Why the Fastest River Crossing Isn’t Always the Shortest

Two boats with identical motors cross a 120 m wide river. One reaches the far bank in 24 seconds but travels almost 140 m relative to the shore. The other takes 30 seconds and travels exactly 120 m.

Which boat do you think pointed straight across the river?

The first one. It made the fastest progress towards the opposite bank, but the current carried it downstream, making its actual path longer.

The fastest river crossing minimises time. The shortest river crossing minimises the distance travelled along the boat’s path. Those aren’t necessarily the same thing.

To cross as quickly as possible, the boat should direct all its available speed through the water towards the far bank. To travel the shortest distance relative to the riverbanks, it may need to point upstream, sacrificing some forward speed to cancel the current.

The distinction comes down to what we’re measuring, which observer is measuring it, and what the boat is actually trying to achieve.

01A river-crossing experiment: what does the stopwatch miss?

Consider a proposed experiment to compare two crossing strategies. This is an idealised investigation, not a report of measurements collected from a real river.

We need:

• A straight river with parallel banks, separated by a known distance.
• A motorboat capable of maintaining a constant speed relative to the water.
• Two markers, A and P, directly opposite each other on the near and far banks.
• A floating marker that drifts with the current.
• A stopwatch to measure the time between leaving A and reaching the far bank.

The boat starts at A. The floating marker moves downstream with the water, while the two bank markers stay fixed relative to the ground.

We’ll take downstream as the positive \(x\)-direction and straight across the river as the positive \(y\)-direction. The starting point is \((0,0)\), and P is directly across from it.

For the first trial, the skipper points the bow perpendicular to the banks. For the second, the skipper points upstream so that the boat’s motion relative to the ground is directly towards P.

Predict the result before we calculate it. Which trial should give the smaller time?

The first trial should be faster. The boat uses its entire speed through the water to move across the river. In the second trial, some of that speed is directed upstream.

But the stopwatch alone can’t establish which route covers the smaller distance. It measures elapsed time, not the length of the track.

Two observers can describe the same crossing differently

An observer standing on the riverbank sees the boat move across the river while drifting downstream.

An observer floating with the water sees the boat move relative to the surrounding water. If the skipper points straight across, the floating observer sees the boat travelling straight across.

Neither observer is wrong. They’re measuring motion relative to different reference frames.

We’ll use three labels:

The velocity of the boat relative to the ground is the vector sum of its velocity relative to the water and the current’s velocity relative to the ground:

\[ \vec v_{B/G} = \vec v_{B/W}+\vec v_{W/G} \]

Each velocity is measured in metres per second (m/s).

The arrows above the symbols mean these quantities are vectors. Each has both a magnitude and a direction.

Their magnitudes are scalar speeds. For example, a boat moving at 5.0 m/s relative to the water in a 3.0 m/s current doesn’t automatically have a ground speed of 8.0 m/s. We must first add the velocity vectors.

The river current can change the direction of the boat’s ground velocity without changing its speed through the water.

That’s the missing piece in the stopwatch experiment.

02Why the fastest crossing uses a straight-across heading

First, consider a uniform current flowing parallel to the riverbanks.

Let:

• \(b\) be the boat’s speed relative to the water, in m/s.
• \(c\) be the current’s downstream speed relative to the ground, in m/s.
• \(d\) be the river’s width, in metres.
• \(\theta\) be the angle the boat points upstream from the direction straight across.

The current acts only in the downstream \(x\)-direction.

The boat’s velocity through the water has two components:

• \(b\sin\theta\) upstream.
• \(b\cos\theta\) towards the opposite bank.

Adding the current gives the ground-velocity components:

\[ \begin{aligned} v_x&=c-b\sin\theta\\ v_y&=b\cos\theta \end{aligned} \]

Here, \(v_x\) is positive downstream, and \(v_y\) is positive towards the far bank. Both are measured relative to the ground, in m/s.

Only the across-river component, \(v_y\), determines how quickly the boat reaches the opposite bank.

For a constant heading with \(v_y>0\), the crossing time is

\[ t=\frac{d}{b\cos\theta} \]

where \(t\) is measured in seconds.

The largest possible value of \(\cos\theta\) is 1, occurring when \(\theta=0^\circ\).

So the minimum crossing time is

\[ t_{\min}=\frac{d}{b} \]

This is the fastest way to reach any point on the opposite bank under our assumptions.

Notice what isn’t in that equation: the current speed.

A current parallel to the banks changes where the boat lands, but it doesn’t change the boat’s across-river velocity when the skipper maintains a particular heading and speed through the water.

03How tracking the boat improves the experiment

The original experiment had a weakness.

A stopwatch tells us the crossing time, and the landing marker tells us the net displacement. If the ground velocity stays constant, those measurements are enough to calculate the length of a straight ground track.

But we haven’t actually established that the boat followed a straight line.

What if the skipper changes heading halfway across? Or the current changes direction across the river?

The boat could drift downstream, recover upstream, and arrive near P. The start and finish positions alone would hide the extra distance.

A revised measurement design

Keep the two fixed bank markers, the stopwatch, and the boat-speed measurement.

Now add a camera fixed relative to the ground, ideally with an overhead view from a bridge. Calibrate its position measurements using known stationary markers.

The camera records the boat’s coordinates at regular time intervals.

The measurement sequence is:

1. Record the boat at A when the clock starts.
2. Record successive positions as it crosses the river.
3. Record the final position when it reaches the far bank.
4. Calculate the distance between each pair of successive positions.
5. Add those distances to estimate the total ground-path length.

Separate tracking of freely drifting markers helps estimate the local current. A heading record and a calibrated through-water speed measurement help distinguish changes in the boat’s propulsion direction from changes in the current.

The revised design tests something the original couldn’t: whether the boat’s actual track agrees with the constant-velocity prediction.

It also exposes a common calculation mistake.

What the investigation can and cannot establish

All the numerical results above are predictions from an idealised model, not collected experimental observations.

If actual tracked trajectories agreed with these predictions, they would support the model under the tested conditions. They wouldn’t prove that every river has a uniform current or that a real boat always follows its bow direction exactly.

Real crossings can involve non-uniform currents, changing speed, wind, steering corrections, and acceleration during turns.

Our calculations assume two-dimensional motion, negligible wind and sideways slip, and a known speed through the water. Where we use straight-line formulas, velocity is constant during that segment.

Those assumptions are especially important when a question asks for distance rather than displacement.

04The decision that prevents most river-crossing errors

The word “shortest” needs an object and a reference frame.

Are we minimising the boat’s ground-path length? The distance relative to moving water? The time to touch any part of the far bank? Or the time to reach a particular jetty?

These lead to different decisions.

For a steady current weaker than the boat’s speed through the water, a perpendicular ground track is achievable by steering upstream.

If the current is too strong, that perpendicular track may not be possible at all. The boat can still cross, but it cannot necessarily land directly opposite.

The upstream heading calculation is developed separately in How to Choose a Boat Heading to Land Directly Opposite.

There’s also an important contrasting case: when the destination is a fixed point, rather than any part of the far bank, the quickest strategy may be different from the fastest bank-to-bank crossing.

For a steady uniform current and a fixed reachable destination, a constant heading can give the shortest possible travel time to that destination. Taking an unnecessary detour doesn’t help.

The practice questions below build towards that distinction, then test what happens when the assumptions change.

05Questions and solutions

All scenarios are illustrative. Unless stated otherwise, use \(+x\) downstream, \(+y\) towards the far bank, and measure ground-path lengths relative to the stationary riverbanks.

Question 1

A boat travels at 4.0 m/s relative to the water across an 80 m wide river. The current flows downstream at 3.0 m/s.

The boat points straight across throughout the journey.

Calculate:

a) The crossing time.

b) The downstream displacement.

c) The boat’s ground speed and actual ground-path length.

Solution 1

The boat crosses in 20.0 s, lands 60.0 m downstream, and travels 100 m relative to the shore at a ground speed of 5.0 m/s.

For part (a), the boat’s entire through-water speed acts across the river:

\[ \begin{aligned} t&=\frac{d}{b}\\ &=\frac{80\,\mathrm{m}}{4.0\,\mathrm{m/s}}\\ &=20.0\,\mathrm{s} \end{aligned} \]

For part (b), the current carries it downstream throughout those 20.0 s:

\[ \begin{aligned} x&=ct\\ &=3.0(20.0)\\ &=60.0\,\mathrm{m} \end{aligned} \]

For part (c), the perpendicular ground-velocity components are 3.0 m/s downstream and 4.0 m/s across.

Therefore,

\[ \begin{aligned} |\vec v_{B/G}| &=\sqrt{3.0^2+4.0^2}\\ &=5.0\,\mathrm{m/s} \end{aligned} \]

The ground-path length is

\[ \begin{aligned} L&=|\vec v_{B/G}|t\\ &=5.0(20.0)\\ &=100\,\mathrm{m} \end{aligned} \]

A tempting mistake is to report 80 m as the distance travelled. That’s the river width and the across-river displacement, not the length of the diagonal ground track.

Question 2

A 90 m wide river flows east at 4.0 m/s. A boat can maintain 5.0 m/s relative to the water.

Two skippers propose different strategies:

• Plan N: Point the bow due north throughout.
• Plan P: Choose an upstream heading so that the boat’s ground track runs directly north to a marker opposite the start.

a) Calculate the crossing time for each plan and the required heading for Plan P.

b) Calculate the distance the boat travels under each plan, first in the riverbank frame and then in the uniformly moving water frame.

c) A person on the bank says Plan P has the shorter path. A person drifting with the water says Plan N has the shorter path. Explain how both can be correct.

Solution 2

Plan N crosses faster, in 18.0 s, and has the shorter water-frame path. Plan P takes 30.0 s but has the shorter ground-frame path. Both observers are correct because distance travelled depends on the reference frame used.

For part (a), Plan N uses the full 5.0 m/s across the river:

\[ \begin{aligned} t_N&=\frac{90}{5.0}\\ &=18.0\,\mathrm{s} \end{aligned} \]

The downstream drift is

\[ x_N=4.0(18.0)=72.0\,\mathrm{m} \]

For Plan P, the boat must supply a westward component of 4.0 m/s to cancel the eastward current.

\[ \begin{aligned} 5.0\sin\theta&=4.0\\ \theta&=\sin^{-1}(0.800)\\ &=53.1^\circ \end{aligned} \]

The skipper must point \(53.1^\circ\) west of north.

The remaining northward ground speed is

\[ v_y=\sqrt{5.0^2-4.0^2} =3.0\,\mathrm{m/s} \]

So,

\[ \begin{aligned} t_P&=\frac{90}{3.0}\\ &=30.0\,\mathrm{s} \end{aligned} \]

For part (b), the ground-path length of Plan N is

\[ \begin{aligned} L_{N,G}&=\sqrt{90^2+72^2}\\ &=115.3\,\mathrm{m} \end{aligned} \]

Plan P travels directly north relative to the ground, giving

\[ L_{P,G}=90.0\,\mathrm{m} \]

Now switch to the water frame.

Because the current is uniform, a drifting observer moves with a consistent water reference frame. In that frame, the boat travels at 5.0 m/s for each journey.

\[ \begin{aligned} L_{N,W}&=5.0(18.0)=90.0\,\mathrm{m}\\ L_{P,W}&=5.0(30.0)=150.0\,\mathrm{m} \end{aligned} \]

The results are:

For part (c), the disagreement comes from comparing distances measured in different frames.

The bank observer sees Plan P travel directly between the two fixed bank markers. The drifting observer sees Plan N move directly across the moving water.

There’s no contradiction. Neither observer is entitled to replace the other’s measurement with their own without specifying the frame.

The tempting route is to treat “distance travelled” as one universal number. In a relative-motion problem, we must say relative to what.

Question 3

A boat has a maximum speed of 5.0 m/s relative to the water. A river is 120 m wide, and its uniform current flows downstream at 3.0 m/s.

The boat must travel from A to P, directly opposite A.

One skipper suggests this strategy:

“Point straight across for the first 12.0 seconds to make fast progress, then turn upstream to recover the drift.”

The boat may change heading after 12.0 s but must not exceed its speed limit relative to the water.

a) Determine the boat’s position after the first 12.0 s.

b) Calculate the minimum possible remaining time to reach P, and hence the total time for the proposed strategy.

c) Determine the minimum possible time from A to P if the skipper chooses the best heading from the start. Explain why a route involving turns cannot beat it.

Solution 3

The proposed detour takes at least 37.5 s, whereas a constant upstream heading reaches P in 30.0 s. Crossing as quickly as possible at first doesn’t guarantee the quickest arrival at a fixed destination.

For part (a), the first 12.0 s produces

\[ \begin{aligned} x&=3.0(12.0)=36.0\,\mathrm{m}\\ y&=5.0(12.0)=60.0\,\mathrm{m} \end{aligned} \]

The boat reaches \((36.0,60.0)\) m, with 60.0 m of river still to cross.

For part (b), let the remaining journey take \(\tau\) seconds.

During that time, the current adds another \(3.0\tau\) metres of downstream displacement.

To finish at P, the boat must therefore supply the following displacements relative to the water:

• \(36.0+3.0\tau\) metres upstream.
• 60.0 metres across.

The magnitude of the required water-frame displacement cannot exceed \(5.0\tau\) metres, because the boat can travel at no more than 5.0 m/s through the water.

Using metres for displacement and seconds for \(\tau\),

\[ (36+3\tau)^2+60^2 \le25\tau^2 \]

Expanding and simplifying,

\[ \begin{aligned} 16\tau^2-216\tau-4896&\ge0\\ 2\tau^2-27\tau-612&\ge0 \end{aligned} \]

The positive boundary of this inequality is

\[ \begin{aligned} \tau&=\frac{27+\sqrt{5625}}{4}\\ &=\frac{27+75}{4}\\ &=25.5\,\mathrm{s} \end{aligned} \]

This remaining time is achievable by steering at one constant heading relative to the water, directly along the required water-frame displacement, at 5.0 m/s.

Therefore, the minimum total time after the initial decision is

\[ t=12.0+25.5=37.5\,\mathrm{s} \]

For part (c), suppose the entire journey takes \(T\) seconds.

To land at P, the boat must generate a total water-relative displacement of \(3.0T\) metres upstream and 120 m across.

Again, the magnitude cannot exceed \(5.0T\):

\[ \begin{aligned} 9T^2+120^2&\le25T^2\\ 16T^2&\ge14400\\ T&\ge30.0\,\mathrm{s} \end{aligned} \]

The lower bound is achievable: maintain a boat-through-water velocity of 3.0 m/s upstream and 4.0 m/s across.

Its magnitude is

\[ \sqrt{3.0^2+4.0^2} =5.0\,\mathrm{m/s} \]

The current cancels the upstream component, leaving 4.0 m/s directly towards P.

So the crossing takes \(120/4.0=30.0\) s.

A detour cannot improve this because the magnitude of the net displacement through the moving water is never greater than the distance the boat can travel through that water. Equality requires a straight water-frame route at full speed.

The tempting argument maximises early northward progress while ignoring the downstream displacement that must later be corrected.

Question 4

A river is 100 m wide. Its uniform downstream current is 5.0 m/s, exactly equal to a boat’s maximum speed relative to the water.

Two possible headings produce the following boat-through-water velocity components:

• Heading A: 4.0 m/s upstream and 3.0 m/s across.
• Heading B: 4.8 m/s upstream and 1.4 m/s across.

Both component pairs have a magnitude of 5.0 m/s.

a) Calculate the crossing time, downstream displacement, and ground-path length for both headings.

b) Can the boat reach the point directly opposite its starting position in any finite time, even if it changes heading?

c) Can the skipper make the ground-path length arbitrarily close to the river width? Explain what happens to the crossing time.

Solution 4

Neither heading reaches the directly opposite point. Heading B produces the shorter ground path, 101.0 m, but takes 71.4 s instead of Heading A’s 33.3 s. No finite-time crossing can travel exactly 100 m in the ground frame, although paths can approach that length arbitrarily closely.

For part (a), Heading A has ground-velocity components

\[ \begin{aligned} v_x&=5.0-4.0=1.0\,\mathrm{m/s}\\ v_y&=3.0\,\mathrm{m/s} \end{aligned} \]

Its crossing time is

\[ t_A=\frac{100}{3.0}=33.3\,\mathrm{s} \]

It drifts downstream by

\[ x_A=1.0\left(\frac{100}{3}\right) =33.3\,\mathrm{m} \]

Its ground-path length is

\[ \begin{aligned} L_A&=\sqrt{100^2+(100/3)^2}\\ &=105.4\,\mathrm{m} \end{aligned} \]

For Heading B, the ground-velocity components are

\[ \begin{aligned} v_x&=5.0-4.8=0.2\,\mathrm{m/s}\\ v_y&=1.4\,\mathrm{m/s} \end{aligned} \]

The crossing time is

\[ \begin{aligned} t_B&=\frac{100}{1.4}\\ &=71.4\,\mathrm{s} \end{aligned} \]

The downstream drift is

\[ \begin{aligned} x_B&=0.2\left(\frac{100}{1.4}\right)\\ &=14.3\,\mathrm{m} \end{aligned} \]

The ground-path length is

\[ \begin{aligned} L_B&=\sqrt{100^2+(100/7)^2}\\ &=101.0\,\mathrm{m} \end{aligned} \]

Heading B produces a path closer to the minimum possible geometric distance, but the boat sacrifices much more across-river speed.

For part (b), the boat cannot reach directly opposite in finite time.

At every instant, the largest possible upstream component of its velocity through the water is 5.0 m/s.

But using all 5.0 m/s upstream leaves no component towards the far bank.

Whenever the boat makes positive across-river progress, its upstream component must be less than 5.0 m/s. The current therefore produces a positive downstream ground-velocity component.

Changing heading cannot reverse that downstream drift, because the boat can never generate an upstream ground velocity.

For part (c), let \(\theta\) approach \(90^\circ\) upstream from straight across.

The downstream and across-river components are

\[ \begin{aligned} v_x&=5(1-\sin\theta)\\ v_y&=5\cos\theta \end{aligned} \]

For a constant heading, the downstream displacement at the far bank is

\[ x=\frac{100(1-\sin\theta)}{\cos\theta} \]

Using a trigonometric identity,

\[ x=\frac{100\cos\theta}{1+\sin\theta} \]

As \(\theta\) approaches \(90^\circ\), this displacement approaches zero.

The ground-path length,

\[ L=\sqrt{100^2+x^2} \]

therefore approaches 100 m.

But the crossing time is

\[ t=\frac{100}{5\cos\theta} \]

As the heading approaches directly upstream, \(\cos\theta\) approaches zero, so the crossing time grows without bound.

The surprising result is that there is no achievable finite-time shortest path. Every finite crossing has a ground-path length greater than 100 m, and the skipper can get closer to 100 m only by accepting an increasingly long crossing time.

The tempting answer is to say that the shortest route must be exactly perpendicular because the banks are 100 m apart. That’s the geometric lower bound, not necessarily a physically achievable route.

Question 5

A 120 m wide river has two regions of steady current.

• From the near bank to the halfway line, 60 m across, the water flows east at 3.0 m/s.
• From the halfway line to the far bank, the water flows west at 3.0 m/s.

A boat can travel at a maximum speed of 4.0 m/s relative to the local water. The skipper may change heading instantly on entering the second region.

The boat starts at A, and the destination P is directly north of A on the far bank.

Two skippers disagree:

• Skipper 1 claims that keeping the bow due north throughout must give the fastest crossing, but cannot land the boat at P.
• Skipper 2 claims that reaching P requires cancelling the current separately in each half, so the shortest path must also be the fastest.

a) If the bow points due north throughout, determine the halfway position, the final landing position, the crossing time, and the actual ground-path length.

b) Find headings that allow the boat to follow a perfectly straight ground path from A to P. Calculate its crossing time and path length.

c) Prove which strategy achieves the absolute minimum possible time to P and which achieves the absolute minimum possible ground-path length, even if the skipper is allowed to vary the heading.

d) For the fastest strategy, calculate the magnitude of average ground velocity and the average ground speed. Explain why these values differ despite the boat landing directly opposite its starting position.

Assume straight, parallel banks, no wind, and no change in the boat’s available through-water speed at the current boundary.

Solution 5

Keeping the bow due north throughout is the fastest possible way to reach P, taking 30.0 s, but its ground path is 150 m long. Cancelling the current in each half produces the shortest possible ground path, 120 m, but takes approximately 45.4 s.

Both skippers are partly wrong because they’re assuming that landing directly opposite tells us what happened along the way.

For part (a), the boat points due north, so its velocity through the local water is 4.0 m/s north.

In the first region, the ground-velocity components are

\[ (v_x,v_y)=(3.0,4.0)\,\mathrm{m/s} \]

The time to reach the halfway line is

\[ \begin{aligned} t_1&=\frac{60}{4.0}\\ &=15.0\,\mathrm{s} \end{aligned} \]

Its eastward displacement is

\[ x_1=3.0(15.0)=45.0\,\mathrm{m} \]

So the halfway position is 45.0 m east and 60.0 m north of A.

In the second region, the current reverses. The boat’s ground velocity becomes

\[ (v_x,v_y)=(-3.0,4.0)\,\mathrm{m/s} \]

It takes another 15.0 s to cross the remaining 60 m.

Its second eastward displacement is negative:

\[ x_2=-3.0(15.0)=-45.0\,\mathrm{m} \]

The total eastward displacement is

\[ x_{\text{total}}=45.0-45.0=0 \]

The boat therefore lands exactly at P after

\[ t_{\text{total}}=15.0+15.0=30.0\,\mathrm{s} \]

But its ground track has two diagonal segments.

Each has length

\[ \begin{aligned} L_1&=\sqrt{60^2+45^2}\\ &=75.0\,\mathrm{m} \end{aligned} \]

The second has the same length.

Therefore,

\[ L_{\text{total}}=75.0+75.0 =150.0\,\mathrm{m} \]

The boat has travelled 150 m relative to the riverbanks, even though its net displacement is exactly 120 m north.

For part (b), a perfectly straight ground path requires zero east-west ground velocity in each region.

In the first region, the boat must provide 3.0 m/s west through the water to cancel the eastward current.

Its remaining northward component is

\[ \begin{aligned} v_y&=\sqrt{4.0^2-3.0^2}\\ &=\sqrt7\\ &=2.65\,\mathrm{m/s} \end{aligned} \]

The required upstream heading is

\[ \begin{aligned} \theta&=\sin^{-1}\left(\frac{3.0}{4.0}\right)\\ &=48.6^\circ \end{aligned} \]

The skipper must aim \(48.6^\circ\) west of north in the first region.

In the second region, the current flows west. The skipper must switch to \(48.6^\circ\) east of north.

The ground velocity stays directly north in both regions, at \(\sqrt7\) m/s.

Thus,

\[ \begin{aligned} t_{\text{straight}} &=\frac{120}{\sqrt7}\\ &=45.4\,\mathrm{s} \end{aligned} \]

The ground-path length is exactly

\[ L_{\text{straight}}=120\,\mathrm{m} \]

For part (c), we must prove the strategies are optimal, rather than merely compare these two attempts.

First consider crossing time.

The current is entirely east-west in both regions. It contributes no northward velocity.

The maximum possible northward velocity of the boat is therefore 4.0 m/s, achieved only when its through-water velocity points due north.

Every crossing must satisfy

\[ \begin{aligned} t&\ge\frac{120\,\mathrm{m}}{4.0\,\mathrm{m/s}}\\ &\ge30.0\,\mathrm{s} \end{aligned} \]

Skipper 1 achieves that lower bound while also arriving at P because the opposite currents produce equal and opposite displacements.

No steering strategy can reach P sooner.

Now consider ground-path length.

The shortest possible geometric distance from A to P is 120 m. Any path containing a sideways deviation has a length at least as great as the straight-line distance between its endpoints.

Skipper 2 achieves this lower bound by cancelling the current separately in each region.

So Skipper 2’s route is globally shortest.

The important distinction is that the fastest path and shortest path both finish at P. Same starting point, same destination, different distances travelled.

For part (d), the fastest route’s net displacement is 120 m north over 30.0 s.

Its average ground velocity is

\[ \begin{aligned} \vec v_{\text{avg}} &=\frac{\Delta\vec r}{t}\\ &=(0,4.0)\,\mathrm{m/s} \end{aligned} \]

So its average ground-velocity magnitude is 4.0 m/s, directed north.

But its average ground speed uses the actual path length:

\[ \begin{aligned} s_{\text{avg}} &=\frac{L_{\text{total}}}{t}\\ &=\frac{150\,\mathrm{m}}{30.0\,\mathrm{s}}\\ &=5.0\,\mathrm{m/s} \end{aligned} \]

The values differ because the boat follows two diagonal segments.

Its eastward and westward displacements cancel in the vector average, but the lengths of those segments add when calculating distance.

A bank observer who recorded only the starting position, finishing position, and time could calculate the average velocity. Those measurements alone wouldn’t reveal the 150 m path length.

This also shows why a single water reference frame isn’t appropriate for the whole crossing: the water moves in opposite directions in the two regions. We must add the boat’s velocity relative to the local water separately in each region.

06From river crossings to navigation problems

The useful question isn’t simply “Which direction should the boat point?”

It’s “What condition must the journey satisfy?”

If the goal is to reach any part of the far bank, maximise the across-river velocity component. If the goal is to minimise ground distance, examine the actual trajectory. If the goal is to reach a fixed destination, satisfy both displacement components, and don’t assume that the quickest bank crossing also gets you to the required point.

The same reasoning applies to aircraft travelling through moving air. The aircraft’s heading, its velocity relative to the air, and its ground track can all point in different directions.

That leads naturally to How to Calculate an Aircraft’s Required Heading from Wind Data, where choosing the correct reference frame determines whether an aircraft follows its intended route or drifts away from it.