Missing data error when no data is missing

[jcarlston]

Hello all,

I'm trying to run a Raven model with MultiData files as forcing inputs (precipitation, daily maximum temperature, daily minimum temperature). When I try to run the model, I get:

Exiting Gracefully: CGauge::Initialize: Raven cannot have blank data in temperature time series
Error Type: Bad input data

So naturally, I spent a long time trying to figure out where the missing data was. I created new files to run one basin at a time and varied the run time until I discovered the offending row of data. It turns out there is no data missing at all! Furthermore, if I change one of the values by just 0.001, it works! This is very reproducible - 100% of the time I run the model with the original data it fails, and 100% if I change the "offending" row, it works. The row causing the failure is (precipitation in mm, daily max temp in deg Celsius, and daily min temp in deg Celsius):

0.083 0.813 -3.282

which are physically reasonable values and which are formatted in the same manner as other lines in the input file.

Has anybody run into an issue like this? If not, is there at least a way to know what line(s) of input the model doesn't like? I'm happy to change the inputs by 0.001 where necessary, as such a change is well within the accuracy of my data anyway, but it's taking me an excessive amount of time to figure out where the problem(s) are located by trial and error. Thanks in advance!

[rchlumsk]

Hi jcarlston

This is an odd error, my first thought is if there is a difference in spaces/tabs between the characters that was changed to make the file work. I am unable to recreate the error without more information, are you able to send me your model files or at least a subset of them so I can take a look at the input files and model run?

Thanks,
Rob
rchlumsk@uwaterloo.ca

[Niksulfibe]

Hi Keystone,

Thanks for the file. I get the same error. Ill post back here when a fix is available.

Best regards,
Aaron.

[SeanWatt]

Hi all, I'm getting this error as well, and no apparent data missing.
I have 4 climate gauge files, they were working fine. 10 years of data, originally 4 columns, Rainfall, Snowfall, DailyMaxT, DailyMinT - and the model was running ok.
I noticed some error warnings, that a few dates had Tmin>Tmax, so I went back to my raw data, and saw a few missing data points in the series that had been filled with zeros. So, I replaced those with -1.2345, as noted in the manual for missing data. And I realized that Raven likes to have DailyMeanT too, so I added that column to my data, since I had it available.
Those are the changes I made, and now I get the error. I don't believe any data is actually missing. Anything else that could have gone wrong?
I wondered whether my columns are overlapping in the rvt file (don't think it should matter as long as there is a space between columns), so I added more spaces between, but no difference.
I'm created the text file, using the EC downloaded csv files, and then @concatenate() to joins the points into rows, to cut and paste into the TXT file. That's how I created the first files that worked, so should be no different there.
Thanks, Sean

[SeanWatt]

Removed the -1.2345 values, replaced them with 0.0. And it works again. So, there is no option for blank data in the Temperature Forcing files? Am I misreading the notes in the manual? It doesn't seem to note that it won't work for time series, but that appears to be what has happened. Next option, since I have multiple gauge stations, that mostly match for rainfall, I will try to replace the missing values with estimates from the other stations. Not the best option, but maybe the next best option?