Abstract. We compared 7 years of local automated weather station (AWS) data to NCEP/NCAR reanalysis data to characterize the modern environment of Lake El'gygytgyn, in Chukotka Russia. We then used this comparison to estimate the air temperatures required to initiate and maintain multi-year lake-ice covers to aid in paleoclimate reconstructions of the 3.6 M years sediment record recovered from there. We present and describe data from our AWS from 2002–2008, which recorded air temperatures, relative humidity, precipitation, barometric pressure, and wind speed/direction, as well as subsurface soil moisture and temperature. Measured mean annual air temperature (MAAT) over this period was −10.4 °C with a slight warming trend during the measurement period. NCEP/NCAR reanalysis air temperatures compared well to this, with annual means within 0.1 to 2.0 °C of the AWS, with an overall mean 1.1 °C higher than the AWS, and daily temperature trends having a correlation of over 96% and capturing the full range of variation. After correcting for elevation differences, barometric pressure discrepancies occasionally reached as high as 20 mbar higher than the AWS particularly in winter, but the correlation in trends was high at 92%, indicating that synoptic-scale weather patterns driving local weather likely are being captured by the reanalysis data. AWS cumulative summer rainfall measurements ranged between 70–200 mm during the record. NCEP/NCAR reanalysis precipitation failed to predict daily events measured by the AWS, but largely captured the annual trends, though higher by a factor of 2–4. NCEP air temperatures showed a strong trend in MAAT over the 1961–2009 record, rising from a pre-1995 mean of −12.0 °C to a post-1994 mean of −9.8 °C. We found that nearly all of this change could be explained by changes in winter temperatures, with mean winter degree days (DD) rising from −5043 to −4340 after 1994 and a much smaller change in summer DD from +666 to +700. Thus, the NCEP record indicates that nearly all modern change in MAAT is driven by changes in winter (which promotes lake-ice growth) not summer (which promotes lake-ice melt). Whether this sensitivity is representative of paleo-conditions is unclear, but it is clear that the lake was unlikely to have initiated a multi-year ice cover since 1961 based on simple DD models of ice dynamics. Using these models we found that the NCEP/NCAR reanalysis mean MAAT over 1961–2009 would have to be at least 4 °C colder to initiate a multi-year ice cover, but more importantly that multi-year ice covers are largely controlled by summer melt rates at this location. Specifically we found that summer DD would have to drop by more than half the modern mean, from +640 to +280. Given that the reanalysis temperatures appears about 1 °C higher than reality, a MAAT cooling of 3 °C may be sufficient in the real world, but as described in the text we consider a cooling of −4°C ± 0.5 °C a reasonable requirement for multi-year ice covers. Also perhaps relevant to paleo-climate proxy interpretation, at temperatures cold enough to maintain a multi-year ice cover, the summer temperatures could still be sufficient for a two-month long thawing period, including a month at about +5 °C Thus it is likely that many summer biological processes and some lake-water warming and mixing may still have been occurring beneath perennial ice-covers; core proxies have already indicated that such perennial ice-covers may have persisted for tens of thousands of years at various times within the 3.6 M years record.