Updated the supplementary report to reference the newly-released Roberts et al. (2024) publication that documents the model. The model itself was not changed.

Completed the supplementary report documenting the details of this model. Corrected the 5 and 95 percent rasters so that they contain the value 0 where the taxon was asssumed absent, rather than NoData. Nothing else was changed.

This model is a major update over version 11. We incorporated additional surveys from all collaborators through 2020. (Due to data limitations, the last month predicted by the model is September 2020.) We reformulated and refitted all detection functions and spatial models. We updated all environmental covariates to newer products, when available, and added several covariates to the set of candidates. Similar to verison 11.1, uncertainty outputs account for variance derived from uncertainty in model parameter estimates and from temporal variability. In all regions, this includes both seasonal and interannual variability, except in August and September north of the Great South Channel, where data limitations necessitated the use of climatological covariates, for which only seasonal variability is derivable. The model was released as part of the final delivery of the U.S. Navy Marine Species Density Database (NMSDD) for the Atlantic Fleet Testing and Training (AFTT) Phase IV Environmental Impact Statement.

In this update, the density surfaces remain unchanged from version 11, but we have added uncertainty surfaces to go with them. Section 5 of Roberts et al. (2021) gives complete details of how uncertainty was derived. As discussed there, the density surfaces represent means for each month averaged over an era, either 2003-2009, 2010-2018, or 2003-2018. The uncertainty surfaces estimate how much density is likely to vary from the mean if a random year was selected from that era, accounting for the estimated statistical error in model parameter estimates and, when possible, the interannual variability in the model covariates over the era. The uncertainty surfaces include standard error (see Roberts et al. 2021 for the specific definition), the coefficient of variation, and the lower (5%) and upper (95%) limits of a 90% confidence interval. These are similar to what we provided with the model up through version 8, but now account for interannual variability. (Version 8 and prior only accounted for the estimated statistical error in model parameter estimates.)

This was a minor update to version 10, undertaken in response to a request from the Massachusetts Division of Marine Fisheries that we reexamine the problem of how best to estimate abundance in Cape Cod Bay for the month of December. We prepared a new estimate based on all surveys conducted by the Center for Coastal Studies during the month of December from 2003-2020. The only predictions that changed during this update were for grid cells of Cape Cod Bay in the month of December. For more information about this, please see Section 4 of Roberts et al. (2021) and the Appendix of that report.

This was a minor update to version 9, undertaken after NOAA NEFSC (B. Shank) reported implausibly high densities in spring months in the vicinity of Massachusetts Bay. To correct this problem we introduced additional survey data collected by NEFSC in spring of 2019, fixed a model term that caused an unrealistic extrapolation into Massachusetts Bay and refitted the "Spring" model in the "North of Nantucket Shoals" region. This fixed the problem in spring months around Massachusetts Bay, while leaving the rest of the Gulf of Maine region essentially the same as version 9. The additional data included additional effort and sightings in the "Hatteras Island to Nantucket Shoals" region, so we refitted that model as well, resulting in slightly higher predictions south of Nantucket. Predictions south of Hatteras Island are the same as version 9. For more details about this update, please see Section 3 of Roberts et al. (2021).

This was a substantial update over version 8. We extended the aggregate database of surveys to extend up through early 2019 (vs. through 2016 in verison 8). The additional surveys contributed by collaborators included: NOAA NEFSC and SEFSC AMAPPS aerial and shipboard from 2016, 2017, and spring 2019; NEFSC pre-AMAPPS HB-07-09 cruise; NEFSC NARWSS for 2017 through spring 2019; southeast U.S. right whale EWS surveys from 2016/17, 2017/18, and 2018/19; VAMSC Navy surveys for 2016-2017; UNCW Navy surveys 2017; HDR Navy surveys for 2018; NYS-DEC/TetraTech NYBWM Years 1 and 2 surveys (2017-2018); New England Aquarium NLPSC 2011-2015 and MMS-WEA 2017-2018 aerial surveys. We shifted the start date of the model forward to 2003 (vs. 1998 in version 8), to make it more recent and to better equalize survey coverage between the north and south. We increased the spatial resolution to 5 km. At NOAA's request, we adjusted the study area to extend farther inshore in certain bays and estuaries, to facilitate better use of the model in NOAA's right whale take reduction decision support tool. We reformulated and refitted all detection functions and regional spatial models. We applied new availability bias corrections formulated using the Laake et al. (1997) estimator; the corrections accounted for survey altitude and speed, regional whale dive behavior, and, where possible, whale group size and composition. We summarized results into two eras, 2003-2009 and 2010-2018, reflecting the apparant major shift in right whale distributions around 2010. For comparison, we also summarized results in to a third era that spanned the entire period 2003-2018. We documented the overall model in a report to the U.S. Navy (Roberts et al. 2020) in summer 2020 and released it for public use in collaboration with NOAA Fisheries. The Navy Marine Species Density Database (NMSDD) was not updated at this time; the Navy will take delivery of the model in Feburary 2022 as scheduled.

Updated density predictions for Cape Cod Bay for January-May with estimates from Ganley et al. (2019). Adjusted subregion definitions and model formulations to eliminate edge effects. Otherwise this version is identical to version 7. Model released in collaboration with NOAA Fisheries following the April 2019 ALWTRT meeting.

Switched back to four season model with new season and subregion definitions (relative to Roberts et al. 2015). Model released as part of a scheduled update to the U.S. Navy Marine Species Density Database (NMSDD).

Switched calculation of monthly 5% and 95% confidence interval rasters to the method used to produce the year-round rasters. (We intended this to happen in version 5.4 but I did not implement it properly.) Updated the monthly CV rasters to have value 0 where we assumed the species was absent, consistent with the year-round CV raster. No changes to the other (non-zero) CV values, the mean abundance rasters, or the model itself. Model files released as supplementary information to Roberts et al. (2016).