Clarification Needed on 'mxharvests' and 'mxsowings' Dimensions in the output History Files

[Jyoti Singh]

While analyzing the output from CTSM history, I came across the dimensions 'mxharvests' and 'mxsowings, ' which prove puzzling. The output files show that the value for 'mxharvests' is 2, and the value for 'mxsowings' is 1. I am trying to understand the implications of these values and how they might affect my analysis of crop outputs.

Could someone please explain:

1. What do these dimensions specifically represent?
2. Why is there a difference between the number of maximum harvests and maximum sowings?
3. How could these values impact the analysis of crop-related outputs?

Any insights or references to documentation that could help clarify these questions would be greatly appreciated.

 

[Jyoti Singh]

I got the following description for the 'mxharvests' and 'mxsowings' dimension from CTSM ChangeLog file:

Tag name: ctsm5.1.dev088
Originator(s): samrabin (Sam Rabin)
Date: Mon Mar 28 10:13:22 MDT 2022
One-line Summary: Add outputs for annual crop sowing and harvest dates

Purpose and description of changes
----------------------------------

Added annual outputs of sowing and harvest dates (`SDATES` and `HDATES`,
respectively). This should simplify the determination of sowing and
harvest date for postprocessing.
- Sowing dates are on new dimension `mxgrowseas` (maximum number of
growing seasons allowed to begin in a year; currently hard-coded to 1).
- Harvest dates are on new dimension `mxharvests` (maximum number of
harvests allowed in a year), which is `mxgrowseas`+1. This is needed
because in year Y you might harvest a field that was planted in year
Y-1, then plant and harvest again.
- The lengths of these dimensions are public constants of `clm_varpar`.

Reading from the Log it explains why this change was needed:

- Non-winter-cereal patches that had live crops at the beginning of the
year did not get planted later that year.
- There was some odd behavior for rice patches at exactly 0 deg latitude
- Crop root depth had unexpected values outside the growing season; now
root depth is set to 0 outside the growing season

Reading from this document I assume, the second value of mxharvests will be useful for
non-winter-cereal patches that had live crops at the beginning of the year?

 

[samrabin]

Hi Jyoti!

I actually introduced these dimensions for postprocessing purposes. I wanted to follow the GGCMI protocol for assigning growing seasons to particular years: If a crop is planted in November 2023 but harvested in May 2024, the yield should be associated with the 2023 "growing season."

With one sowing allowed per calendar year (i.e., mxsowings = 1), some gridcells can experience two harvests in a calendar year. That's why mxharvests (maximum number of harvests) is two. If a gridcell has two values for (e.g.)
GRAINC_TO_FOOD_PERHARV in a single calendar year's output file, the first one was planted in the previous calendar year. You can confirm this by looking at SDATES_PERHARV (day of year of planting, 1-366) and SYEARS_PERHARV (year of planting), which together tell you the sowing date associated with each harvest.

Before introducing mxharvest-dimensioned outputs, the only way to determine what year a crop was planted in would have been to look at daily outputs and trace backwards from harvest date to sowing date—no fun.

The mxsowings-dimensioned outputs are all related to sowing: start and end of sowing window, actual sowing date, and sowing reason. The latter two also have mxharvest-dimensioned equivalents, but it's nice to have mxsowings-dimensioned versions to use if you're interested in sowing per se.

How these affect your analyses depends on what variables you're looking at. Let me know and I can elaborate.

Thanks for pointing out the lack of documentation about this, by the way. I've filed an issue to remind myself to add some.

 

[samrabin]

I should add—outputs with dimensions of mxsowings or mxharvests should only ever be saved annually. They don't really make sense otherwise.