23ISD006_Addendum__1.pdf

Notice to Prospective Proposers

Request for Proposals (RFP) No.

ADDENDUM NO. 1

March 01, 2024

The California Air Resources Board (CARB) is issuing this Addendum No. 1 to post Questions and Answers to the RFP and a revised RFP incorporating new Key Action Dates.

Proposers are advised to review the new Key Action Dates as they affect your submittal response.

All other information in the RFP shall remain the same.

This addendum, along with the RFP is published online in the California State Contracts Register at https://caleprocure.ca.gov/pages/index.aspx. To ensure receipt of any addenda that may be issued, interested parties are encouraged to register online at https://caleprocure.ca.gov. Any questions are to be addressed to the contact listed below:

Contact: Sakinah Carter Phone: (279) 208-7804 Email: S @arb.ca.gov

Thank you.

RFP 23ISD006 Questions and Answers

Question #1: Does CARB envision on-the-ground monitoring data to include field samples (e.g., soil organic carbon (SOC), GHG measurements), flux tower measurements, or observations (tillage practices, crop type, etc.) that are needed as training or validation dataset for developing the remote sensing product and the BGC model? Should resources required to obtain such monitoring data be included in the budget estimates?

Answer #1: This RFP does not include the establishment of any new on-the-ground observation networks. That being said, in-situ empirical observations are required to develop accurate remote sensing and modeling processes and products. To that end, if potential contractors for see the need for data that is not publicly available or does not currently exist, but is crucial to the success of this framework, then resources should be built in to either purchase or collect the requisite data.

Question #2: Would CARB facilitate any collaborations that would allow successful bidders to connect to groups that have on-the-ground training datasets for CA croplands? Or is the expectation that the potential bidders should already have connections with such groups at the proposal stage?

Answer #2: CARB staff can help the contractors make connections with potential data providers.

However, the contractors should identify the organizations that possess the needed information, and should not rely on CARB to find the data or organizations for the contractors.

Question #3: Should the proposed methodology/ framework incorporate long-term field or RS-based monitoring components in order to better detect management-induced changes in GHG/SOC? If so, would cost estimates for such monitoring efforts be constrained for practical purposes?

Answer #3: The proposed methodology should not include any new long-term field data requirements that do not currently exist. However, if an established field data collection network currently exists and has a reasonable expectation of sustainment, such as the NEON, AmeriFlux, and Fluxnet networks, then the framework can incorporate that data. The framework shall include a remote sensing component. One objective of this RFP is to develop an agricultural monitoring network that is remote sensing based and less reliant on reported or in-situ information. The remote sensing data that acts as the foundation of the long-term monitoring and modeling system should be open and freely available. However, if the contractor may need some non-public or proprietary data for model development, calibration, or validation, then those costs should be built into the budget. Such data can be used for these initial development purposes, but non-public or proprietary data that has costs associated with it, shall not be the foundational data on which the ongoing monitoring framework is build. For example, the contractor may purchase or collect Lidar data if needed for model development, calibration, or validation for a framework that is built upon something free, like Landsat or Sentinel data. In this way, the framework can continue into the future regardless of State budgets and funding for such a program and be used by others that have no prospect of ongoing funding.

Question #4: Does the suggested annual time-step monitoring scheme include SOC which might not change so quickly?

Answer #4: Yes, estimates of CO2, CH4, and N2O fluxes, as well as changes in SOC should be reportable at least on annual time steps.

Question #5: Is CARB interested in monitoring management impacts on C and GHG emissions, which would need to be monitored at a finer scale? Or is CARB mostly interested in getting aggregated estimates of management effects and the associated uncertainty at the county-level?

Answer #5: The monitoring and modeling framework shall be able to quantify the impacts of nature based solutions on California’s agricultural systems. However, the accuracy of such a quantification scheme should be maximized at programmatic, county, and state scales. It is not necessary that this framework be able to perfectly capture the impacts of individual projects. This framework will not be used for carbon credit quantification for example. However, the State of California will be increasing deployment of climate smart agricultural practices and this framework must be able to estimate the impact of that increase and how implementation may be adjusted to enhance climate benefits. The intended users of this framework are county and state governments for landscape scale emissions and carbon accounting purposes, not farmers or carbon project developers.

Question #6: Does CARB have requirements for the choice of the BGC model? Can a proprietary model be used if the model is freely available but the code is not open-source?

Answer #6: The model shall be open-source in that all of the code used within this framework is open and available for public consumption. Any proprietary systems used within this framework should be ancillary to the main framework and swapable. For example, if some sort of simple visualization system is built upon an ESRI platform, that is possible as long as it is not critical to the framework functionality and it can be swapped out with another visualization system if desired. However, all of the code used for scientific computing and processing, and data development shall be open to all with one final version given to CARB which can be maintained and manipulated by CARB staff through time.

Question #7: Does the monitoring framework need to be able to quantify historical emissions? If so, how far back is required?

Answer #7: This is not required, however, it would be beneficial to the overall monitoring and modeling framework to have some historical measurements, and is most likely necessary for validation and calibration of the framework. CARB would request estimates as far back as the methodology and data allow, however, there is no stipulation as to how far back this would have to go. If the contractors intend to use a brand new space-based sensor, then very few historical estimates will be requested. The contractor must demonstrate accuracy through a calibration and validation process, which most likely would require several years of historical data..

Question #8: Key Deliverable #3: what is the timeframe for the monitoring data needed to run the BGC model? For example, annual data from 2015 to present?

Answer #8: The monitoring data in key deliverable #3 should constrain the model in key deliverable #2 to result in a monitoring framework for all California agricultural systems. In this way, the monitoring data required is dependent on the model that the contractor(s) develop. If the model requires sub-annual remotely sensed data then the contractor will need sub-annual monitoring data. For example, MOD17 is a simplified BGC model that estimates GPP/NPP on MODIS temporal/spatial resolutions. MOD17 ingests MODIS derived LAI and FPAR, which constrains and drives the MOD17 algorithm. The timeframe of how far back the data should go, is also dependent on the data used by the contractor to drive their algorithm resulting from this contract. If they are using a brand new space-based sensor, then they obviously cannot generate estimates too far back. However, the contractor must demonstrate accuracy through a calibration and validation process, which most likely would require several years of historical data.

Question #9: Task 2.6 Inventory Demonstration and tutorial. What years would the inventory cover?

Ex: would the demonstration to calculate the inventory be for 2024?

Answer #9: No exact specification for how many years from the past must be included, and will depend on the data and methods used to generate the framework. Further, the lag time between when estimates can be generated and today, is also not specified. However, framework proposals with shorter lag times, meaning how soon to present day can estimates be generated, will be more competitive. That being said, accuracy and functionality are more important than lag time, and this framework certainly does not need to generate anything close to “real-time” or even “near real-time” data.

Question #10: Is there a requirement for the latency of the monitoring products? For example, for 2025 crop production year, do the monitoring products need to be available by Q1 2026? When in the calendar year would the annual GHG inventory be modeled?

Answer #10: The lag time between when estimates can be generated and today, is not specified. However, framework proposals with shorter lag times, meaning how soon to present day can estimates be generated, will be more competitive. That being said, accuracy and functionality are more important than lag time, and this framework certainly does not need to generate anything close to “real-time” or even “near real-time” data.

Question #11: Will CARB be able to provide ground reference data to enhance and validate the monitoring data workflow? Ex field-level management practices that can be used to validate and report uncertainty of the monitoring products.

Answer #11: CARB has no special access to any agricultural data. It will be the responsibility of the contractor to obtain all of the necessary data for development, calibration, and validation.

However, CARB will help the contractor facilitate the initial contacts with contractor-identified data providers. CARB can also help provide contact information with collaborators in sister agencies, such as the California Department of Food and Agriculture, who may know where the contractor might find additional data.

Question #12: Does open source include sharing the source code publicly (ex in a GitHub repo)?

Answer #12: Yes.

Question #13: Instead of open-sourcing software and remote-sensing models that creates primary data that are publicly available (https://www.ctic.org/OpTIS), would making open-source code used to process existing available monitoring data sets, that would include a table showing cropland management data by county and crop type, be compliant with CARB goals regarding transparency and availability?

Answer #13: All code used for scientific computation and data development within this framework shall be made available to the public. Ancillary processes to package or visualize outputs may consist of proprietary products, or non-public code as long as it is, agreed upon with CARB staff, not essential to the foundational modeling and monitoring system and could be swapped with another product or code if desired. The remote sensing data products that are used as the foundational constraint for the framework must be publicly available. However, the entire processing chain that produces that publicly available derived data product does not have to be made publicly available. For example, if the contractor would like to use a USGS or NASA developed cloud-free, research ready remotely sensed product, it is not necessary that the entire processing chain that makes this product be publicly available. However, the derived product must have a history of being publicly and freely available, comes from a reputable and sustainable source with some assurance of continued development and openness, such as a federal agency, and its underlying algorithms are published in scientific journals or government reports.

Question #14: Would CARB consider extending the submission timeline? An extension by even a few weeks to March 26th would enable our team and other interested bidders to provide a stronger and more detailed proposal.

Answer #14: CARB will provide an extension to the Final Date for Proposal Submission/Receipt by CARB. See updated key action table for new dates.

Question #15: Could you please clarify the expected relationship between the monitoring system described in the RFP and the existing NWL? Does CARB have a preference whether data from the 2018 NWL is used for validation of the new model, or if the new inventory is kept independent of NWL?

Answer #15: There is no need for the contractor to take into consideration the previous version of the Natural and Working Lands Inventory for croplands. CARB staff will determine how previous inventory estimates will be considered in future inventory reports.

Question #16: The RFP mentions "developing" or "creating" a model in some places and "using a simplified" model in others, implying modification of an existing model. Could you please clarify whether using an existing model (with or without modifications), rather than creating a new one, would be considered responsive?

Answer #16: Either creating a brand new model from scratch, or modifying an existing model are both viable options. It is unlikely that existing models without modification will meet the requirements of the framework. However, if potential contractors find such options that will meet all of the needs of this framework, then that is also a viable option. CARB is unaware of any process based open-source models that are constrained by remote sensing data and suitable for estimating CO2, CH4, N2O, and SOC in agricultural systems which are also responsive to climate and management changes on the ground, and which can also be used for future projection modeling given climate and management scenarios.

Question #17: The management actions to be included in modeling are stated as "conventional agriculture as well as all healthy soils practices that are funded by the CDFA, stacking of these practices, and organic projects". Since the RFP is clear that the project is focused on "annual and perennial cropland", are we correct to interpret this as incorporating HSP categories I (cropland) and II (orchard or vineyard) but not category III (grazing land)?

Answer #17: Yes, that is correct.

Question #18: Is there a lower limit to the degree of model simplification? As a hypothetical example, would a model with carbon fluxes that are sensitive to management but with N2O and/or CH4 predictions derived from county-level emission factors be considered responsive?

Answer #18: Yes, there is a limit to model simplification and, no, county level emissions factors are not sufficient. All emissions and carbon stock changes should be consistent and derived using an integrated algorithm and be responsive to changes in climate and management. For this reason, the term “biogeochemical model” is used within this RFP. This implies that the underlying processes that result in these fluxes and stock changes are linked and influence one another. However, the degree to which these processes are represented within the model is dependent on the methods proposed by the contractor and maybe simplified up to, but not including, the point where the model is no longer constrained by the remote sensing data, is not responsive to climate and management changes, and is not sufficiently temporally or spatially explicit.