What actually does ‹elevation gain› equal?

[Victamon]

Elseplacesߴ usage of the terminology with ‘cumulative’ prepended to it suggest that it represents the total climb in altitude ignoring portions of dips. Going by that meaning, ʼtwould be conceivably possible to walk on a relatively flat course that varies repeatedly between a maximum vertical range of |𝑐| ft. (for some constant 𝑐 small enough so as to qualify for "relative flatness" gradient for length of the course) yet legitimately count an "elevation gain"≕⦗Δ⁺ ∘ ⟮𝑎 ≔altitude⟯⦘ of a substantial magnitude times |𝑐|: for example, if 𝑐=^±3 with #undulations=15 then 3ft<Δ⁺𝑎≤45ft (with ⟮|Δ^±∘𝑎|≕ Δ⁺(𝑎)⟯ closer to elevational supremum 45ft as the small distinct undulations become more uniform in magnitude). However, the corresponding Δ⁺(𝑎) of a recent recorded short(4m44s⊶0.26㎞) Walk of mine seems to be less than the identified 𝔤≔"elevation gain"— i.e., apparently 𝔤_𝛋≩⟮Δ⁺(𝑎)⟯_𝛋 for a particular parambulatory activity of mine identified as 𝛋 ﹘ as seen below:

Notice that፦ following the forward direction of time, the track appears to (albeit not monotonically )progress from the lowest point (0m5s in) to a highest point (1m53s in) approximately 11½ft higher (from altitude min–max of ∼ 213–225 ft.); yet, the stated "elevation gain" is merely 5ft.፤ Wherefore?

[Triletics]

Algorithms estimating an elevation gain usually use a minimum gain threshold so relatively flat surface stays flat. For that reason, the same data will bring slightly different elevation gain (different data smoothing applied to the elevation gain calculation and gain threshold).