Panama: when a single lake sets the price of global trade.

Optical baseline. Satellite optical imagery photographs Gatun Lake and the upstream Alajuela (Madden) reservoir whenever the sky is clear: down to roughly 10 m ground sampling on the visible and near-infrared bands, refreshed every few days. As the lake falls, the shoreline retreats. Mudflats and exposed lakebed show up where there was water just weeks before, and that receding edge is basically a direct, visual measurement of stored volume.

Water-extent mapping, or turning pixels into surface area. A water-extent index contrasts the green and near-infrared bands so water goes strongly positive and land goes negative: the formula McFeeters published in 1996 as the Normalized Difference Water Index. A later refinement, Xu's 2006 Modified NDWI, swaps in a shortwave-infrared band instead, which suppresses false positives from built-up land and better separates a muddy, receding shoreline from the water itself. Threshold it per pixel and you get a clean water mask. Count the masked pixels, multiply by pixel area, and out comes lake surface area in square kilometres, tracked image by image. Pair that time series with the reservoir's known depth-area-volume curve and the flat 2-D surface turns into a volume estimate.

Drought context. A dry month and a structural decline can look similar from a single reading. Standardised drought indices, precipitation anomalies and vegetation-condition indices over the watershed, place that single low reading in its multi-year context, so the two aren't confused. In 2023, Gatun sat near 80 feet through the back half of the year, against a more typical ~88 feet at the close of the wet season.

Why radar works here. All-weather radar, synthetic-aperture radar, emits its own microwave pulses and measures the echo, so it images day or night, straight through cloud and rain. A steel hull happens to be a strong corner reflector. Microwaves bounce off the metal and the deck structure and come back as a bright point against the radar-dark sea, so ships light up as discrete targets even when the optical scene is nothing but overcast.

Counting the queue. Detect and count those bright targets in the anchorage off Balboa (Pacific) and Cristobal/Colon (Atlantic) and you get a vessel count at anchor. Optical imagery from August 2023 shows the same signal: dozens of ships parked offshore. Track how the count rises and falls week to week and you have an instant, independent supply-chain throughput indicator. A growing anchorage means congestion is building, and it shows up before any official advisory confirms it.

Why the trees matter. Porous, forested soil absorbs heavy tropical rain and releases it slowly through the dry season, smoothing the flow into the reservoir. Strip the forest, though, and bare soil sheds rain in fast, muddy pulses. The water floods through and runs to sea instead of being banked. And the runoff carries sediment that silts up the lake, eroding the very storage capacity the canal depends on.

What EO monitors. Optical time series flag deforestation and land-clearing as they happen. A vegetation index, the near-infrared-minus-red contrast that scales with healthy green biomass, tracks canopy vigour and stress across the watershed. So falling vegetation vigour over a slope becomes an early flag, first for erosion potential, then further downstream for reservoir sedimentation.

Measuring water height from orbit. Radar and laser altimeters fire a pulse straight down and time the return to fix the water-surface height. Reference-grade radar altimetry, and newer wide-swath interferometric altimetry that maps a water body in two dimensions rather than along a single ground track, can resolve inland water-surface elevation to the centimetre class. That is an actual lake level, not a surface outline.

The limit worth knowing. Centimetre accuracy is real, but it only holds for water bodies large enough to fill the sensor's footprint cleanly. Published validation work puts uncertainty at roughly 8–10 cm for large lakes, widening toward the decimetre range for small or narrow ones, and reference-mission repeat cycles typically run 10–27 days. The NASA/CNES SWOT mission, launched in December 2022, pushes this further with a Ka-band radar interferometer imaging swaths up to 120 km wide rather than a single ground track: published validation puts its water-surface-elevation error at roughly 10 cm for non-vegetated water bodies larger than 1 km², degrading sharply below about 0.5 km². Gatun Lake, at roughly 425 km², sits comfortably inside that reliable range. But a single altimetry pass won't catch a sudden lock-release or a flash-flood pulse between overpasses, which is why a level series is best fused with the optical shoreline record and the watershed water balance rather than read from altimetry alone.

Why centimetres are money. The Panama Canal Authority sets a maximum permissible draft against the available depth over the lock sills. In the 2023 drought the Neopanamax draft limit was cut from the normal 15.24 m (50 ft) toward 13.41 m (44 ft): nearly two metres of cargo a vessel could no longer carry through. A bulk or container operator can convert that lake level directly into a loadable draft, and from there into a cargo mass or TEU ceiling, weeks before the ship even reaches the lock.

Ship-type identification. A vessel's radar signature, how strongly it reflects radar, scales with its size and superstructure, and its polarimetric response differs by hull form. All-weather radar can transmit and receive in dual polarisation, and the ratio and texture of those channels, fed to a machine-learning classifier, is often enough to separate a bulk carrier from a container ship from a tanker through complete cloud cover. No optical view required.

Dark vessels via radar and AIS fusion. Most ships broadcast their identity and position over AIS. Fuse the radar detections with the AIS feed, and what gets interesting are the mismatches. A hard radar target with no matching AIS signal is what the industry calls a dark vessel, a ship that has switched off or is spoofing its transponder. Radar sees the physical object regardless of what the transponder claims.

Radar interferometry for the structures. Comparing the phase of repeat radar passes measures ground and structure displacement at millimetre-to-centimetre scale. Over the lock chambers, approach walls and surrounding terrain, radar deformation analysis can flag subsidence and slow movement: an early, remote check on the stability of the fixed infrastructure itself.

Soil moisture. Microwave soil-moisture radiometry measures the moisture in the top layer of soil across the watershed. Dry, parched soil absorbs the first rains rather than passing them downstream. Soil already near saturation just lets the rain run off into the lake. So soil-moisture state ends up being a leading variable for how effective the coming wet season will actually be at refilling Gatun.

Gravimetry for the macro deficit. Satellite gravimetry detects tiny changes in Earth's gravity field caused by shifts in total water mass: surface water, soil moisture and groundwater combined. The NASA/DLR GRACE and GRACE-FO missions, which measure this by timing the changing distance between a twin pair of satellites, resolve it natively at a coarse ~300 km scale: basins on the order of 60,000–250,000 km² are what the raw signal can carry with useful accuracy, so a single small watershed like Panama's isn't resolved directly. At regional scale, though, this reveals a cumulative water deficit or surplus that a single lake gauge simply can't, putting one reservoir's level inside the basin-wide water budget.

The choice. A vessel can join the queue and wait. It can bid for a priority slot at auction, which during the 2023 drought hit a reported US$4 million for a single passage (officially ~US$3.98M, paid by Japan's Eneos Group for an LPG carrier on 8 November 2023, on top of standard tolls). Or it can reroute entirely, around Cape Horn or the Cape of Good Hope, trading days at sea and extra fuel and CO2 for certainty of passage. For a major line such as Maersk or MSC, the difference runs to millions of dollars per ship and weeks of schedule.

EO as a predictive input. Lake level, soil moisture, anchorage counts and transit cadence are the live inputs to a supply-chain digital twin, a running simulation of the network. Feed the orbital signals in, and the twin can model, roughly 30 days ahead, the fuel cost, CO2 and time-loss of waiting versus rerouting for a given vessel and cargo. That's what turns a gut call into a costed decision.

What it is. The centre is a regional hub that brings Earth-observation data and services to governments across a region highly exposed to climate-driven hazards. Its mandate is climate resilience, so the point is turning satellite signals into decisions rather than piling them up as archives.

Why it matters for the canal story. It is built as an operational early-warning capability. Among its thematic services are dedicated drought and wildfire monitoring alongside flood mapping, drawing on the same satellite signals described above and feeding regional disaster-risk-reduction decisions. The hydrology that governs the canal sits inside exactly the drought-and-forest signals such a hub is designed to watch.