BRAT_SOW_Jordan_2.docx

STATEMENT OF NEED

NFFP7100-19-03067

PURPOSE:

statement of Need The beaver restoration assessment tool (BRAT) is an established, peer reviewed decision support and research tool for quantifying the restoration potential and risks associated with beaver. BRAT calculates both the present and historical beaver dam capacity for individual reaches. It also identifies potential management issues that might arise when beaver are present.

The current version of BRAT is open-source, research-grade software that is currently in its third iteration (http://github.com/riverscapes/pyBRAT). It is operational but has several limitations that prevent the model from being run at a regional scale. However, even more crucially, technical issues prevent users from easily implementing and interpreting the model results. These limitations involve dependencies on proprietary software tools and the form of the spatial data outputs that are not easily shared or accessed.

Therefore, an updated version of the BRAT project is required that addresses the aforementioned limitations. The overarching goal is to make the model easier to run at a regional scale. This represents improvements in two essential elements of the current software: 1) Make the model easier to run, and 2) Make it easier to disseminate the results in easy to consume products.

Background Beaver modifications on the landscape, especially their dams that impound water and sediment, can dramatically affect both the biotic and abiotic characteristics of the stream network in which they occur. These effects can produce numerous benefits to plants, fish, and wildlife in terms of habitat creation and increased complexity that are often desired in degraded systems in need of restoration. River restoration project goals often focus on improving the hydrologic functions, geomorphology, vegetation, water quality, and habitat for many aquatic and terrestrial species. Dam-building activity by beaver, or human-made beaver dam mimics (beaver dam analogues), can be utilized to reach these goals by taking advantage of their habitat modifying effects.

The potential benefits of employing a beaver-based stream restoration approach are numerous; however, so too are the potential for beaver-human conflicts. The benefits and conflicts arise from an overlap of preferred habitats by both humans and beavers, misunderstandings of how beavers modify their habitats, and a lack of planning or use of adaptive management on restoration projects. The Beaver Restoration Assessment Tool (BRAT) was developed to support decision making around the use of beaver dam building as a stream restoration modality by assessing the natural (current and historic) dam density across a river network and evaluating the potential for conflicts between human land-use and beaver dam building activity.

The core algorithms underlying the calculations that BRAT performs are valuable, vetted and have widespread potential for managers trying to make decisions related to beaver and stream restoration. However, these same managers are unable to use the model in its current form because it is difficult to configure and run for new regions. And even where the model is run successfully, the results are extremely cumbersome to use and extract any value from. Decision makers trying to understand the potential and risks associated with beaver could benefit greatly from an improved model that was easier to use and consume.

Specifically, this task requires addressing each of the following limitations in the current form of the watershed assessment model.

· Dependencies on proprietary GIS software that both preclude cost-effective batch processing, but also drastically limit the end user audience who can use the results to those possessing specific, and potentially, expensive software.

· Inability to package the BRAT results into widely distributable cartographic products that are easily accessible to end users.

· The research focus during the development of the BRAT up to this point has resulted in a model that does not scale when run at the regional scale:

· Model run configuration involves complex, location specific processes,

· Obtaining and configuring the model inputs is manual and labor intensive,

· Monitoring large batch runs is virtually impossible without considerable effort,

· Tracking where the model encountered errors is itself error prone,

· The outputs do not possess sufficient metadata to allow integration of results across regions or use in post processing tools.

TECHNICAL SPECIFICATIONS:

Model Improvement framework A new version of BRAT is required that addresses the aforementioned limitations. The overarching goal is to make the model easier to run at the regional scale. This involves two essential elements: 1) Make the model easier to run, and 2) Make it easier to disseminate the results in easy to consume products.

Making the model easier to run requires three main technical tasks. First, all proprietary ESRI dependencies should be removed so that the model can easily be batch processed, and also freeing the model results to be consumed by those without this specific software. Second, a suite of coding best practices should be implemented that will standardize and simplify the way the model is configured and executed. This will greatly improve the reliability of each run and provide transparency to the process. Third, an improved process is needed to streamline the way model inputs are downloaded and prepared. Currently, a great deal of effort is expended simply locating and downloading the required data. The input data then needs to be formatted correctly, often involving fragile cartographic re-projection and clipping steps to transform the original national datasets. Overcoming these limitations would greatly improve the feasibility of running BRAT at larger scales.

Improving BRAT outputs to better disseminate the results involves both improved cartographic outputs, as well as providing a way for users to interrogate these outputs to learn why the model is working the way it is. The current version of BRAT produces a suite of ESRI layer packages that are difficult to share because of their size but also extremely brittle due to their association with a specific version of ESRI ArcGIS. BRAT analysts struggle to share these results with end users and then frequently run into problems when the files don’t work on their systems due to version inconsistencies. Even when they do work, these outputs are static and fail to educate users about how the model arrived at a particular result. Ultimately the outputs fail to help decision makers because the results are cumbersome to use and transmit, and are static and lack sufficient depth to support decision making.

The next version of the model will include the ability to publish the results to a simple, publicly available web map for anyone to use. This will work in any web browser, without any proprietary software. Several improvements are required to the BRAT outputs to make this possible, in addition to more comprehensive metadata that describes these outputs. A robust and repeatable workflow is also needed to execute the publishing step that feeds into the public web map. Finally, a simple series of interactive views will be required for exploring the results and helping decision makers understand the inner workings of the new BRAT model; what were the inputs, what rules were applied, what limiting factors are affecting beaver densities, where are beavers most suitable, what are the risks to introducing beaver in specific reaches?

In essence, the goal is to enhance the existing BRAT model to realize its full potential as a regional decision making tool.

Technical requirements All work done on this task must be documented as part of the open-source code repository that currently houses the BRAT project. All functionality developed must be compatible with the underlying BRAT metric calculations and logic models. All functionality developed will be based only on open-source tool sets such that no proprietary dependencies are built into the project. Documentation of all methodological details and development decisions must be sufficient to allow model functionality and output documentation and peer-evaluation or reproduction. The final work product must be open-source code and mark-down documentation and must be published to the existing BRAT GitHub open source code repository (https://github.com/Riverscapes/pyBRAT/releases/latest).

The specific product deliverables shall be.

1. Enhance the existing BRAT model with the following improvements. The resultant model will retain the current open source GNU license and be published publicly for anyone to download and use.

a. Remove all proprietary dependencies. All GIS tool dependencies must be migrated to alternative, open source geoprocessing technologies.

b. Abstract all model inputs and parameters to external configuration inputs file.

c. Make it possible to run the model from the command line without the need for a user interface.

d. Implement logging to improve transparency into the model operations.

e. Implement error handling such that all errors are tracked in the log.

f. The model outputs must include a fully populated project metadata file that validates against the published Riverscapes schema.

g. Migrate all outputs to publicly accessible formats and not require any proprietary software.

2. Automate the download, reprojection and clipping to the study area of the following BRAT inputs:

a. National Hydrographic Dataset (NHD) for an individual eight digit Hydrologic Unit Code (HUC)

b. National Elevation Dataset (NED) 10m digital elevation models (DEM)

c. National Transportation Dataset (NTD) roads, railways and canals

3. Verification of the updated BRAT model against prior runs of the model for three (3) eight digit HUCs within the state of Oregon. The verification will be delivered in the form of a report that clearly states the ability of the updated model to reproduce the results of the preceding version including:

a. Reach segmentation

b. Drainage area for each reach

c. Existing dam capacity

d. Historic dam capacity

e. Distance to conflict

4. Develop the ability for anyone to publish BRAT model runs such that the results can be consumed publically:

a. Display model runs in web accessible cartographic views showing:

i. Existing beaver dam capacity

ii. Historic beaver dam capacity

iii. Management and risk concerns

b. Construct interactivity that identifies individual reaches and shows the intermediate model data that led to the observed model result for the specific reach.

c. Build output as downloadable model run in a zip archive.

5. Populate the BRAT publishing system with the existing BRAT results that already exist for the state of Oregon in the following locations:

a. John Day River

b. Powder River

c. Burnt River

6. Update the publicly available BRAT documentation wiki (http://brat.riverscapes.xyz/) to reflect the new version of the model and other deliverables that are part of this work.

Contractor Minimum Qualifications:

· PhD in GIS and Remote Sensing or related field and 10+ years experience in spatial data software development for natural resource management applications.

· Track-record in developing cloud-based open-source spatial data analysis software and ability to make source-code changes to open-source codes underlying models specified for use in producing the project deliverables above (i.e. proficiency in Python, R, C#, XML, GDAL).

· Documented proficiency in developing open-source GIS applications with 10+ years experience in working with and acquiring national scale remote sensing based landscape and river network data and delivering ecological modelling analyses via interactive web user interfaces.

· Documented proficiency and experience in running and evaluating Beaver Restoration Assessment Tool model builds and runs.

DELIVERABLE SCHEDULE:

Delivery of all final work products must be by 9/30/2020

POINT OF CONTACT (POC):

Chris Luecken Chris.luecken@noaa.gov

(206) 860-3424