Open julieshortridge opened 1 year ago
Original case study results ========================= ============================ Rain + snow : 82.32 Interception : 2.58 Runon : 0.00 Runoff : 1.68 Irrigation : 0.00 Transpiration : 24.44 Bottom flux : -34.85 Soil evaporation : 16.00
Crack flux : 0.00 ========================= ============================ Sum : 47.47 Sum : 44.70
CASE STUDY with just one year simulated Water balance components (cm)
In Out
========================= ============================
Rain + snow : 64.49 Interception : 0.66
Runon : 0.00 Runoff : 0.00
Irrigation : 0.00 Transpiration : 21.79
Bottom flux : -27.39 Soil evaporation : 16.54
Crack flux : 0.00
========================= ============================
Sum : 37.10 Sum : 38.99
Case study using 2021 tidewater (observed) met data. Also, I had to change the crop planting and harvest dates in the .swp file because now the dates are different. For the same reason (different dates), I changed the bottom flux to zero as I did for the previous simulation in the simple crop growth model. Groundwater levels (for multiple dates within a year) were provided in the original case study file for the bottom flux component which is not available in our study. However, the results were fine Water balance components (cm)
In Out
========================= ============================
Rain + snow : 57.41 Interception : 0.17
Runon : 0.00 Runoff : 10.57
Irrigation : 0.00 Transpiration : 22.08
Bottom flux : 0.00 Soil evaporation : 23.15
Crack flux : 0.00
========================= ============================
Sum : 57.41 Sum : 55.98
Case study with tidewater met data and heat flow excluded from simulation. This needs to be checked before updating soil properties with tidewater data because heat flow simulation using the numerical approach (currently being used) requires us to input additional soil properties for different soil layers. By changing the method of oxygen stress calculation in the .crp file, heat flow simulation could be excluded from the model (these two processes are interdependent)
In Out
========================= ============================
Rain + snow : 57.41 Interception : 0.21
Runon : 0.00 Runoff : 9.04
Irrigation : 0.00 Transpiration : 22.91
Bottom flux : 0.00 Soil evaporation : 24.86
Crack flux : 0.00
========================= ============================
Sum : 57.41 Sum : 57.01
Case study with tidewater met and soil data. Also, I changed the GWL from the existing -68 cm to -500 cm as we used before. A fair amount of transpiration, but the water balance seems to get worse because of less water going out of the system. This was likely because of the runoff component as we can see in the above water balance. Makes sense as those can be related to different soil hydraulic parameters. Water balance components (cm)
In Out
========================= ============================
Rain + snow : 57.41 Interception : 0.12
Runon : 0.00 Runoff : 0.00
Irrigation : 0.00 Transpiration : 18.83
Bottom flux : 0.00 Soil evaporation : 19.35
Crack flux : 0.00
========================= ============================
Sum : 57.41 Sum : 38.30
Changed method of ET estimation from the use of crop height to crop factor. Result didn't change much but transpiration reduced from 18 to 16 cm. Major difference is in May period- there is no rise in ET values as in the above graph Water balance components (cm)
In Out
========================= ============================
Rain + snow : 57.41 Interception : 0.10
Runon : 0.00 Runoff : 0.00
Irrigation : 0.00 Transpiration : 16.44
Bottom flux : 0.00 Soil evaporation : 19.63
Crack flux : 0.00
========================= ============================
Sum : 57.41 Sum : 36.17
Using historic met data on the latest updated model (Just simulated year 2021). Starting to see a reduction in transpiration, but not insignificant as we saw in our meeting (was around 0.15 cm). Forage yield reduced substantially, but is not insignificant. The rainfall amount here is twice the amount in our observed data for the same 2021 year (this seems too high for 10 months period). Please ignore the red dots in all graphs (observed values in CN case study-I didn't delete these from the folder so that I won't have an issue running the R code directly from the windows command script). Water balance components (cm)
In Out
========================= ============================
Rain + snow : 121.55 Interception : 0.00
Runon : 0.00 Runoff : 35.20
Irrigation : 0.00 Transpiration : 8.49
Bottom flux : 0.00 Soil evaporation : 45.37
Crack flux : 0.00
========================= ============================
Sum : 121.55 Sum : 89.06
Great, thanks Suman for pulling all these together so quickly. Clearly something seems to be going on when you introduce the historic met data (by this I'm assuming you mean the 20-year simulated data that I sent you, please correct me if I'm wrong about that). Can you share/send me (or point me towards the files/locations on the shared drive) the met files for you used for the last two simulations that you posted? I'll see if there's an issue with the method I'm using to generate the synthetic weather data.
Also can you edit your posts above (click on the three dots at the upper right hand side of the comment) to provide, for each model run except the initial multi-year case study, a plot of potential versus actual ET as well as the rain, ET, and 10cm soil water content (see the last two plots on the "Crop Growth Module" issue for an example of what I'd like to see).
While I'm looking into the met data, I'd suggest looking into the documentation and see if you can figure out what needs to change to plot maize grain yields. Thanks!
@julieshortridge The historic met data that I used for the last two simulations (observed and synthetic, respectively) is located here: (Metdata.csv and Synthetic_2021) G:\Shared drives\GCC Agroeconomic Modeling\SWAP\SumanWorking\Tidewater_metdata
I am running the model simulations from the following location (the input weather data for all simulations can also be obtained in their respective InputData\Weather folder) G:\Shared drives\GCC Agroeconomic Modeling\SWAP\SumanWorking\Model versions\08_CN_CaseStudy_multiyear\InputData\Weather
Just realized that those files are in plain text when I exported from R. I have quickly converted them to csv as follows. I hope this helps. 1) For the second last simulation, the observed tidewater met data for 2021 is Metdata.csv
2) For the last simulation, the synthetic tidewater met data for 2021 is Synthetic_2021.csv
When I was trying to generate CSV outputs for soil moisture and ET for desired depths, I realized the older version (which we are using currently, 4.0.1) does not support specifying and exporting such CSV outputs directly. In the past few days, I was able to generate those CSV outputs because somehow I was using a newer version of swap.exe (4.2.0) within the old version SWAP directory (which was possibly creating some issues) to run the model. I just had to modify my R code to import output text files and extract specific depths (among different compartments) and make the newer plots for WC, ET, and rainfall.
Thanks Lal. Can you also share your converted version of the 20-year synthetic data? In the 2021 synthetic data the rainfall is higher than the observed, and I want to see if that's the case for other years too.
I'm still looking at this data but one thing I noticed is that at least in the csv files you shared, there is an additional column at the far left of the synthetic data (eg starting with 6576). Maybe that just happened when it exported to csv but you might want to check and see if that's causing some kind of issue. We also need to double check our units for wind speed (it's easy to get miles per hour and m/s mixed up) but I doubt that would have a large enough impact to be noticable in your results.
The additional column is just the row numbers in R that showed up while exporting into CSV. I checked the plain text file used by the SWAP and it seems fine. Here's the 20 years of the synthetic data file. Sure, will check for any potential issues like that. Thanks! Synthetic_allyears.csv
Bottom boundary condition changed from bottom flux equals zero to free drainage of the soil profile. While looking into the groundwater level over time, there were positive values (mostly during May). A better water balance result and reasonable plots were obtained using the same synthetic met data for 2021. Forage yield is improved. Water balance components (cm)
In Out ========================= ============================ Rain + snow : 121.55 Interception : 0.10 Runon : 0.00 Runoff : 0.00 Irrigation : 0.00 Transpiration : 23.08 Bottom flux : -51.26 Soil evaporation : 28.91 Crack flux : 0.00 ========================= ============================ Sum : 70.29 Sum: 52.09
Running the model with 20 years of data. There's no fertilizer application in the field at this point.
Addition of 180kg/ha of mineral N feritilizer at the planting date to each of 20 years cropping period
Simulated yield has improved a lot but something went wrong with the addition of fertilizer. Simulation stopped on 2010.
Surprisingly the model ran successfully just by changing the fertilizer volatility fraction from 0 to 0.25. Updated the plots by replacing forage yield by grain yield (dry weight kg/ha of storage organ: CWSO). I also removed the groundwater component in the plot and all potential yield component right now. Values on 2022 seems very small, possibly because simulation stopped on October and not all the way till December.
Great, thanks Suman. These are looking good. Regarding the low yields in 2022, we specify harvest sometime in September, right? Assuming that's the case I don't think the shortened simulation would impact yields. However, 2022 has been really dry and from your earlier soil moisture plots it looked like the soil profile was drying out too quickly in the simulations, so it might stem from that. From my initial look at this both the yield numbers and nitrogen uptake numbers seem low, but we can figure out what's causing that down the road (that could also stem from low soil moisture values). Here's what I'd like to have done for the meeting Friday: Update results output:
Just adding irrigation based on pressure head: critical pressure head being -800 cm and irrigating to bring soil back to FC. Tried different scheduling criteria, but none made a significant difference (yield was similar for different methods). Yields in 2007 and 2022 have been improved.
Just adding rainfall plot (April-September) for all years and converted yields in bushels/acre
Thanks Suman. I'm a little confused by the 800cm irrigation threshold but I'll look through the SWAP documentation to get a handle on that. I'm surprised irrigation didn't make more of a difference. Can you show a plot of the total depth of irrigation applied in each year through time?
It would be interesting to compare the plots of soil moisture, ET and rainfall for the years where irrigation made a difference. Could you generate those plots for 2007 for the rainfed and irrigated simulations. On the plot for the irrigated simulation, could you add irrigation information so it shows how much was applied and when?
Including the total depth of irrigation applied each year in the above plot
Plotting irrigated vs non-irrigated simulation yields on the same plot (had to import yield values column from non-irrigated simulation results).
2007 simulations under irrigated conditions
2007 simulations under rainfed conditions
Simulations with fixed weekly irrigation, rootzone to field capacity (TCS=6).
The threshold for weekly irrigation=1.0 [0.0. . 20.0 mm, R]; only irrigate when deficit is higher than threshold
Thanks Suman. There seems to be something weird going on with the 2007 rainfed soil moisture plots, I'm not sure why you'd see so much more variability at 30 cm relative to 10cm. Can you double check those different depths are being plotted correctly.
In any case, this is interesting to see because there appears to be something depressing plant production, and I thought it might be water stress but looking at these results that doesn't appear to be the case. So we'll need to spend some time looking through the different parameters within the detailed crop model and see what might be causing that (may be something similar to what we found with nitrogen applications).
Can you adjust the nitrogen uptake and leaching charts so that the irrigated/nonirrigated are plotted on the same axes (same as they are for yields)?
Thanks, Dr. Shortridge for pointing out that issue. I reran the same code and it fixed the soil moisture plots - not sure how I got that one before. I updated both the soil water and rain-ET plots
Updated plot with irrigated/non-irrigated N uptake and leaching on the same axes
Updated plot with few sensitive parameters changed for improved yield and N uptake under different irrigation levels. Posting it here just to compare the total values with the above plots generated in the past
FC set to -330 cm, irrigation threshold 50% of AWC
FC set to -60cm, irrigation threshold 50% of AWC. Increased leaching and reduced N uptake with increased irrigation depths
Please post results from each iteration to this issue.