2021/10/25 and 2021/11/01 HARP Work Session

[rburghol]

• Visualizations / QA analytics
  • [x] a histogram of precip days
  • [x] Days over time
  • [x] Is precip intensity and frequency changing over time?
    • Create bar graph of number of precip days and precip values
  • [ ] Geospatial trends
    • Divide state into sections and plot a few land segments in each section
  • [x] Seasonal plots to determine seasonal trends
    • Bar graph of number of precip days
    • Total Precipitation
• WDM
  • [x] CLC: cloud cover, Gopal thinks the fortran code creates CLC from solar radiation (who can verify that in the source code? someone who hasn't touched the code this could be a good exercise to team with someone who has) - It does appear to be calculated in wdm_insert_ALL.f
  • [ ] Document CLC algorithm
  • [ ] Generating some full WDMs. - progress tracked here (currently just missing CLC): https://github.com/HARPgroup/HARParchive/issues/129#issuecomment-951099709
  • [ ] Testing full WDMs
  • [ ] Note: Gopal pointed out that the calibration was performed with a different met source, the XYZ data set, and that since the models infiltration etc. are adjusted to fit stream gage data, that there will be uncertainty as to what distortions this new met set may introduce.
• Documentation of Factors that Determine Base Flow in Virginia:
  • The long term study of the topic of rainfall, baseflows and extreme drought
  • Do multi-year droughts exacerbate base flow deficits?
    • Are there observable 1-year, 2-year, 3-year base flow recharge effects?
  • How well do the models replicate the 1-year, 2-year, 3-year base flow recharge patterns?
  • The HARP annual projects have focused on different facets of this during many project years
  • Essentially, what we have been doing is devising an analytical and modeling approach to answering these questions using readily available data.
• Project planning (Note: analysts can work on different tasks concurrently):
  • [ ] Wrap up documentation on these approaches from QA/QC perspective
  • [ ] Then document all the additional questions we could seek to answer
    • [ ] Documentation of precip trend and data observations for decision making
  • [ ] Continue pushing for the WDMs to get into the model (b/c this is the primary focus)
  • [ ] Schedule/Timeline planning
    • [ ] ~6weeks of work left before exam/holiday time
    • [ ] Rest of the semester: holidays, exams, breaks.

[kylewlowe]

CLC algorithm update:

• Value is calculated inside wdm_insert_ALL.f function using precipitation and radiation data
  • Don't understand fortran well enough to determine exactly how it is doing this
• Currently running into error on line 857
  • Can't open .RNMAX csv file
  • It seems like RNMAX corresponds to highest radiation
  • Line 43: real RNMAX(12,31,24), PRECIP(NDAYSMAX*24)
  • @rburghol Do you know what exactly the line and the real function does? Also, do you know what would be in the .RNMAX timeseries csv so we can begin to put one together

[rburghol]

Hmm. Well, RNMAX could be the max radiation I suppose. But to give you some hints on this, the lines that you have identified are just declarations, in other words, in Fortran (and many other languages) you have to declare what kind of variable something is before you can use it -- unlike in R which usually infers what a variable is by the way that you declare it.

• Line 43 declares 2 variables as "real", that is they are "real" numeric types not complex or imaginary.
• the variables it declares are RNMAX and PRECIP (if you wanted to declare more real variables you could just add them on that line with commas separating them)
• RNMAX is a 3-dimensional array that is 12x31x24, and I am assuming this means it might hold a single value for each month, day and hour (see below for my idea on why that is needed)
• PRECIP will hold one value for each hour of the entire simulation.
• RNMAX is being calculated in the block of code between lines 844-1039 and does indeed use PRECIP (since does something with sunny days),note that the presence of precip here does not indicate that it is using precip to calculate RNMAX, though it may be, but again, this is just a declaration and they may have combined these for convenience sake.

So, next steps would be to:

1. post up the command that you are using that gets the error (just for grins)
2. Search the code for occurrence of RNMAX, since it seems that we might be on the hook for supplying that! If so, we want to know what the format it expects is.
3. See if you can find (via searches), any code that works with the CLC variable, for that matter, have you found a CLC variable in there?
4. I think you can omit step 3 above, since I indicated the block of code that calculates that. My read on this is that line 954 is where CLC is being calculated. Because solar radiation will vary by latitude day of the year, and hour of the day, somewhere they are storing the maximum solar radiation observed in the entire dataset for each hour of each day of the year in the variable RNMAX, then calculating the cloud cover at any time based on the percentage of the maximum ever observed for that location. Thus, I suppose that we must do that too -- or search the code Gopal has already given us for this thing. I think the process would be
   1. we can simply load our solar radiation file for each land segment in R
   2. use sqldf to do this: select month, day, hour, max(solar_rad) as RNMAX from solar_rad_dataframe GROUP BY month, day, hour
   3. save that in our data directory that the fortran file is trying to load (described in line 854)

[alexwlowe]

^Here is the command that gets the error:

 ./wdm_insert_ALL
A51175 nldas2 harp2021 1984010100-1984123123 1984 1984 1 1 1 1 1 1 1

The error says:

STOP ERROR opening .RNMax file

Which means it occurs on line 857

[katealbi11]

Graph for Multiple Land Segments

[katealbi11]

Graphs with Summer and Winter Data Only

[alexwlowe]

11/2/2021 Update:

Kyle finished the RNMax script yesterday and pushed the script to github, and was able to get all the files on the terminal.

Upon attempting to create full WDMs, I received an couple of errors:

The first error stated "error opening RNMax file" and I was able to past this by switching the directories for which the RNMax files where stored.

The second error stated "error reading RNMax file". I received this error writhe before the end of the day yesterday and did not get a chance to investigate why it occurred, but will take a look at it when the new server is up and running. I'm thinking it could be some differences in formatting between our RNMax csvs and the csvs the wdm populating function expects.

Once we have Gopal's LongTermAvgRNMax.cpp function compiled and running we may just use that instead.

[katealbi11]

La Nina Updates:

• Generally wetter winters in the Northern US and warmer, drier conditions in the south
• Temperature effects of 2018's La Nina Winter (January-March)
  • Precipitation effects of 2018's La Nina Winter (January-March)

[alexwlowe]

TO DO AFTER 11/3:

• Run some QA on solar radiation and cloud cover (wdm function does not like cloud cover values - values outside of acceptable range)
• Double check our RNMax files with new RNMax