search

JamesSample / icpw2

New decade, new workflows
http://www.icp-waters.no/
0 stars 0 forks source link

Analysis summary (Thematic report 2023) #5

Open JamesSample opened 1 year ago

JamesSample commented 1 year ago

This issue summarises the analyses requested by Rolf in Bestilling.docx (e-mail received 15.04.2023 at 14.17).

Site selection

  1. Remove sites that did not pass the criteria for estimating monotonic trends according to Garmo et al. 2020.
  • Time series had at least one data point within the first 5 years (start of 1990 to end of 1994)
  • Time series had at least one data point within the last 5 years (start of 2016 to end of 2020)
  • Time series included data from at least of 18 of the 31 years.

Completed. As previously discussed, the selection criteria are not exactly as described above. See this notebook for details.

Maps

  1. Maps showing the stations that passed the criteria for estimating trends from 1990 to 2020. (Figure 1 in Garmo et al.)
  2. Maps of the stations (subset of sites shown in Figure 1) included in the analysis of three 15-year time series (similar to Figure 2 in Garmo et al.)

First draft completed. At present, we are using the same sites for all periods (as discussed). A preliminary map is here. Once the site selection is finalised, I'll tidy this up.

Trends

  1. Detn. trends in annual median values of the parameters between 1990 – 2020 in each of the 13 regions, incl. percent decline for the period. Note that we are not using sea-salt corrected values.

Partly completed. See notebook 04 and the by_region worksheet of this Excel file. Note that we have fewer than 13 regions given the current site selection. I have not yet calculated the percentage change for each period, but this should be fast.

  1. Difference between the annual median value and the annual median of the first year in the time series, divided by the standard deviation for each of the parameters. Moving average representing lowess (Locally Weighted Scatterplot Smoothing) smoothers.
  2. Figure of time series trends between 1990 – 2020 for the list of parameters based on statistical data from 5. (i.e., similar to Fig. 3 in Garmo et al.).

Completed. See notebook 05 and the PNG file here.

  1. Regional median levels for the list of parameters in each region for the periods 1990-2004, 1998-2012, 2008-2020.
  2. Monotonic trends, using Mann Kendall test for the list of parameters in each region for the periods 1990-2004, 1998-2012, 2008-2020 from the value of the test statistic (Z-score).
  3. Detn. slopes for the list of parameters in each region using the Sen estimator (Sen, 1968) for the period 1990 – 2020 and for the periods 1990-2004, 1998-2012, 2008-2020
  4. Detn. regional median slopes and statistical significance of the trends in annual median values for each of the parameters using the Regional Kendall Test.
  5. Sen slopes for each region for the periods 1990-2004, 1998-2012, 2008-2020 in the list of parameters

Completed. See the by_region worksheet of the Excel file linked above.

  1. Sen slopes for each site for Ca, Mg, Ca+Mg, (Ca+Mg)/SO4, SO4, Al, TOC, H+, and ANC.

Completed. See the by_site worksheet of the Excel file linked above.

  1. Number of sites for each region showing positive or negative trends (p<0.05, MK-test) in for the list of parameters (similar to Table 5 in Garmo et al.)

Partly completed. All relevant data are in the by_site worksheet of the Excel file linked above, but I have not yet summarised this into the requested table.

  1. The relative change in 2008 – 2020 compared to the period 1990-2004 (Table 4 in Garmo et al.)
  2. Test for differences for each parameter between Sen slopes for the three periods using a paired t-test based on statistical data from 10.

Not completed. I haven't looked at these yet.

@RolfVogt