There is a bit of confusion about my answer. I'm updating the answer with clarifications and additional information. The quotes are my original answer.
TL;DR
I think the best answer is: you can't.
You can't realistically get close to 24-hour accuracy with any method mentioned. The only methods that close require atypical knowledge and very specific circumstances. More realistically, you're probably looking at an error window of 30-60 days.
Try to stock up and stay as close to the gate as possible for those last couple months.
Assumptions
The question specifically asks how a person could know what day to return. My assumption is this is an Earth day, and I'm going to be generous and assume $\pm$24 hours is acceptable. Said another way, we're looking for accurate enough timing that you could walk away from the stargate for 364 days, then come back and be confident that you're not late, and no more than 24 hours early.
(Note the question was edited so it's no longer so specific. The revised information tries to calculate the realistic accuracy and why this was my answer to begin with.)
We know the planet has breathable air, but nothing else. For all we know, you're literally the only living thing on the planet. But since there's a presumption that survival is possible, I'm going to assume this planet is fairly Earth-like, is populated by plants and animals similar to Earth-life (particularly, they're made of proteins, vitamins, etc. that are digestible by humans), and that the environment near the stargate is similar to some environment here on Earth.
The question asks about a "stargate", so I'm going with a Stargate approach: a few million years ago, some ancient almost-humans explored the galaxy, constructed stargates on thousands of planets, and terraformed quite a few of them. In that time, other aliens have moved in, invaded, crash-landed, etc., and natural evolution has occurred as well. The planet is similar to Earth, but may have marked differences from Earth.
The question specifies that "I" got stuck on the alien planet. I don't know much of anything about the writer, so I'm going with a statistical approach. The writer obviously has access to a computer and the internet, so I'm assuming we're looking at a statistically-normal first-world citizen who is probably reasonably smart, understands at least the basics of core subjects at about a high school level, and to make the answer interesting, I'm assuming this person has a strong desire to live.
Clearly, some of the suggestions below would fail if the planet were less Earth-like. For example, if it's tidally locked to the local star and the star is too bright to see any other celestial objects, sundials and the like wouldn't work. That just makes my answer more right, so it's not really important here.
Most of the suggestions here would get you in the ballpark, but none of them would be highly accurate. Circadian rhythms, menstrual cycles, resting heart-rates, etc., are all based on many factors. First, none of those are the same from person to person, so you'd have to know your own. Second, they're all affected by stress, metabolic rates, nutrition, and so forth, so the instant you hit the new planet they change. Third, none of them are going to be the same from day to day or month to month with extreme accuracy even in a modern, normal, predictable life, much less on an alien planet.
Before we begin, note that there are two separate parts to this answer. The first part is counting time on the new planet. The second part is determining the ratio of local time to Earth time.
Local Time Keeping
Very few people would have the knowledge or skillset to build a pendulum clock or anything more complex, so you'd basically be restricted to sticks and shadows, which would be accurate to a few minutes, but not much more. And unless the planet is in a perfectly circular orbit with zero axial tilt, the scale of your clock will change every day. So you'd need a local year or two to really calibrate the clock.
First, let's look at suggestions for determining local time. It's pretty easy by comparison.
Count the Days. The easiest counting method is alien days. You've got a giant alien clock right under your feet. Just tick off how many days have passed, apply a conversion factor, and voila! You could also use a readily-identifiable constellation, a moon, whatever. (Technically, distant stars are best, because highly-eccentric orbits don't affect that measurement.)
Use a Sundial. But you need to correlate alien days to Earth days. It's unlikely you can maintain any particular counting method for the entire day (moreso to do so without errors), so you'll want a shorter interval. A sundial will give you a good way to track time in hour-ish increments, and possibly as good as minute-ish increments. Sundials are extremely easy to implement, so that's good. Use one of the Earth-time methods below to see how long 5 alien minutes, or one alien hour, etc. are.
The bad part of a sundial is that the ratio (probably) changes from day to day, so you'll have to take that into account. It's best if you can get the timing of an entire day/night cycle, so you'll want a "stardial" of some type. Using a raised, flat rock with sand on top, you can put a pointy rock near the edge, then sight along it from below and beside the flat rock. Put a mark in the sand directly between your eye and the rock, while "aiming" the end of the rock at some notable star as it comes off the horizon, then another mark every so often (every thousand heartbeats, or every tenth time you finish Mary Had a Little Lamb, whatever). Make another mark when the star goes behind the horizon. By comparing the arc length of several smaller intervals to the arc length of the full night, you can estimate the length of the night.
Unfortunately, neither of these is going to be terribly accurate. You either need to build a really big sun/stardial (this is possible, and probably a good idea), or make very precise, very tiny marks each time. Otherwise each mark will span minutes, perhaps tens of minutes.
Build a Pendulum. A really nice way to tell shorter time periods is some type of periodic motion. Like a pendulum. Unfortunately, keeping the pendulum swinging is a rather large feat of engineering unless you want to push it manually, in which case you'll introduce more error. Still, this is probably a pretty decent approximation given your options. But you need to make sure to use a thin, light string of some sort, or get some really slick goop to "oil" the contact if you slide, e.g., a stick onto another stick with a hole. Otherwise the stiffness or friction is going to introduce errors that will be hard to account for later.
For most people, the pendulum will be adequate for measuring time over several hours to more accurately calibrate the day, but it's unlikely many people could construct an actual clock any time soon.
A Water Clock. This requires you to be capable of moving water in sufficient quantities to fill at least one bucket, and preferably several, and is a pretty decent way to estimate the length of a day. You wouldn't want to keep it running constantly, but you don't really need to so that's fine.
Find a Local Clock. It's possible there is an existing civilization who has existing clocks. This makes your job trivial, but isn't remotely guaranteed. Plus, they aren't guaranteed to be friendly if they do exist.
I think there are probably some other methods I missed which would work decently. All of them will have some margin of error, but it's plausible they would be good enough.
The worst problem here is that you don't have very long to build an accurate device. If you spend 6 months constructing an elaborate grandfather clock, any physiological processes you might have used to calibrate the clock will have long since changed.
Relating Local Time to Earth Time
If you could engineer stuff, and you happened to have an object of known mass, one of known length, and one with a known spring rate, you could use all of that to determine the local gravity and make a clock from there. But you'd have to remember of bunch of physics equations off the top of your head and be able to apply them with rudimentary technology. All while not getting eaten by a Grue.
There are probably ways to make chemical clocks, but I doubt a layperson could do it at all, and even a chemist would have trouble without a lab on the planet. The speed of sound and electricity are variable depending on the medium, so wouldn't work very well. The speed of light is pretty constant through air, but you'd really need to know either time or distance for it to help. And good luck building a device to measure any of the above out of sticks and stones.
So now we need to get the Earth duration of some unit of alien time. This is where it gets extremely difficult. I'm going to cover the major ideas presented on the page so far, and why none of them are particularly great. I'll also try to give reasonable estimates of how accurate they are. Naturally, these don't cover every single variation of answer, but the same premises will apply to similar answers.
Count Your Heartbeats. My resting heart-rate, as measured by professional nurses with actual medical equipment, while I'm as calm and rested as I can get, has ranged from ~70 to ~110 BPM. It tends to be in the 80-90 range. I don't know what a typical deviation is for a single person, but it's probably at least $\pm$5 bpm.
So 85 $\pm$ 5 is an error of $\pm$ 5.9%. After a year (I'm using 365.2422 days, although I doubt the extra precision matters here), that 's a $\pm$ 21.5 day deviation.
That assumes the person going through knows their own average resting heart-rate. It also doesn't take into account the incredibly significant effects of stress (you just got tossed through a stargate and have to survive here for a year), the alien environment (different gravity, oxygen levels, air pressure, etc. will have an effect on your biology), or, as days and weeks pass, the different nutrition you'll be getting, your physical condition, etc.
Realistically, most people probably won't be able to be certain of an error margin of anywhere near the $\pm$ 5 BPM range, driving the deviation even higher.
Measure Your Period. It's quite typical in normal, everyday Earth life for a woman's period to be off from average by $\pm$ 2 days or more. Average period length is around 28 days, which gives an error range of $\pm$ 7.1% or $\pm$ 26.1 days. Wikipedia puts $\pm$ 4 days as more typical, which is an error range of $\pm$ 14.3% or 52.2 days. The highest typical period length is around 45 days for adolescents, or 31 days for adults. 45 days $\pm$ 2 days gives an error of $\pm$ 4.4% or $\pm$ 16.2 days, while 31 days $\pm$ 2 days is $\pm$ 6.5% or 23.6 days.
Like heart-rate, period length will be greatly affected by stress, diet, etc. It's actually a bigger issue here, because the next period will be days to weeks away, meaning your body will have a lot of time to adjust to the new conditions, while the heart-rate measurement could be taken within minutes of arrival.
And, period length is usuable to less than half the population. About 50% of the planet is male, and while I didn't bother looking up numbers, young girls and older women don't menstruate, so actual numbers will be less than 50%.
Another issue my room-mate mentioned is birth control. Anyone currently taking birth control will have substantial hormonal changes going on once they hit the new planet and have no more pills.
Use Your Favorite Music. Obviously, not everyone has heard of John Philip Sousa, let alone memorized any of his songs, but most people know a decent number of songs. Great! But does that help? Not really.
I highly doubt the average person can tell you the BPM or running length of any particular song they own. I can guarantee that only a (relative) handful of people are capable of singing through an entire song without missing a beat. Then they have to get the exact right tempo.
Certainly there are plenty of musical professionals who could do a pretty decent job. But even then I doubt they'd hit the 24-hour mark. That answer says 120 BPM $\pm$ 2-10 is reasonable for someone with moderate to little musical training. That's an error margin of $\pm$ 1.7% to 8.3%, or 6.1 to 30.4 days. Even the tight end of that scale misses the mark by quite a bit, but is looking good compared to other answers.
I'm not a musical expert by any means, and had no luck looking up statistics relevant to this, so I can't comment on how accurate that $\pm$ 2 BPM estimate is. But something tells me that's after sitting in front of a metronome or similar device, and most people probably program that timing with a musical instrument. The lack of any such tools would probably throw the estimate off by more.
Again, stress and local factors could have a huge impact on the perception of time and the ability to maintain a consistent rhythm, even for the pros.
Measure Hair or Nail Growth. To add to the idea, when I went through basic training the new toe nails were much thinner than my old nails, and I could visually see the time passing.
Hair grows around 6 inches per year, while nails grow around 1.44 inches per year. It's probably a lot easier to measure hair than nails in this case, but it's easier to cut into nails to keep them marked in more precise intervals. And there's a big difference between the length of neatly-brushed hair and caveman hair.
However, as Wikipedia points out, actual growth time can depend on many factors, such as the stress, etc. on our alien planet. More importantly, I highly doubt many people could tell you their personal growth rates right now with any significant accuracy.
It's hard to put an estimate on the accuracy here, but it's probably at least $\pm$ 10%. Hair on different parts of your head doesn't grow at the exact same rate, so you'll have to average it out a bit. Not everyone will have just gotten a haircut when they got pushed through, so the unevenness will be even higher. Plus the biological factors.
From freshman astronomy, you remember that the period of Delta Cephei (a cepheid variable) is 5.36 days, doubling its brightness at maximum. This is a pretty good trick, but that requires the person to have learned this fact (or one like it) at some point, still remember it to an exact enough value, and be able to find the star in question. Given that stargates could dump you all over the galaxy (or even other galaxies), this is a very niche technique.
If you can use the trick, the $\pm$ 0.01 day accuracy is good to about 2.4 seconds at the end of the year, although you'd likely be off by at least a few hours in trying to determine the center of the maxima and guessing at how long it had been between arrival and the first maximum. But certainly a very good technique.
But if you can't use the trick, it's worthless, and the odds of the trick working are very slim.
Practice a Speech or Clapping Your Hands. This sounds like a neat idea on its surface, although I doubt it's terribly accurate. Remember that alien worlds could easily affect your speech and muscles by a considerable margin.
But the worst problem is it requires preparation. If I was going to prepare, I would bring something with me that was far more accurate, like a watch. Even simple things like calibrated springs, weights and rulers would do a better job. If preparation is allowed, it becomes trivial to calculate time. But that's not what the question asked.
Dial a Random Gate for 38 minutes. I thought of this one, and it's not a bad idea. There are still problems. First, this narrows the assumption from "Stargate-like" to "this is the Stargate universe", which is probably too narrow. Second, most people probably never knew about the 38 minute limitation, let alone remember it.
But more importantly, it took trained professionals (in-universe, which we need to consider meaningful if we're using the 38-minute rule) decades to get one correct address. The odds of successfully dialing out in the entire year you're there are vanishingly small.
However, if it worked and you happened to get a gate open, $\pm$ 1 minute translates to about $\pm$ 2.6% accuracy or $\pm$ 9.6 days. So we've still missed our window, but it's not a bad estimate.
Calculate the Period of a Pendulum. For starters, your height is probably much less variable, and a longer length translates to better accuracy in the long run. So I'd use that rather than an erection.
But can we actually do it? Not really. The formula for the pendulum makes several assumptions already, so we need to be careful. But let's pretend we can meet those assumptions using weeds and sticks and so forth.
We still need to find both L and local g. L is easy enough with something like your own height. Pretty much everyone knows their own height, which is a huge plus, but it's not ridiculously accurate. Best case measuring accuracy is around half an inch, and most people are probably between 5' and 6.5' (60-78") tall. That gives us an error margin of $\pm$ 0.8% to 0.6%, which is quite good, leading to a $\pm$ 3.0 to 2.3 day total error.
But we still need to be able to replicate that height on the alien planet, which is probably going to double our error margin to around $\pm$ 6.0 to 4.7 days. But we're not finished. We still haven't found local g.
The proposed method won't work once you're on planet. It requires you to know the initial velocity of an object, which requires you to measure distance over time, and time is the things we're searching for. Two unknowns, one equation.
If we happened to know how far and how fast we could throw a particular object on Earth (as suggested), it could work. But there are several variables to consider. According to this random site, MLB pitchers (they're literally pros at throwing things) throw a baseball about 92.15 mph, $\pm$ 2.6 mph. That's $\pm$ 2.8% or 10.3 days of error. Then remember that most people aren't pros. And we're throwing random rocks or sticks that won't be the exact same mass and shape.
Then there's throwing distance. I can't find any good references, but we're adding at least a few percent here.
Finally, there's throwing angle. According to Wikipedia, MLB pitchers had an easy time hitting a strike zone between a batter's shoulders and knees. Let's call that a 4 foot height, which is at a distance of 60.5 feet from the pitcher. That's an angle of $\pm$ 1.9°.
Using the $d = \frac{v \cos \theta}{g} \left( v \sin \theta + \sqrt{v^2 \sin^2 \theta + 2gy_0} \right)$ given, with a velocity of 70 mph (we're not all MLB pitchers), $y_0$ of 6 ft, Earth gravity, and a horizontal throw, we get distances between 52.8 and 74 feet. That's 63.4 $\pm$ 10.6 feet, giving an error margin of $\pm$ 16.7% or 61.1 days.
Now, the errors given so far have been per measurement. We need to add them all together to get the final measurement. With a tall person (6.5'), that's a total error of $\pm$ 20.8% or 76.1 days. And, yet again, local factors will cause more deviation.
(Of note, adding them together means you get a more condensed curve, so it's actually a little better when accounting for whatever arbitrary probability cutoff we're using. $\pm$ 15% or 54.8 days might be a better estimate.)
Drop an Object Off a Cliff.
Measure Echo Distance. These suffer the same problem that they require a precise, working clock to make your measurements, which negates the purpose of the experiment.
Measure Time to Run a Distance. The problem here is it assumes people can accurately determine a half-mile distance and know how fast they run it. But both of these assumptions are flawed. First, normal people don't do that much running to begin with, and can probably guess their half-mile speed to $\pm$ a minute or something. But anything substantially better than that isn't likely. Second, I've met many people who couldn't judge 100 yards within 20%, let alone 1700. This second problem can be somewhat overcome with time (use your height to measure a length of wood, then use that to measure half a mile to decent accuracy -- probably $\pm$ 2-5%).
Additionally, even professional runners are unlikely to maintain close to their normal pace when you put them on a random alien planet on rough terrain. Even if they can easily tell they're slower than normal, they won't have any particularly reasonable way to determine how much slower they are. You're still going to be off by at least several percent, and probably 10-20% depending on just how rough the terrain is.
This is another place where it's pretty hard to make any estimates, but probably a minimum of $\pm$ 10% or 36.5 days.
Find an Atomic Clock on the Planet. This would certainly be an excellent method of telling time, because it's the same frequency anywhere in the universe.
However, you'd have to remember off the top of your head that Cesium oscillates at around 9.19 billion cycles per second, and hope the aliens are using Cesium, and that you know enough chemistry to verify that they're using Cesium (they aren't likely to call it Cesium).
And there's no guarantee there are people on the planet, let alone an advanced civilization. Using my "Stargate-like" assumptions, we know that most of the gates lead to extremely primitive planets, if the planet is even habitable.
So this is another case where the timing is more than sufficient if everything lines up, and worthless if it doesn't.
Use the Dimensions of a Dollar Bill. Assuming you have any such items on you, this is certainly an excellent suggestion. But you have to know the length, weight, etc. of such items for them to be useful (I have no more idea of the length of a dollar bill than my own finger). And you have to have these components on you as you get shoved through.
This isn't an answer so much as a "how to help with other answers" suggestion. It's certainly good advice, but the question doesn't specify we have these kinds of tools so we can't presume we do.
How To Maximize Survival Odds
Consider that your daily activities are going to be hunting, gathering, fishing, etc, just to stay alive, and possibly running and hiding from alien predators. Your first few weeks on the planet will be one hellish blur, and by the time you get into a normal routine (assuming you even survive, which, let's be fair, you probably won't), any connection between Earth time and alien time will be long gone.
One of the comments suggests that survival is trivial. It might be. You could land in the literal Garden of Eden that was taken from us and placed on this planet thousands of years ago.
But it's probably not. You have no way of knowing what's edible, what's poisonous, what's edible but totally worthless, etc. You can certainly just eat a bunch of everything, then use binary search methods to figure out what made you sick. But if you're throwing up all your food you might not survive the process.
There will be alien bacteria your body isn't used to fighting. You won't have immediate access to distillation methods, and most people probably wouldn't figure out anything better than boiling the water for treatment at all. That super-clear river water will probably make you very sick at first.
You could land in the middle of a desert. Or the arctic. Or anything in between. There could be extremely dangerous predators on the planet, or human-sized carnivorous plants. Or RouS's.
Another comment mentions that "it's the only ticket back" and assumes that means people will automatically find a way. Unfortunately, that only works in the movies. While perseverance and the will to live certainly help, they can't magically turn a person into a chemist or engineer or teach them advanced physics or math skills. Nor can they generate a fire out of solid ice, or food out of famine.
Your best bet is to stockpile food and water in an area near the stargate, then camp out right next to the gate for the last several months. But that assumes you have the materials and knowledge to keep food fresh for a couple months. Depending on the local season, you might be able to harvest fruit from trees that grow near the stargate.
Various sources I've read say that it takes between 1 and 10 square miles of land to support one person doing hunter/gatherer survival methods, so you could be very far from the stargate when it opened if you were out hunting. You would really want to transplant as many fruit and vegetable plants as possible to an area right next to the stargate, and keep it protected from the local fauna. You could set up traps and lures to hopefully get meat with spears or a bow and arrow, but you wouldn't want to leave the immediate area once you were close to the jump window.
A commenter suggests 10 mi² is a circle of radius 2 miles. However, it doesn't really work that way. First, the wildlife isn't going to happen to congregate in a perfectly-symmetrical circle around the stargate so you can slaughter them. They're more likely to be concentrated near lakes, rivers, and valleys in a more linear arrangement. Second, the nearest concentration of wildlife might be many miles from the stargate. Third, animals tend to migrate with the seasons, and random events, and the fact that you're murdering them, so you'll have to keep up, which could lead you hundreds of miles away.
The same comment also suggests you could always run 2 miles in 30 minutes. There's no guarantee the gate will be open 5 minutes, let alone 30, so you don't want to rely on that. But there's also no guarantee you happen to be in an area with relatively flat ground between you and the gate. If there are a few cliffs in the way, it could take hours to go 2 miles. And there's always the possibility of a major strain or break.
Also, there's the issue of season. It might be great near the stargate when you arrive, but the middle of the worst winter ever when the gate reopens. You'd really want some kind of greenhouse going near the stargate for this, although it could be nearly impossible to pull off.
Depending on the planet, and where the stargate is located, it might be impossible to survive very long near the stargate. In this case, you'd best hope you can run from your camp to the stargate before it shuts down again. And practice that run so you can do it as fast as possible when the time comes.
As an addendum, it's always possible the nearest reasonably-habitable zone is dozens to hundreds of miles from the stargate. A comment somewhere on the page suggests Earth might send someone through to re-open the gate from your side. In this case, it would be advisable to leave messages around the stargate explaining where you were so they could signal you or come get you since you probably won't happen to check the gate just as it opens.