You are not alone by noticing HD transfer rates decline. I am afraid, that is not an issue, but the way things work IMHO our equipment is not optimized for a bulk data transfers but for a lot of (relatively) small reads and writes.
As far as I can judge high initial speed is due to a lot of buffering and caching involved. Buffering and caching do not work well with bulk data transfers. And their effect is that they report a finished HD write transaction as soon as data are put down in a buffer/cache and not actually written onto media. That is why there are huge transfer rates shown.
On the other hand, as soon as buffers and caches are overflowed with a stream of non-reusable bulk transfer data we do see what is (very likely) just pure physical transfer rate.
Your transfer rate is a constant 833 MB/s . This is the physical limitation of your storage medium connected to, and in combination with, your motherboard.
Your copy command is completely oblivious to the buffering that the kernel is doing and has no idea that it is filling a write buffer in RAM instead of writing directly to disk.
This makes it think that the transfer rate is much much higher than it really is, because RAM is much much faster than NVME/SSD/HDD.
The rates you see (1.4 GB/sec and 833 MB/sec) are the average over the total transfer. Actual speeds vary at any given moment. These numbers are the total bytes transferred divided by the number of seconds since the job started.
If you want to know your maximum burst transfer rate (the one that storage mfrs advertise) go to Applications -> Accessories -> Disks
In the upper right hand corner of Disks is three vertical dots. Click on that and select Benchmark. Follow the instructions and it will give the transfer speeds and latency.
IMHO our equipment is not optimized for a bulk data transfers but for a lot of (relatively) small reads and writes.
