Add management practice priors by divine7022 · Pull Request #3 · ccmmf/scenarios

divine7022 · 2026-01-30T16:58:51Z

Summary

Adds documented prior distributions for CA ag management parameters, derived from NRCS practice standards, CDFA program guidelines, and UC ANR extension publications.

data/management_priors.yaml - Machine readable priors for ensemble sampling
R/sample_priors.R - Functions to sample from prior distributions
docs/management_priors.md - Documentation with source justifications
scenarios.yaml - Updated parameter values based on verified sources and added comments

Verification

All priors verified against primary sources:

NRCS conservation practice standards (329, 340, 345, 449)
CDFA compost application rates white paper (Gravuer 2016)
UC ANR publications (3396, 3470)
USDA NASS census 2022

Implements - #161

dlebauer

This is a solid step toward a structured priors system. It is clear that you have put a lot of thought into the specification, and this is a first step toward a framework that could be used for monitoring, future scenarios, and creation of test runs.

management_priors.yaml appears aligned with the yaml example I drafted in ccmmf/organization#170, although we skipped the explicit "Before implementation the first step is to discuss and have the specification reviewed." step.

Given the near-term need to simulate management inputs, I recommend that we address a few points, focusing on those that will more clearly define the approach and align with existing PEcAn functions and data (like events_schema.json and PEcAn.data.land::look_up_fertilizer_components).

Let's capture feedback from @mdietze, @sarahkanee and others, and triage what should be included in this PR vs what belongs in new issues.

To fix before merging:

Consistent Practice IDs. And these should be consistent with names in events.json where practical. We may need to expand the list of practices in events_schema.json; ideally we don't need a separate practice dictionary.
- specifically, in management_priors.yaml, practices uses 'cover_crop_annual', 'compost_application', 'irrigation_drip' whereas adoption_rates has 'cover_crop', 'compost', and compatibility has 'cover_crop', 'irrigation_flood', 'tillage'.
Make sure that README is consistent with scenarios.yaml; remove scenario_parameters.csv if obsolete.

Future work:

Gather feedback from group and triage - decide what's needed in this PR vs what can be clarified in new issues for future implementation.
- criteria for merging: meets near term needs to simulate management inputs and does not cause unnecessary technical debt (i.e. downstream pipelines that will not be transferable to monitoring and projection workflows, beyond what is necessary to learn through the process of implementation and application)
Consider what this would look like as a spreadsheet from the perspectives of end users, data validation, and code complexity.
Expand the priors to better represent the common crops/PFTs in CA, and regions. Also need to have generic CFT fallbacks. Coverage doesn't need to be comprehensive at the crop level, but should reflect what is available from CDFA/UCANR/NRCS sources for major CA crops.
Adoption rates will be useful in the short term for simulating monitoring workflows. Where applicable, these should be directly informed by initial conditions + transition matrices derived from LandIQ data.
- but note that even though LandIQ says it has mixed cropping systems, it does not, and this is one place that adoption rates (e.g. % of orchards with ground cover) will be helpful.

dlebauer · 2026-02-04T21:25:41Z

+
+### STIR Scale Background
+
+NRCS uses the Soil Tillage Intensity Rating (STIR) on a **0-200 scale**:


add definition of STIR - and confirm how this is linked to SIPNET tillage efficiency.

dlebauer · 2026-02-04T22:00:05Z

+
+### Our Model Parameter
+
+Our parameter `tillage_eff_0to1` represents SOM pool transfer fraction, not direct STIR. The conversion is approximately `tillage_eff ≈ STIR / 200`.


I think it is a good idea to use STIR to inform the SIPNET parameter, especially to inform the relative impact of different tillage tools and practices.

However, there isn't a direct mapping to a SIPNET parameter, and the parameter would probably be best calibrated using a time series of CO2 flux.

So for an MVP I think it would be sufficient to create a table of priors that represent no-till, reduced-till, and conventional

Suggested change

Our parameter `tillage_eff_0to1` represents SOM pool transfer fraction, not direct STIR. The conversion is approximately `tillage_eff ≈ STIR / 200`.

The SIPNET parameter $f_\text{till}$ (`tillage_eff_0to1`) represents the transient increase in decomposition rate that exponentially decays over time[^1]. SOM pool transfer fraction, not direct STIR. The conversion is approximately `tillage_eff ≈ STIR / 200`.

[^1]: SIPNET Documentation https://pecanproject.github.io/sipnet/user-guide/model-inputs/#tillage

dlebauer · 2026-02-04T22:01:02Z

+We use Beta distributions because:
+1. Parameter is bounded [0, 1]


need to separate bulleted list by blank line for the list to render correctly

Suggested change

We use Beta distributions because:

1. Parameter is bounded [0, 1]

We use Beta distributions because:

1. Parameter is bounded [0, 1]

dlebauer · 2026-02-04T22:42:07Z

+        distribution: beta
+        a: 7
+        b: 13  # mean=0.35
+      date:


typically tillage will occur some number of days before planting. It may make more sense to anchor tillage dates to planting.

It may sense to anchor other management types to planting dates, e.g.

irrigation only between planting and harvest (perhaps a few days ahead)

pre-planting fertilization; post-planting side-dress

dlebauer · 2026-02-04T22:47:07Z

+
+## Compost Application
+
+### CDFA Official Rates


PEcAn.data.land::look_up_fertilizer_components provides default values for different fertilizer and organic matter types, including the C:N ratio and mineral N content. Would that be helpful here? If CDFA specifies these very broad "compost types" it may not be necessary, but if there are statistics on the frequency and application rates of specific NCC types, it could be useful.

dlebauer · 2026-02-04T22:51:05Z

+
+CalRecycle lab data (n=1364 samples) shows typical finished compost C:N = 14-20.
+
+**Prior:** Normal(16, 3) -> 95% CI [10, 22]


should there be separate priors for high and low C:N OM additions?

dlebauer · 2026-02-05T18:06:26Z

+    - [reduced_tillage, cover_crop]
+    - [compost_application, reduced_tillage]
+    - [irrigation_drip, cover_crop_orchard]
+  incompatible:


what makes these incompatible? Need to add sources.

dlebauer · 2026-02-05T18:08:16Z

+
+metadata:
+  sources:
+    - id: nrcs_cps_329


For linking priors to references, I think it would be better to include the id, and then use a bibtex file to store the references. Create a group in Zotero with files and export the bibtex.

dlebauer · 2026-02-05T18:18:17Z

+    source: CDFA Gravuer 2016 Table 2
+    parameters:
+      org_c_kg_m2:
+        # Table 2: annual crops + lower N compost = 4.0-5.3 dry tons/ac


we should leverage and / or extend PEcAn.data.land::look_up_fertilizer_components to support this workflow (look up N & C composition)

organic matter ammendments typically have some mineral N (NO3 &/or NH4)

dlebauer · 2026-02-05T18:20:00Z

+        b: 15  # ~25%, flood dominant
+
+# Crop-specific baselines
+crop_baselines:


Need to have fallbacks for crops and regions not covered.
Regions should have an associated polygon [or perhaps could be a set of county names]

divine7022 added 3 commits January 30, 2026 11:56

update org_c_kg_m2 and org_n_kg_m2 value

349e065

add management priors

d5cbc01

Functions to sample from management priors

db25855

divine7022 marked this pull request as draft January 30, 2026 17:17

divine7022 changed the title ~~Management practices for different crops~~ Add management practice priors Jan 30, 2026

docs for management practice priors

c2a5c79

divine7022 mentioned this pull request Jan 30, 2026

Identify typical management practices for different crops in California ccmmf/organization#161

Open

2 tasks

divine7022 marked this pull request as ready for review January 30, 2026 19:17

divine7022 requested review from dlebauer, infotroph, mdietze and sarahkanee February 4, 2026 18:09

dlebauer reviewed Feb 5, 2026

View reviewed changes

dlebauer requested a review from ashiklom April 10, 2026 17:57

dlebauer mentioned this pull request Apr 11, 2026

Add White/Salinas 2020 cover crop + compost dataset entry ccmmf/uncertainty#5

Draft

1 task

AritraDey-Dev mentioned this pull request Apr 25, 2026

Pipeline: managements table → events.json with gap filling (MVP) ccmmf/organization#234

Open

5 tasks

divine7022 mentioned this pull request May 19, 2026

calval workbook to events.json pipeline ccmmf/uncertainty#6

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add management practice priors#3

Add management practice priors#3
divine7022 wants to merge 4 commits into
mainfrom
management_prac

divine7022 commented Jan 30, 2026 •

edited

Loading

Uh oh!

dlebauer left a comment •

edited

Loading

Uh oh!

dlebauer Feb 4, 2026

Uh oh!

dlebauer Feb 4, 2026

Uh oh!

dlebauer Feb 4, 2026

Uh oh!

dlebauer Feb 4, 2026

Uh oh!

dlebauer Feb 4, 2026

Uh oh!

dlebauer Feb 4, 2026

Uh oh!

dlebauer Feb 5, 2026

Uh oh!

dlebauer Feb 5, 2026

Uh oh!

dlebauer Feb 5, 2026

Uh oh!

dlebauer Feb 5, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants


		### STIR Scale Background

		NRCS uses the Soil Tillage Intensity Rating (STIR) on a 0-200 scale:


		### Our Model Parameter

		Our parameter `tillage_eff_0to1` represents SOM pool transfer fraction, not direct STIR. The conversion is approximately `tillage_eff ≈ STIR / 200`.

-Our parameter `tillage_eff_0to1` represents SOM pool transfer fraction, not direct STIR. The conversion is approximately `tillage_eff ≈ STIR / 200`.
+The SIPNET parameter $f_\text{till}$ (`tillage_eff_0to1`) represents the transient increase in decomposition rate that exponentially decays over time[^1]. SOM pool transfer fraction, not direct STIR. The conversion is approximately `tillage_eff ≈ STIR / 200`.
+[^1]: SIPNET Documentation https://pecanproject.github.io/sipnet/user-guide/model-inputs/#tillage

		We use Beta distributions because:
		1. Parameter is bounded [0, 1]


		CalRecycle lab data (n=1364 samples) shows typical finished compost C:N = 14-20.

		Prior: Normal(16, 3) -> 95% CI [10, 22]

Conversation

divine7022 commented Jan 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Verification

Uh oh!

dlebauer left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

divine7022 commented Jan 30, 2026 •

edited

Loading

dlebauer left a comment •

edited

Loading