporting a new parameter 'dailyrainin' along side, I decided to compare the new 5-minute increments to the 5-minute reported rates.  To my schock and horror I found most of these tipping buckets are glogged or something.  Some gages seem to report the 5-minute rates as the average rate over the past hour and some are just what ever fits their fancy.

The take away here is that if in doubt, use the parameter 'dailyrainin' accumulation values because rainfall rates are a lost art in the backwoods of these here Blue Ridge Mtns. You be the judge:

Wundergage #1:

Gage #1 looks like a traditional hourly rate with some slop, just one really bad rate at the start.

Wundergage #2:

Gage #2 looks like a lagging average rate for the preceeding hour that eliminates the variability.

Wundergage #3:

Gage #3 looks like a lagging average rate for the preceeding hour that eliminates the variability.

Wundergage #4:

Gage #4 looks like a lagging average rate for the preceeding hour that eliminates the variability.

Wundergage #5:

Gage #5 looks like a lagging average rate for the preceeding hour that died.

Wundergage #6:

Gage #6 looks like a lagging average rate for the preceeding hour that eliminates the variability.

After further review of all the Wunderground gages I've recently used in other studies, the Beech Mtn #1 gage above is the only one I can find correctly reporting the hourly rainfall rate 'HourlyPrecipIn' so my apologies to the Blue Ridge Mtn folks.  Apparently the new standard rainfall rate 'HourlyPrecipIn' reported each 5-minutes is total rainfall for the past 60 minutes which who the heck cares!  That's a crying shame.  To get the 5-minute increments, 'dailyrainin' must be differenced.

For my other current study at Ft Lauderdale, FL, I have had to redownload and replaced all 4 gage's unit values using the new 'dailyrainin' parameter.  My GetRealtime.exe 07/02/2012 has been updated to make this an easy task.  Live and learn. 

For more radar adjustment stategies and comparison of N0Q and Level 2 go here or for a list of my studies go here.