Emirati Arabic remains one of the least-resourced Arabic dialects, with lexical information scattered across corpora, morphological databases, and dictionaries developed under incompatible annotation schemes. We present Gahwa, a machine-readable lexicon constructed by extracting, enriching, and standardizing evidence from an annotated Emirati corpus, a Gulf Arabic verb database, and a Gulf Arabic Wiktionary, unifying their input into one lemma-POS keyed scheme. We evaluate Gahwa through automatic coverage analysis of three Emirati datasets, GumarAE, Mixat, and the Ramsa Corpus, against existing Arabic morphological resources, and through analysis of out-of-vocabulary (OOV) forms. Results show that Gahwa has high word coverage across the three datasets, achieving 89.1-94.2% effective dediacritized lemma-POS token coverage, while the remaining OOV cases primarily reflect lexical gaps, orthographic variations, fused expressions, and morphological gaps.