Reading all the answers, I realized that the confusion comes from accurately determining the height. On page 153 of the CLRS book, the height is defined as follows:
Looking at the heap as a tree, we determine the height of the node in the heap to be the number of edges on the long simplest descending path from node to leaf ...
Now look at the original heap provided by Nishant. Nodes 8, 9, 10, 6, and 7 are at a height of 0 (i.e., leaves). Nodes 4, 5 and 3 are at a height of 1. For example, between node 5 and its sheet, node 10 has one edge. There is also one edge between node 3 and its leaf node 6. node 6 looks like at a height of 1, but at a height of 0 and therefore on the sheet. node 2 is the only node at height 2. You might wonder that node 1 (root) is the two edges from node 6 and 7 (leaves), and they say that node 1 is also at height 2. But if we look back at definition, the word of the bold word "longest" assumes that the longest simple descending path from root to leaf has 3 edges (passing node 2). Finally, node 1 is at height 3.
Thus, there are 5, 3, 1, 1 nodes at a height of 0, 1, 2, 3, respectively.
We apply the formula to the observation we made in the previous paragraph. I would like to point out that the formula proposed by Nishant is incorrect.
It should be a ceiling (n / 2 ^ (h + 1)) not a ceiling (n / (2 ^ h + 1). Sorry for the awful formatting. I still cannot post the image.
In any case, using the correct formula,
h = 0, ceiling (10/2) = 5 (nodes 8, 9, 10, 6 and 7)
h = 1, ceiling (10/4) = 3 (nodes 4, 5 and 3)
h = 2, ceiling (10/8) = 2 (node ββ2, but this is normal, because the formula predicts that at a height of no more than 2 nodes).
h = 3, ceiling (10/16) = 1 (node ββ1)
With the correct determination of height, this formula works.