I am very squeezed and surrounded.
I think you exaggerated here.
There exist 3 flood plain wheats near Almarikh in the picture you posted above. 2 of them couldn't get used by any city even with a cultural expansion. A flood plain wheat plus a mined desert near a capital likely gets the city to 2 shields per turn, and the use of another tile may well get it to 3 shields per turn. Either way, towns near there still end up decent for producing workers.
I don't understand why you didn't place a city like Mandalgovi there. If the city were at the 8-7 spot from Ulaanbaatar (starting from Ulaanbaatar if you controlled a unit and pressed '8' and then '7' on the numpad, then the unit would reach the spot I refer to), the deer could have gotten used as soon as the city got founded.
And the worker roading the forest just outside of Mandalgovi? Are you automating workers? Have you thought about how to optimize worker moves before? It
almost never ends up best to road a forest tile before chopping it. Maybe even to make the rule more clear, I would do better to say NEVER. Never road a forest. Chop a forest first. Then road.
As another suggestion, I see Darhan on a granary. But, if you have forests near a city, those forests can get chopped to get the granary in. For a 1 shield city with forests there exist a few different possibilites:
1. Build the granary without any workers. That could take 60 turns.
2. Put out a worker. Then chop a forest, have the worker at least road the tile afterwards, if it does not seem worth it to improve the tile. With just one worker produced, the city produces one worker in 10 turns, then after 1 + 4 = 5 more turns it has 10 shields back from the worker having chopped a tile. So, in 15 turns, the city has 15 shields back from the worker, and thus has recouped the shield cost of the worker taken off the granary. Given that the worker can chop one other tile reasonably, the city will then have 20 shields towards the granary in comparison to the 10 shield cost of one worker!
2 workers may take 20 turns to build, but 2 chops recoup their costs, and can improve tiles also. And there's enough turns for one worker to either add back into the city for more shields per turn from the city to have the granary finished sooner because of workers getting produced before a granary.
I am about to hit libraries and I am in desperate need of space.
I see space for another city, if not two, west/left of Kazan. From Kazan at the 4-1 spot could go another city and claim unused tiles.
Backing up even more, I see that you generally place cities at a CxxxC sort of pattern. The empire ends up with a fair amount of unworked tiles that way for a while. It does not suit military games as well as a CxxC sort of pattern and even CxC spacing.
Instead of Kazan where it sits, it going at the 8-8-8 spot from Karakorum could work better. You wanted to grab the horses with it's placement? Then placing a city at 8-8-8 from Karakorum, then another at the 7 spot from Kazan could work. Or from Karakroum, a city at the 8-8-9 spot instead of the 8-8-8 spot. Ta-Tu instead placed at the 7-7-7 spot from Karakorum, then a city where Ta-Tu sits now or at the 1 spot from where it now sits. A city at the 6-6 spot from Karakorum would make a lot of sense. A city at 1-1-4 from Karakorum, then Tabriz pushed further onto the coast also.
Was that 4 more cities? Darhan also looks far from Hovd to me. But with more cities and more tiles worked, you might just not feel so cramped.