Drifter-derived seasonal climatology of Sea surface temperature in the Tropical Indian Ocean

Drifter processing

Figure 1 summarise the main analytical step used in this post. This data model is modified from Wickham & Grolemund (2016). Thre model consists of mainly three steps. The first, the dataset was downloade from NOAA as ascii file. Second, the dataset was imported in R using the dplyr’s read_table() function. Third, once the dataset was in the workspace, it was manipulated and transformed. These dataset has twelve variables, I cleaned the dataset by removing some variables not needed for this post. I kept the common id, longitude, latitude, year, month, day and sst variables. Fourth, the observations were later grouped based on monsoon seasons. All drifter observations recorded between April and September fell into southeast monsoon season (SE). And those between October and March fell in the northeast (NE) monsoon season. Table 1 show drifter observation after being procesed.

Figure 1: The data analysis model

# read the file from the local drive
drifter = read_table2("./drifters.txt",
                      col_names = FALSE, comment = "#")%>%
  select(id = 1, lon = 2, lat = 3, drogue =4, u = 5, v = 6, 
         sst = 7, year = 8, month = 9, day = 10, hour = 11)

drifter = drifter%>%
  mutate(date = make_date(year, month, day),
         velocity = sqrt(u^2 + v^2),
    # make season variable
         season = month,
    # convert months into seasons
         season = replace(season,season %in% c(10,11,12,1,2,3), "NE"),
         season = replace(season, season %in% c(4:9), "SE"))%>%
    # select variables of interest and drop others
  select(date, season,id,lon,lat, u,v,velocity,sst)

Create Simple Feature

Because this post deals with spatial analysis, I converted the drifter dataset from data frame to simple feature and assign the dataset with WGS84 as the coordinate sysem.

drifter.sst = drifter%>%select(-c(u,v,velocity))

drifter.sst.sf = drifter.sst%>%
  st_as_sf(coords = c("lon", "lat"))%>%
  st_set_crs(4326)

Divide the Area into Grids

Because drifter observations are inhomogenous distributed within the Tropical Indian Ocean region, to obtain homogenous distribution, I divided the region into 900 grids covering the entire area.

# creat a grids of the area
grid = drifter.sst.sf%>%st_make_grid(n = 30)

Populate Grids with Drifter Observations and Median SST

Once the grids were created, I computed the statistics in each grids. The first statistic metric computed was the number of observations in grids . Then the second metric was the calculation of median SST in a grid.

Visualization

The Western Indian Ocean Region is fairly crossed with Lagrangian drifter observations (Figure 2a). Grids with drifters above 300 are found North the Madascar Island and extend to the west of the Island toward the East African coast. The largest part of the area has grids with less than 100 drifter per grid. The climatology sea surface temperature from drifters range between 25 ^oC and 30^oC with mean tempeature of 27.80 ^oC. The warmest water lies between latitude 4 to 10 ^oS (Figure 2b). This warmest region lies in the southern gyres as shown in the previous post.

Figure 2: Climatology a)Total number of drifters and b) the median SST in the Tropical Indian Ocean region

Seasonal Variation

Figure 3 show the seasonal median sea surface current within the tropical Indian ocean region. Like other studies, drifter observations reveals the northeast season has fairly higher sea surface temperature compared to the southeast monsoon season (Figure 3).

Figure 3: The sea surface current during the northeast (NE) and southeast (SE) monsoon season

Southeast Monsoon Period

All drifters crossed the region between April and September were combined to form the southeast monsoon. The pattern of drifters distribution during the climatology southeast monsoon period (Figure 4) matches closely to those in figure 2a, though the number is less. However, the climatology sea surface temperature during the southeast monsoon period does not show clear patter, even the dongue of warm water is not visibl (figure 4b)e. This is because both the East African Coastal Current and Somali CUrrent both flow northward during the southease monsoon and weaken cyclonic flow around the equator leading to disappearing of southern gyre.

Figure 4: Climatology drifter observation during the southeast monsoon season. a)Total number of drifters and b) the median SST in a grid

Northeast monsoon Period

Unlike the southeast, the climatology northeast period had few drifters throughout the region. Most of the area had drifters below 75 except some few grids west of north Madagascar Island (Figure 5a). Yet although northeast season has few drifter but the pattern of areas with large number of drifter is similar to those in figure 2a and 4. We also notice that the the pattern of sea surface temperature during the northeast (Figure 5b) differs with the southeast season (Figure 4b). The northeast monsoon is associated with warmest water above 30 ^oC along the coastal areas.

Figure 5: Climatology drifter observation during the northeast monsoon season. a)Total number of drifters and b) the median SST in a grid

Conclusion

As we have seen there are more drifters during the southeast monsoon (Figure 4a) than the northeast monsoon (Figure 5a). Drifter reveals the presence of warm temperature southern gyre (Figure 2a). Threfore, drifter observation from Global Drifter Program provides reliable in-situ data that can be used to validate and calibrate satellite-derived sea surface temperatue in the tropical Indian Ocean Region.