A great many Buddhist works are known today in a Chinese translation and in a Sanskrit original, and a good number are known in the translation alone. Where both survive, comparing them is one of the most productive things a scholar can do, and the Siddham letters are frequently what makes the comparison possible at the level of individual words.
The reason is the untranslated material. A translator rendering a sutra into Chinese translated the prose and transcribed the mantras and dharani, and those transcriptions are anchored to particular Sanskrit sounds. Where the Sanskrit original also survives, the two can be lined up, and where it does not, the transcription plus a knowledge of the transcription conventions gives a partial reconstruction of what stood there. Manuscripts in Siddham preserved in Japan have been used in exactly this way, since they sometimes carry the Sanskrit for a passage whose Indian witnesses are lost.
The work has limits that are worth stating. A transcription is approximate, the conventions changed over time, and reconstructing a word from one is a reasoned inference rather than a reading. Editions built that way say so, and mark what is recovered as distinct from what is attested. The atlas follows the same principle: what a source establishes is one thing, and what it makes plausible is another.