Solar · Through winter
The array works better on a clear January morning than on a hot July afternoon. The trouble is that January has fewer of those mornings, the sun sits lower when it arrives, snow can cover the panels entirely, and the battery may refuse the charge when it does come.
The paradox
Photovoltaic modules produce more as they get colder. Output rises as cell temperature falls, at a rate given by the temperature coefficient on the datasheet. A cold clear day is genuinely good conditions for an array.
That is where the good news stops. Winter brings less light, a lower sun, snow that blocks everything, batteries that will not accept charge when cold, and a voltage rise that can damage the controller.
Every other page in this section has left a winter problem hanging. This page is where they collect, because in practice they arrive together rather than one at a time.
Less light
Winter output is not simply proportional to daylight hours. Three separate effects stack, which is why the drop feels steeper than the calendar suggests.
The obvious one. The sun is above the horizon for less of the day, so there is less time in which to collect anything.
Light arriving at a shallow angle spreads across more panel area and passes through more atmosphere on the way. Both weaken it before it reaches the cells.
In much of the country winter is also the cloudiest season. This is why two sites at the same latitude can have very different December figures.
Together these are why the array was sized against the worst month rather than the annual average on the sizing page, and why the tilt was biased steeper on the mounting page. Both of those decisions were made for this season.
Snow
Not a reduced amount. Nothing. A worst month that the arithmetic treated as dim can instead contain several consecutive days of zero, and the bank has to carry all of them.
Three things decide how much this costs a property, and all of them were set when the array was mounted rather than when the snow arrived.
A steep array sheds much of what lands on it before it consolidates. The winter-biased angle chosen for light collection pays a second time here.
An array reachable from the ground with a soft brush gets cleared the morning after a storm. A roof array in January does not, and nobody should be up there attempting it.
A ground mount set low fills its own bottom edge with the snow it just shed, plus whatever drifts. Mounting height comes from local snowfall records, not from the racking catalogue.
Fresh snow is highly reflective, and bifacial modules that collect on their rear face gain from that reflected light. NREL's PVWatts version 8 added a bifacial module option and albedo inputs, so this is something you can model for your own site rather than take on faith.
The bank in winter
This is the winter failure that surprises people most, because nothing is broken. A lithium bank below roughly freezing refuses charge, and its management system is blocking it deliberately to prevent the permanent damage covered on the battery banks page.
First sun arrives at the coldest hour. A bank that spent the night below the threshold cannot use it, and by the time cells warm enough to accept charge, hours of a short day are gone.
Self-heating cells draw from the system to warm themselves, in the season the system has least to spare. That draw belongs in the daily figure that sized the array.
Discharging cold is permitted, but a cold bank delivers less than its rated capacity. Days of autonomy quietly shrink in exactly the week they are needed.
Putting the bank somewhere already heated solves all three at once and costs nothing extra to run. It is usually a better answer than heated cells, if the siting rules allow it.
Where a bank may be installed is a code question rather than a preference, and it is one of the four things worth asking your local authority about on the wiring and code page.
Voltage rise
Panel open-circuit voltage rises as temperature falls. That is the flip side of cold weather helping output, and it has a hard limit attached.
A string comfortably inside its controller's maximum input voltage on a mild afternoon can exceed that limit at dawn on the coldest morning of the year, before any load is drawn. Exceeding it destroys the controller.
This is why string length is calculated against the record low temperature for the site rather than the average, using the panel's voltage temperature coefficient. The calculation belongs to the controllers page, and it is a winter decision made in advance.
The compound week
Each problem above is manageable alone. The week that actually tests an off-grid property is the one where they land together, and the sequence is worth picturing before it happens rather than during.
A storm arrives and covers the array. Output goes to zero, not to a fraction. The bank begins carrying the whole property, and because it is cold it holds less than its rating.
Behind the storm comes clear cold air. The sun returns, low and brief, but the cells are well below freezing at first light and cannot accept a charge. If they are heated, the heater is drawing from a bank that is already low.
Meanwhile household demand is at its annual peak. Pumps run against frozen ground, heating runs constantly, and daylight hours are short enough that lighting is on for most of the waking day.
None of this is a fault. Every component is behaving exactly as designed. The system is simply meeting the conditions it was sized for, which is why the sizing pages insisted on the worst month rather than the average.
What to do
The single highest-value winter decision. Conditioned space removes the charge cutoff, the capacity loss, and the heater draw in one move, subject to what the code allows.
Sheds snow, catches a low sun, and keeps the bottom edge clear of drifts. Three winter problems answered by one mounting decision.
Clearing snow safely from the ground is worth more in December than any efficiency gain elsewhere in the system.
Decide in advance what gets shed in a hard week, and in what order. A household that has already had that conversation makes better decisions on day four than one improvising.
Sizing storage to survive the worst compound week unaided is possible and expensive. A generator running a few hours a year is often cheaper than the battery capacity it replaces.
Everything on this page is knowable in advance from local weather records and equipment datasheets. A property that treats its worst week as a specification rather than a surprise spends less and worries less than one that discovers it in January.
Next
The point where roof space, shade or winter load pushes a household into property-scale planning, and what the options are past it.
The engine side of property power, including the backup generator this page recommends considering.
Back to Solar at Property Scale, or to sizing an array for the worst-month arithmetic this page describes in practice.
Sources
Winter behaviour varies enormously between sites at the same latitude. Every figure that matters here belongs to your location and your equipment rather than to a general rule.