Background

Below is a summary of the data comparisons between the Beck et. al 2022 paper and the NERRS data. Their analysis consisted of 3 steps. 1) Ran a gam analysis on the biweekly/monthly samples to estimate a smooth temporal pattern. 2) Used the smoothed gam to get a daily timeseries. 3) Took seasonal estimate averages from the daily times series. The goal of this is to propogate uncertainty through the analysis.

They used that estimates to evaluate linear trends over a moving average through different time periods (10 years). They interpreted the slope of the regression a representative for the central year in the moving avg block. Our timeseries is only 12 years so a 4yr moving window was used to get three distinct blocks like the Beck paper.

Data Summary

Parameter Beck et. al 2022 NERRS data from stations used in Rhegan’s Analysis
Time Series Length 30 years (1990 - 2019) 12 years (2007 - 2019) * earlier samples removed due to poor qaqc scores

Data only presented for Wells Bay, Maine and Waquoit Bay, Massachusetts because these stations had the fewest years missing.

GAM meta analysis

Wells Bay (Data Rich)

kts = 144 (12months * 12years) Based on Beck paper calculation

Timeseries GAM

Monthly

DOY

Summer seasonal average with regression

Summer seasonal average log slopes (window = 4)

Waquoit (Data Poor)

kts = 84 (7months * 12years) Based on Beck paper calculation

  • Seasonal plots error out because too little data.

Timeseries GAM

Monthly

DOY

Summer seasonal average with regression

## Error in smooth.construct.tp.smooth.spec(object, dk$data, dk$knots) : 
##   A term has fewer unique covariate combinations than specified maximum degrees of freedom
## Error in smooth.construct.tp.smooth.spec(object, dk$data, dk$knots) : 
##   A term has fewer unique covariate combinations than specified maximum degrees of freedom
## Error in smooth.construct.tp.smooth.spec(object, dk$data, dk$knots) : 
##   A term has fewer unique covariate combinations than specified maximum degrees of freedom
## Error in smooth.construct.tp.smooth.spec(object, dk$data, dk$knots) : 
##   A term has fewer unique covariate combinations than specified maximum degrees of freedom
## Error in smooth.construct.tp.smooth.spec(object, dk$data, dk$knots) : 
##   A term has fewer unique covariate combinations than specified maximum degrees of freedom
## Error in smooth.construct.tp.smooth.spec(object, dk$data, dk$knots) : 
##   A term has fewer unique covariate combinations than specified maximum degrees of freedom

Summer seasonal average log slopes (window = 4)

## Error in smooth.construct.tp.smooth.spec(object, dk$data, dk$knots) : 
##   A term has fewer unique covariate combinations than specified maximum degrees of freedom
## Error in smooth.construct.tp.smooth.spec(object, dk$data, dk$knots) : 
##   A term has fewer unique covariate combinations than specified maximum degrees of freedom
## Error in smooth.construct.tp.smooth.spec(object, dk$data, dk$knots) : 
##   A term has fewer unique covariate combinations than specified maximum degrees of freedom
## Error in smooth.construct.tp.smooth.spec(object, dk$data, dk$knots) : 
##   A term has fewer unique covariate combinations than specified maximum degrees of freedom
## Error in smooth.construct.tp.smooth.spec(object, dk$data, dk$knots) : 
##   A term has fewer unique covariate combinations than specified maximum degrees of freedom
## Error in smooth.construct.tp.smooth.spec(object, dk$data, dk$knots) : 
##   A term has fewer unique covariate combinations than specified maximum degrees of freedom