this means that the weather/climate data set you're working with offers a maximum spatial resolution of 1 square kilometer, i.e., a 1 km by 1 km grid cell. So your data points will each pertain to one such 1 km × 1 km grid cell. Consequentially, you won't be able to say anything about spatial temperature variations WITHIN any single grid cell. Thus, you can only compare whole (1 km x 1 km) cells with each other. In short: Your data set only offers ONE single ("aggregate") figure for a given atmospheric parameter for a whole 1 square kilometer grid cell.
Hope this helps a bit and please do not hesitate to get back should you have any further questions.
It means that the measured temperature is assumed to be the same everywhere in the area 1 km x 1 km. The temperature field is continuous but the meteo-stations installed only at discrete geographic points, they cannot be everywhere. Therefore those measured temperatures are only discrete (so called spatial) approximations of the continuous temperature field.
Notice that daily mean temperature has no physical significance. Mean or average temperature does NOT have meaning of temperature at all because there is no physical object to which the average temperature can be applied. No continuous field can be substituted by only one number, "the average".
You can talk about the daily temperature range, say from 10 C at midnight night to 20 C at noon, but averaging it to get 15C is physically meaningless, it is just statistical property of the values 10 and 20. Temperature is an intensive thermodynamic variable and cannot be added and then divided by the number of measurements.
I highly recommend the fundamental paper by Essex et al
"Does a Global Temperature Exist?" It is a MUST READ for everybody involved in mean temperature discussion.
It means the overall measured temperature is assumed same within the grid/spatial resolution (1 km x 1 km). It's not possible to make measurement every cm or m, so the best way to represent the meteorological variable is through grid cell of larger spatial resolution.