Thanks for the clarification. Your problem has two parts. 1) What is the spacial relationship between the ground station and the satellite and 2) given a spacial relationship between two radios, how 'good' is the connection between them.
Problem 1: Taking the Easy Efficient Route
The spacial relationship between a satellite and a receiver can be calculated by hand (a common home work assignment in aerospace classes, and another discussion in itself), but the fastest way to find satellite coverage windows is to use a computer. There is plenty of free software available that can do this, but I'll recommend the free version of STK. Here is what you can do with the free version, and here is where you can download it.
Problem 2: Elevation Angle
The system could be designed in various ways. Data is often broken into packets, which can be transmitted individually. The satellite needs to know which packets the ground station has received successfully and vice versa. Successful reception of a packet is acknowledged with an 'ACK' transmission. To downlink, the satellite could be commanded to transmit a packet repeatedly until it is received successfully, then commanded to transmit the next packet. To uplink data, the ground station would transmit a command or packet and wait for an ACK, then transmit another packet. These methods are agnostic of communication angle and path loss. You simply start when the satellite is approaching the window.
Problem 3: Relating SNR and Path Loss to Elevation Angle
First, Power. Free Space Path Loss (FSPL) is used in conjunction with the Friss Transmission Equation to compute signal power at the input of the receiver (not the receivers antenna, the actual receiver). Given the location of the radios and some information about the receivers this is straight forward. Including atmospheric and other effects can be done if you want a more realistic model (let me know if you're curious).
Next, Noise. Noise is a little complicated since it involves the actual noise of the transmitter and receiver, as well as external noise. Let's ignore external noise and look at thermal noise. Electrical noise is often measure in temperature because thermal energy translates to electron motion, which is electrical noise. Noise power density (noise power per unit bandwidth) is found as $N_0=kT$ where $N_0$ is in watts, $k$ is the Boltzmann constant in joules/kelvin, and $T$ is the receiver system noise temperature in kelvin. A simple approximation of $T$ on Earth is 290 K. (This is ignoring a few things, but it's an okay first attempt).
Noise power spectral density is found as $\frac{E_b}{N_0}=\frac{S/R}{N_0}=\frac{S}{kTR}$ (not in dB) where $S$ is signal power, and $R$ is the bit rate. This is the Energy per bit over the noise density, or the SNR per bit. Defining this in terms of bits will help us relate it to data rate. Further, $SNR = \frac{S}{R}=\frac{E_bR}{N_0B}$ where $B$ is the bandwidth of the signal. Changing your encoding will give different relationships between noise and bit error rate.
increasing the SNR will require increasing the power transmitted
Remember decreasing system noise also will increase SNR.
Final Thought: Data Rate
The Shannon-Hartley theorem defines the maximum possible information rate over any (AWGN) channel. $I<B \log_2{\left(1+\frac{S}{N}\right)}$.
You can now relate elevation angle with path loss, SNR, and even maximum data rate. Hurray!