Pacea package: NPI and ONI monthly anomalies?

[SOLV-Code]

Intro

Most of the monthly data sets for coastwide indices in pacea have anomalies. oni has both values and anomalies. The npi_monthly data set has only values, but not anomalies.

How anomalies are calculated varies substantially between indices. Started a note re: Trends vs. Anomalies

Have 2 specific items to resolve:

• trying to calculate the anomalies for NPI and need to resolve among alternative versions
• trying to understand the base values used for the ONI anomalies

A general request:

• wherever possible, provide the annual value as well as the anomaly in the pacea data objects.

NPI Anomalies

trying to recreate NPI Anomalies

testing in this script

• discrepancy in base reference value? mean of monthly values is different from mean of annual values

# ?npi_annual states that anomalies are calculated relative to 1925-1989 mean
base.mean <- mean( npi_monthly %>% dplyr::filter(year %in% 1925:1989) %>% select(value) %>% unlist(),na.rm=TRUE)
base.mean  # comes out as 1012.503
# different from the 1008.9 hPa value in the npi_annual help file?
# mean of the annual values matches
mean(npi_annual %>% dplyr::filter(year %in% 1925:1989) %>% select(value) %>% unlist(),na.rm=TRUE)

• shouldn't the monthly anomaly be calculated relative to the mean for that month?

This script generates a comparison plot of the three versions:
Summer months show positive anomalies when compared to a single overall base value, but not when compared to a month-specific base value.

Request for NPI Anomalies

• put in a request to include the correct anomalies for the npi object in pacea

ONI Anomalies

the oni object has monthly values and anomalies. The helpfile at ?oni states "The Oceanic Niño Index (ONI) is a 3-month running mean of sea surface temperature (SST) anomalies in the Niño 3.4 region (5 deg N to 5 deg S, 120 deg W to 170 deg W) plotted on the center month. The SST anomalies are calculated based on 30-year base periods that are updated every 5 years, which accounts for global warming and some of the decadal-scale SST variability (as seen in the Pacific Decadal Oscillation index)."

Because the oni object includes both values and anomalies, the base value used for each monthly anomaly can be calculated. Figure below, generated with this script shows annual base values for August and January.

• some parts of the series have constant base value for 5 year increments, but others do not. How come?
• More generally is the detrended ONI series useful for salmon models, or should the underlying increase be captured as well? Or should the overall increase be a separate covariate? (Note: The PDO index is handled the same way, with warming trend removed. See discussion here