How Solar Panels Perform on Cloudy Days

A photovoltaic array responds to light, not to blue sky. On an overcast day the modules still see the sky, still receive photons, and still produce a voltage. The output is lower than on a clear day, often substantially, but the system has not stopped working. Most of what looks like a failure in cloudy weather is the array behaving as the physics requires.

This page explains what changes under cloud, what does not, and how to tell a normal weather loss from a fault in the monitoring data. It does not cover net metering, interconnection, or export compensation; those rules are set by state and by utility, they change, and nothing here should be read as a statement about any particular jurisdiction.

Direct and diffuse irradiance

On a clear day, most of the light reaching a module is direct irradiance: sunlight arriving in a straight line from the solar disc. A module’s output is roughly proportional to the irradiance in its plane, so a clear noon in summer delivers the highest current the array will see.

Under cloud, the direct beam is scattered. What remains is diffuse irradiance, light arriving from the whole sky rather than from one direction. Diffuse light is weaker per unit of sky area, but it is also less directional, which has two consequences. First, the array produces less current, because the total irradiance in the plane of the modules is lower. Second, the light is more uniform across the array, so mismatch between modules is often smaller than on a partly cloudy day.

The drop is real but it is not a switch. A silicon module that produces several amps under full sun still produces a fraction of an amp under heavy overcast. The inverter remains on, the meter still turns, and the monitoring portal still shows a curve. The curve is simply lower and flatter than the one the reader remembers from last week.

The cloud-edge effect

A partly cloudy sky can produce brief periods when array output exceeds the clear-sky value for that time of day. This happens when direct sunlight passes between clouds and reflects off the edges of nearby clouds onto the array, adding to the direct beam. The effect lasts seconds to a few minutes and appears on a production graph as a narrow spike above the smooth clear-sky curve.

The spike is normal and is not a sign of a fault. It is also not a reason to expect higher yield; the energy in those spikes is small compared with the loss during the cloud passages around them. What matters for the reader is that a graph with spikes and dips on a partly cloudy day is describing the weather, not an inverter problem.

Rain, soiling, and self-cleaning

Rain washes dust, pollen, and bird droppings from module glass. A module that has gone weeks without rain accumulates soiling, and the loss from soiling can be larger than the loss from thin cloud. A steady rain event therefore often restores production to something closer to the clean-module value.

The effect is not a substitute for cleaning in all climates. In regions with long dry seasons, or on arrays mounted at a shallow tilt, rain may not remove all soiling, and the residue can bake on. The point is that a production increase after rain is expected and does not indicate that the array was faulty before.

Winter, low sun, and a week of poor production

Winter reduces production for reasons that have nothing to do with cloud. The sun is lower in the sky, days are shorter, and the angle of incidence on a fixed array is less favourable. Snow cover on the modules can reduce output to near zero until it clears. A week of low production in December or January is usually the sum of these effects, not a failure.

The natural mistake is to compare a winter week with a summer week and conclude that something has broken. A fair comparison uses the same weeks in previous years, or the same month, and accounts for weather. Monitoring portals often provide a clear-sky model or a weather overlay for exactly this purpose; if the measured curve tracks the model’s shape at a lower level, the array is working.

Separating weather loss from a fault

The table below lists common observations on cloudy days and whether they indicate a fault. The final column is the condition under which the cause or fix does not apply.

A string that is consistently lower than its siblings under all conditions deserves attention. A string that is lower only under cloud, and equal under clear sun, is more likely to be a shading or orientation difference than a failed module.

Reading the monitoring data

Most residential monitoring systems report inverter AC output, not module DC output. Under heavy cloud the inverter may shut down if the DC voltage falls below its start threshold, and the portal will show zero even though the modules are producing a small voltage. This is normal behaviour, not a fault, and it is one of the reasons a cloudy-day graph can look like a flat line.

A quick way to check whether the array is alive is to look at the inverter status, not the production curve. If the status is ‘waiting’ or ‘standby’ during dark weather, the inverter is doing what it was designed to do.

# Example: reading a typical inverter status page
# Status: Waiting  -> DC voltage below start threshold; normal in heavy cloud
# Status: Producing -> inverter is converting; output may still be low
# Status: Fault    -> inverter has recorded an error; check the code

Every one of these checks is made from the monitoring portal or the inverter’s display. None of them involves the DC side of an array, which produces dangerous voltage whenever light falls on it and cannot be switched off at the panel. Work on the DC side, roof work, and any mains connection is licensed work.

If the status shows a fault code, the code should be checked against the inverter manufacturer’s documentation. A fault code is not the same as low production.

What people get wrong

The common misdiagnosis is to treat a cloudy-day production drop as evidence that the array is failing, or that the installer overstated the system’s output. This is a natural mistake because the proposal the reader was given usually quotes an annual figure, and the first cloudy week arrives before the reader has any sense of the seasonal range. A system that produces a fraction of its rated output on an overcast December day is not underperforming; it is producing what the irradiance allows.

The second misdiagnosis is the reverse: to assume that because the array still produces something under cloud, the cloud loss is negligible. It is not negligible over a year, and a system sized on a sunny-climate assumption will produce less in a cloudy one. The annual figure in a proposal should already account for local weather, but the reader should confirm what weather data the figure was based on.

The third is to read a monitoring gap as a production gap. A communications dropout, a firmware update, or a portal outage can leave a flat line on a day when the array was producing. The inverter’s own display, if it has one, is the cross-check.

What to check locally

Anything involving net metering, interconnection, export compensation, permitting, or incentives is set by state and by utility and changes over time. This page does not state those rules for any jurisdiction. The reader should check with the local utility and the state regulatory body for the current rules, and should confirm with the installer what weather data and assumptions underlie the production estimate.

For the weather side, the useful checks are the same everywhere: compare like months, use the monitoring portal’s weather overlay if it has one, and treat a single cloudy week as weather rather than as a fault.