30 Day Map Challenge 2026 - Day 13 (Icon Map Pro) - Interactions

How do the warm Agulhas and the cold Benguela interact around South Africa, and how does that set up the fishing year?

For Day 13 of the 30 Day Map Challenge, the theme is Interactions. After Day 12's map of trade power, I went offshore. South Africa sits between two very different oceans: the Agulhas, one of the fastest warm currents in the world, runs down the east coast, while on the West Coast the Benguela brings cold water to the surface. Where and when they meet decides where the sea is cold and productive, and that shapes everything from sardine spawning to the KwaZulu-Natal Sardine Run.

The story turned out to be one of cause and effect: the wind sets the sea temperature, and life follows.

Explore the interactive report

The page has four maps side by side: January (summer), April (autumn), July (winter) and October (spring). Each shows the typical month, built from several years of monthly data.

  • Pick a map layer in the dropdown and all four maps switch together: sea temperature, sea temperature against the average at its latitude, chlorophyll, or chlorophyll against its yearly average. The legend under the dropdown switches with it.
  • Black lines are surface currents, traced through the month's mean current, with arrows pointing downstream.
  • White arrows are the monthly mean wind: the longer and thicker the arrow, the stronger the wind.
  • Magenta markers are fishery events for that season, such as anchovy spawning or the Sardine Run. Hover over one for when it happens, what happens, how it links to the ocean, how confident the sources are, and the sources themselves.
  • Show or Hide overlays with the three buttons for surface currents, wind and fishery events.
  • The chart shows the West Coast month by month: southerly wind as columns, and how much colder the coast is than its latitude as a line.

Interactive report: Day 13, Interactions

What the map doesn't show: anything below the surface. Every layer is a surface measurement, so cold water lying on the seabed under a warm surface layer, or the low-oxygen water behind lobster walkouts, doesn't appear.

Starting with sea temperature and chlorophyll

Sea temperature is NOAA OISST v2.1, the monthly mean on a 0.25° grid (about 25 km), read straight from NOAA PSL's server with xarray. I kept 1991–2025 for South Africa's waters (10–35°E, 26–42°S), which came to about 11 MB. The map shows the 2016–2025 average for each calendar month.

Chlorophyll, the green pigment in plankton, is the NOAA S-NPP VIIRS monthly science-quality product, 2018–2025. It's a 4 km grid; I kept every second cell (about 8 km), which made eight yearly files of about 3.5 MB each.

Both are monthly averages of satellite measurements, so they're measured data. Winter cloud leaves gaps in the chlorophyll far south, so a small 3×3 median filter removes the speckle, and pixels with fewer than 10 months of data are left blank on the "against its yearly average" layer.

Currents and wind from Copernicus Marine

Currents are Copernicus Marine's GlobCurrent total surface current: the geostrophic flow from satellite altimetry plus the wind-driven part, monthly at 0.25°, 2018–2025. Wind is Copernicus Marine's monthly sea-surface wind, ERA5 corrected with scatterometer data, also 0.25°. Each download was about 2.5 MB. Both need a free Copernicus Marine account, and both allow commercial use with attribution.

From a grid of arrows to traced currents

The first version drew one arrow per 0.5° cell, about 13,000 arrows across the year. It was accurate but unreadable, especially on four small maps.

So I traced streamlines instead. For each month, the script starts from the fastest water and follows the flow forwards and backwards in small steps until it reaches land, still water, the edge of the map, another line, or loops back on itself in an eddy. Lines are kept about 0.75° apart, which gives about 50 flowing lines per month instead of a thousand arrows. Because the Agulhas core is traced first, it always comes out as one continuous line, with the retroflection and the return current looping back east south of the Cape.

Each line is named by its position and direction (Agulhas Current, Agulhas Retroflection, Agulhas Return Current, Benguela Current), and the names show on the map as labels.

Finding the upwelling in the data

This was the part that surprised me. On the plain sea temperature layer, the West Coast didn't look colder in summer than in winter, even though summer is the upwelling season. The numbers explained why. Along the West Coast (31–34°S), the coastal water only warms from about 14.3 °C to 16.6 °C through the year, while the open ocean about 550 km offshore warms from about 16.3 °C to 21.4 °C. Upwelling doesn't make the coast cold in absolute terms: it stops it warming while everything around it heats up.

The fix was a new layer: each cell's temperature minus the average of all ocean at the same latitude that month. That removes the seasonal heating and leaves what the currents and wind are doing. The West Coast strip turns deep blue in summer and autumn, and the Agulhas shows red down the east coast all year.

Raster edges that follow the coast

Coarse 25 km grids give a staircase edge along the coast, with gaps in the sea and colour spilling onto land. Before each raster is drawn, the nearest ocean value is carried up to 0.5° inland, the grid is resampled to 2.5 km, and every pixel on Natural Earth's 1:10m land is cut away. The edge now follows the real coastline, and the thin strip near the shore shows the nearest real cell's value.

Four seasons instead of a play axis

The plan was one large map with a Play Axis stepping through the months. Watching it animate, you lose the comparison: you can't see January and July at the same time. Four fixed maps, one per season, make the comparison instant. I chose mid-season months (January, April, July, October) so each map sits inside its season rather than on an edge.

Each map is an Icon Map Pro visual filtered to its month, and all four share one fixed camera so they line up exactly. The raster is a Cloud Optimised GeoTIFF read straight from Azure Blob Storage. A measure picks the right file for the map's month and the chosen layer, using the dynamic-layers pattern, so one dropdown switches 48 rasters (four layers × 12 months) across four maps.

A gradient legend that switches with the layer

Four layers means four colour scales. I followed the Icon Map blog post on multi-stop gradient legends: a small table of colour stops, a DAX measure that builds an SVG gradient bar from them, and a card visual that shows the SVG as an image. My table holds all four layers, and the measure filters it to the picked layer, so the legend switches with the dropdown. The colour stops are sampled from the same colour scales the rasters were drawn with, so the legend and the map always match.

Overlay buttons with symbols

The static legend for currents, wind and events became a button slicer: three buttons, each with its own symbol, that switch an overlay off and on across all four maps. Each map row is tagged with its overlay, and a small Map overlay table connects the slicer to the map. The symbols are small SVGs stored as URL-encoded data URIs in a column with the Image URL data category, and bound to the slicer's image field in each button state, so the symbol stays when a button is hovered or selected.

Wind barbs: tried and dropped

I also tried proper meteorological wind barbs, drawn as rotated image markers. They worked, but at this scale they were hard to see, and a general reader has to know that the staff points to where the wind comes from. White arrows whose width grows with the wind speed read instantly, so the arrows came back.

From fish farms to the fishing year

The initial plan was to focus on aquaculture: where is the water cold enough for abalone, mussel and oyster farms? But the open farm data is thin. I could cite nine sites, most of them abalone farms on land, and OpenStreetMap only tags a handful. So I dropped the farms and asked a better question: what is the fishery doing in each season, and why?

I compiled six events from published research, agency reports and news, each with a confidence rating:

  • Spring: sardine and anchovy spawning on the Agulhas Bank, with anchovy peaking in mid-November; the tuna pole-line season opens, running from September to May.
  • Summer: a jet along the shelf edge carries eggs and larvae to the cold, food-rich West Coast nursery; red tides and rock lobster walkouts at Elands Bay, most recently about 500 tonnes in January–February 2026.
  • Autumn and winter: snoek move inshore on the West Coast, with the line-fishing peak from May to July.
  • Winter: the Sardine Run up the Eastern Cape and KwaZulu-Natal coast in May–July, following a strip of cool water inside the warm Agulhas.

Design choices

  • Four rasters, four colour scales. Red–yellow–blue for temperature, blue–red (centred on white) for colder or warmer than the latitude average, viridis for chlorophyll, and brown–teal for chlorophyll against its yearly average.
  • Dark currents, white wind. Near-black current lines read over every raster, white wind arrows stand apart from them, and wind width grows from about 2.5 px to 6 px with speed.
  • Magenta for life. None of the four colour scales uses magenta, so the event markers and their halo labels stand out on every layer.
  • A quiet basemap. Icon Map's dataviz basemap keeps land and water neutral, so the rasters carry the colour.
  • Labels placed by hand. Markers were nudged so their labels clear each other and the town names, and every marker is checked to be at sea.

What the map reveals

The wind comes first. Along the West Coast (31–34°S, measured over the sea out to 15°E), the southerly wind averages about 6.3 m/s in February and drops to about 0.3 m/s in June.

The cold follows a month or two later. The coastal strip is about 6.3 °C colder than the average at its latitude in March and April, and only about 3.4 °C colder in October. The lag is clear in the chart: the wind peaks in January–February, the cold in March–April.

So does the bloom. Coastal chlorophyll peaks at about 5.3 mg/m³ in February, against about 2.1 in June.

The Agulhas runs all year. Its core is the fastest water on the map, averaging about 1.3 m/s along the line, and it keeps the east coast warmer than its latitude in every season.

These are monthly averages of surface conditions over several years. They show the typical seasonal cycle, not any single year or a week of strong wind, and the fishery events are approximate in place and time.

What I learned from Day 13

1. Remove the obvious signal to see the interesting one. Seasonal heating hid the upwelling. Subtracting the latitude average made it the clearest thing on the map.

2. Side by side beats animation for comparison. A play axis shows change over time, but four fixed maps let you compare seasons at a glance.

3. Trace the flow, don't grid it. Fifty streamlines a month tell the current story better than thousands of arrows.

4. Let the data decide the angle. The aquaculture data was too thin to carry the map, and the fishery calendar was a better story anyway.

Final thoughts

Interactions is a good theme for oceans, because nothing on the map acts alone: the wind drives the current and the upwelling, the cold water feeds the plankton, and the fish follow. Putting the cause beside the effect is what makes a map like this work: the wind and the temperature on the same map, and the wind and the cooling in the same chart. The same approach works for any business with a seasonal driver and a seasonal response, such as weather and demand, rainfall and yields, or holidays and footfall: map them side by side, season by season, and the lag between them becomes visible.

Sources and attribution

  • Sea temperature: NOAA OISST v2.1 monthly mean, 0.25°, 1991–2025, via NOAA PSL. Huang, B. et al. (2021), Improvements of the Daily Optimum Interpolation Sea Surface Temperature (DOISST) Version 2.1, Journal of Climate 34, 2923–2939. Public domain. Downloaded 11 November 2026.
  • Chlorophyll: NOAA CoastWatch, S-NPP VIIRS chlorophyll-a, Science Quality, monthly (nesdisVHNSQchlaMonthly), 2018–2025. Free and open. Downloaded 11 November 2026.
  • Currents: E.U. Copernicus Marine Service Information, Global Total Surface and 15 m Current (MULTIOBS_GLO_PHY_MYNRT_015_003, dataset cmems_obs-mob_glo_phy-cur_my_0.25deg_P1M-m), 2018–2025. Free, including commercial use, with attribution. Downloaded 11 November 2026.
  • Wind: E.U. Copernicus Marine Service Information, Global Ocean Monthly Mean Sea Surface Wind (WIND_GLO_PHY_CLIMATE_L4_MY_012_003, dataset cmems_obs-wind_glo_phy_my_l4_P1M), 2018–2025. Free, including commercial use, with attribution. Downloaded 11 November 2026.
  • Land: Natural Earth 1:10m land polygons, public domain.
  • Fishery events: compiled by hand from published sources, including SARP research on sardine and anchovy spawning (African Journal of Marine Science), Veitch et al. (2009) in Progress in Oceanography, Griffiths (2002) in Fishery Bulletin 100, the IOTC South Africa national report (2012), Fréon et al. on the KwaZulu-Natal Sardine Run, and Cape Argus/IOL and EWN reports on the 2026 lobster walkouts. Every event lists its sources in the map tooltip.

Methodology note: All layers are monthly climatologies of surface conditions: sea temperature is the 2016–2025 mean for each calendar month, and chlorophyll, currents and wind are 2018–2025 means. "Sea temperature against its latitude average" is each 0.25° cell minus the mean of all ocean cells at the same latitude that month. "Chlorophyll against its yearly average" is each pixel's month divided by its own mean of the 12 months on a log scale, and is left blank where a pixel has fewer than 10 months. Current lines are evenly spaced streamlines (about 0.75° apart) traced through each month's mean current and cut against Natural Earth land. Wind arrows are drawn every 2° north of 37°S for winds of at least 1 m/s. The West Coast chart averages the wind over the sea at 31–34°S, 15–18.5°E, and the temperature over the three ocean cells nearest the coast in each 0.25° row. Raster edges carry the nearest ocean value up to 0.5° inland before land is cut away at 2.5 km. Fishery events are approximate places and typical timings, mostly from research of the 1990s to 2002; the Sardine Run now arrives later and the snoek run has declined.