Chairmen of two House subcommittees, upset over a series of after-the-fact changes in reports from Union Carbide Corp. about toxic emissions from its pesticide plant in Institute, W.Va., yesterday asked the General Accounting Office to investigate the Environmental Protection Agency's data-gathering methods.

Reps. Henry A. Waxman (D-Calif.) and James J. Florio (D-N.J.) began looking into spills at the West Virginia plant after methyl isocyanate (MIC) escaped from its sister plant in Bhopal, India, killing more than 2,000 persons. The Institute plant, the only one in the United States that manufactured the pesticide ingredient, halted production of MIC after the Dec. 3 tragedy.

Since then, Waxman's Energy and Commerce subcommittee on health and the environment has received a set of documents that Carbide gave the EPA about the West Virginia plant. On three occasions after subcommittee staff members questioned large releases of MIC or the equally toxic chemical phosgene, they were told that the company had made mistakes in compiling the information for the EPA.

The latest change involved a reported release of five tons of phosgene at the plant in August 1980. When asked about it, the company said the figure should have been 600 pounds.

Earlier, the firm had downgraded reported spills of MIC from 840 pounds to five pounds, and from 12 pounds to .62 pounds.

In their letter to Charles A. Bowsher, who heads the GAO, the lawmakers said the discrepancies "raise serious questions about the reliability of any data" given to EPA, especially because "Union Carbide has a reputation of being one of the more safety-conscious" chemical companies.

Florio, chairman of the commerce, transportation and tourism subcommittee, said he wants the GAO to "evaluate the quality control" used by EPA in gathering information from chemical companies on the manufacture, storage and transportation of toxic substances.

Waxman also has asked 87 major chemical companies to tell his subcommittee about leaks and spills, safety surveys, health-effect studies and control technology in the nation's 300 largest chemical plants.

A Waxman spokesman also criticized the release yesterday of an EPA report of environmental pollution in West Virginia's "chemical valley," including the Institute plant, that was completed last August but apparently not made available to two top EPA officials who testified in December at a Waxman hearing in Institute.

The report, made public yesterday by EPA after it had been leaked to The Wall Street Journal Thursday, identified the Institute plant as a major industrial source of pollution in the Kanawha Valley.

It noted that until last summer, there had been no federal or state regulatory limits on airborne volatile organic chemicals. The report, an update of a 1977 study, said low concentrations of several of those substances -- benzene, chloroform, carbon tetrachloride and vinyl chloride -- are known or suspected to cause serious health problems.

The study found toxic substances being disposed at 50 sites in the thickly populated area around Charleston, the state capital, and that at several of the landfills, ponds, dumps and strip mines where the materials were being discarded, toxic substances had leached into the ground and surface waters.

Cleanup efforts had been made at few locations, the report said, noting that only one site, Fike Chemical in Nitro, is a priority site for cleanup under the federal "Superfund" program.