Many existing maps where water is prevalent or dominant don't work very well in Zero-K, especially for 1v1 or small teams.
This isn't criticism of any map developers - most of them were not designing with ZK in mind at all, let alone the most recent sea design iteration. Even for those that were, ZK sea is poorly explored, poorly understood, and perhaps more sensitive to mapmaker decisions than ZK land, so the task of making a good sea map is much more challenging.
The following are some speculative thoughts about the ideas which might lead to a well-behaved water map.
In one sense, sea maps have a less rich variety of terrain than land maps - the water surface must always be entirely flat, and there are practical limitations on what underwater terrain features are a good idea.
That being said, there is still a surprising amount of variety if you look closely enough. A lot of this variety isn't really used in current maps, since the water depths required to achieve them are fairly precise and specific to ZK.
In deep water (around depth 50 to 90) Commanders and Grizzlies are fully submerged. Something in this range ia a good baseline for a major body of water.
Significantly deeper water (depth 100+) may have unfortunate effects on amphibious combat (due to increased float times). As long as you don't go overboard this is probably fine.
Deep trenches with steep cliffs are unpathable for Amph but do not obstruct ships, submarines or hovercraft. Floating amphbots may get stuck in them, and they may be difficult to discern depending on your water settings. Limited use of trenches can be OK, flooded Titan Duel is a nice map.
Shallow water allows pathing of land units, allows Commanders to fire their weapons, and Amphbots will generally fire their land-based weaponry. Ideally it should be obvious by looking at a sea map which areas are shallow and which are not.
Water of depth 6+ can be entered by most ships, but NOT submarines or the shipfac building itself. Most land units can fire their weapons. Limited use of this might be OK. Remember that terraform can be used to make water deeper.
Water of depth 16+ can be entered by submarines and shipfac. Commanders can still fire their weapons but smaller units may struggle.
Most land units can't walk below depth 20 or so.
At depth 30-50 Commanders are unable to fire their weapons but are still vulnerable to surface weapons. This is pretty awkward so I'd advise avoiding the use of large bodies of water around this depth.
A gentle beach is usable by amph and hover. Hills near the coast are usable by amph but not hover. A coast with nearby underwater cliffs is usable by hover but not amph (without the use of Lobster). Cliffs at the water edge are only usable by Lobster and Recon commanders.
Since amph is generally supposed to have better pathing than hover, and underwater cliffs may not be easy to see, I would avoid the third option unless you have a good reason for it.
The behaviour of your shoreline should be consistent and clearly signalled to the user through textures. Indonesia is an otherwise reasonable map ruined by unreliable hover pathing on shores.
Ships and hovercraft are pretty quick so a reasonably-sized sea map is potentially a tad larger than its land-map equivalent. Somewhere around 12x12 with diagonal starts is a reasonable 1v1 map, for instance. 10x10 is likely to be on the small side. Several otherwise-reasonable existing sea maps are either tiny (Coastal, 12x4, particularly narrow) or huge (Inculta Wet, 16x12, also metal dense).
Some land maps have distinct areas where one type of factory is stronger than another. If the distinction is between vehicle-favoured and bot-favoured domains, or spider-favoured and bot-favoured domains, this is not a big issue since bots can function reasonably well on flat ground and spiders do OK on moderately hilly ground. If the distinction is between spider- and vehicle-favoured domains it's more of an issue.
EXAMPLE: Cold Snap - In 1v1 any land factory you choose is going to be very weak on one area or another of the map. It's kind of OK since all of the metal spots are accessible to all factories, but if your opponent controls the hills with spiders, and you control the flats with vehicles, many of the metal spots will be difficult for either of you to hold.||
Other maps have separate fronts where it is very difficult to transition from one front to another, because of distance or very inaccessible terrain.
EXAMPLE: Onyx Cauldron - It's pretty difficult to move from the top-right side of the map to the bottom-left.||
If a map is being played by less players on each team than it has domains, it will often feel awkward - you won't be able to affect all of the important parts of the map, and blind rock-paper-scissors in start locations and factory choice comes into play.
Both of these issues are more severe and more common in the context of sea maps. The fact that land units cannot move on water, ships cannot move on land, amphbots are slow, and hovercraft cannot move on hills results in a very large number of domains occurring very easily.
Obviously some domains don't actually matter. For instance, Charlie in the Hills has a lot of mountainous terrain but it's not generally important to control it.
The facts which make a domain matter are (1) access to resources, (2) influence over resources and (3) strategic position, in approximately that order.
Designing a map for 2v2/3v3 which has a combination of land and sea domains is a reasonable thing to do. Just be aware that in 1v1 the map will probably be amph or hover only which tends to make games boring. (e.g. Rapids is amph only, Finns Revenge is mostly hover only, Cull is amph/hover only.)
Using a combination of domains which straight-up prevent movement of ships, and domains which straight-up prevent movement of land units, should only be done with extreme caution in a map intended for 1v1. The lakes of Onyx Cauldron and the islands of Shimmershore are small enough (and contain/influence few enough mex) that land units and ships respectively can still influence them; lakes and islands of substantially larger size would probably be a big issue. For instance the land masses on the edges of Cull, which are large and contain significant quantities of mex, make it very awkward to play ships in 1v1.
Your metal layout determines which parts of the map are important to control, so it's important to bear in mind which factories your metal layout encourages, and whether this is consistent with the map as a whole.
To wit: If your map is intended for ships, but a sizeable number of metal spots are inaccessible by Mariners and/or undefendable by ships, playing 1v1 on this map is going to feel really bad. Even small teams might feel pretty awkward, depending on your layout.
The message is that if you're trying to make a tournament-feasible ship map, make sure most (preferably all) of the mex can be built by Mariners, and ships can effectively defend those mex. The mex don't necessarily have to be in the water.
It's more OK for geothermal vents and moderate numbers of reclaimable features to be less accessible - geos are less important until later (when factory switching can occur) and generally features are a minor concern.
Maps for teams
Multiple disconnected large water bodies frequently don't work well unless there are multiple people playing in each body. Game outcomes are decided on a single front and there's very little which players on other fronts can do to affect it. This is at least partly a problem with bad map choices for given game sizes, rather than purely a map design issue. Small Supreme Battlefield and Tangerine play pretty badly unless your teams are quite large for this reason.
These are some comments illustrating how I applied the above ideas to Shimmershore.
The starting area of shallow water allows Commanders to defend themselves against early raids. All areas with shallow water are clearly defined by the different texture and colourful reflection.
The large area of shallow water in the middle of the map is an interesting terrain feature; some amphbots, particularly Scallops, are much more potent here. Between the shallows and the islands Amph may have an advantage in the middle of the map.
The islands and shallows give some tactical flavour to the map and some tools for amph/hover to work with, but the map doesn't revolve around them - the mex layout directs attention towards the "neutral" areas of water, where ships are stronger. The handful of mex that aren't in the water are generally pretty defendable by ships (the mex on the large islands are the only partial exception).
All of the deep water is sufficiently deep to fully immerse Commanders and Grizzlies. There are a few somewhat deeper areas but they are not large, not too deep and the slopes don't impact on Amph movement too much.
There is no relevant part of the map that cannot be influenced by any of the water factories.