Fig. 1.
Location map showing Lake Abashiri on (a) a wide scale and (b) a regional scale, showing three observation sites on the lake, two Japan Meteorological Agency (JMA) operational sites: Abashiri Meteorological Observatory and Memanbetsu Airport, the photograph spot at the top of Tento-san mountain and the near shore site for the past ice thickness record (Yobito). Modified from Ohata et al. (2016)
Fig. 2.
A schematic picture of the installation of the 12 sensor thermistor string for measuring air, dry snow, slush, ice and water temperature at site 3. The vertical line represents the sensor cable with thermistor sensor depths referenced to the slush/ice interface on 29 January 2016.
Table 1.
Model parameters and constants. Modified from Ohata et al. (2016).
Fig. 3.
Time series of daily (a) air temperatures and (b) snow depths at AMO in 2012/13, 2014/15 and 2015/16. The vertical dashed lines in (a) indicate the freeze-up dates and the horizontal dashed lines in (b) indicate the snow depths on the freeze-up dates.
Table 2.
Meteorological conditions from the freeze-up date to 20 February.
[i] Note: The mean air temperature and mean snow depth (given by the snow thickness increment from freeze-up date) are the averages at the two meteorological stations; Abashiri Meteorological Observatory and Memanbetsu Airport. Increase of snow depth just after freeze up is the mean increase of snow depth for 2 days after freeze-up.
Table 3.
Ice thickness and snow depth at sites 1, 3 and 6.
Fig. 4.
Model results of thickness evolution of total ice, SI and snow with observations shown by triangles (snow depth), squares (SI) and solid circles (total ice thickness) in 2012/13 (modified from Ohata et al., 2016), 2014/15 and 2015/16. The light blue curve indicates the snow depth on land obtained from the daily meteorological snow depth data by taking a 5-d running mean (the day and 4 d ahead). The purple, red and blue curves show the snow depth on the lake, SI thickness and total ice thickness predicted by the model, respectively. Freeze-up dates used in the model and break-up dates calculated by the model are indicated, along with those estimated with MODIS images (in parentheses).
Fig. 5.
Observations of ice thickness (at site 3 on 19 February 2013, 24 February 2015 and 19 February 2016) and the ice thicknesses on 20 February 2013, 2015 and 2016, predicted by the model for three cases; (1) snow accumulates with SI formation (control run), (2) snow accumulates without SI formation and (3) no snow accumulates.
Fig. 6.
Schematic pictures showing the effect of snow and snow ice formation on the lake ice thickness for (a) the same snow amount and different air temperatures (cited from Ohata et al., 2016), and (b) different snow amounts and the same air temperature.
Fig. 7.
The correlation coefficient between modelled and observed lake ice thickness at Yobito on 15 January and 15 February for 2005/06–2015/16. Diamonds and squares show ice thicknesses on 15 January and February, respectively. Correlation coefficients and p-values are shown by r and p in the figure. Yobito data are from the Abashiri tourism association.
Fig. 8.
The relationship between total ice thickness on 15 February and mean air temperature or mean snow depth from freeze-up date to 15 February, recorded at the Abashiri Meteorological Observatory for 2000/01–2015/16. The solid straight lines are the regression lines of best fit. Correlation coefficients and p-values are shown by r and p in the figure.
Fig. 9.
Freeze-up and break-up dates from 1961/62 to 2015/16, determined by the model and MODIS images. Error bars show the time ranges estimated for freeze-up and break-up dates from MODIS images. Diamonds and squares show freeze-up dates used for the model and break-up dates calculated by the model, respectively.
Fig. 10.
The thicknesses of total ice, SI and CI on 15 February, determined by the model. The black solid line, dashed-dotted blue line and dashed red line show total ice thickness, CI thickness and SI thickness predicted by the model, respectively. The black solid straight line is the regression line of best fit for total ice thickness. The trend, shared variance (r 2 ), correlation coefficient (r) and p-value (p) are shown in the figure.
Fig. 11.
The relationships between total ice thickness (hi), SI thickness (hsi) and CI thickness (hci) on 15 February for 55 years: (a) hi vs. hsi, (b) hi vs. hci and (c) hsi vs. hci. Correlation coefficients and p-values are shown by r and p in the figure.
Fig. 12.
Time series of temperatures of (a) air, (b) snow, ice and water at site 3 from 29 January to 11 March 2016.
Fig. 13.
Time series of temperature gradients within snow and ice at site 3 from 29 January to 11 March 2016.
Fig. 14.
A schematic picture of snow ice growth at site 3. The vertical axis represents the heights with sensor names referenced to the ice/slush interface on 29 January. The horizontal axis represents time. The red line represents the boundaries between SI and slush or dry snow. The green line represents the snow surface. The red dashed line represents the supposed boundary between SI and dry snow which was not observed. The black dashed vertical and horizontal lines represent the dates and the corresponding heights of slush, respectively.
Fig. 15.
Meteorological conditions after freeze-up date to 15 February for (a) the mean air temperature and (b) the mean snow depth. The black solid straight line is the regression line of best fit. The trends, shared variances (r 2 ) and p-values are shown in the figure.
Fig. 16.
The relationship between total ice thickness on 15 February estimated by the model, and the mean air temperature or the mean snow depth from freeze-up date to 15 February for 1961/62–2015/16, recorded at the Abashiri Meteorological Observatory. The solid straight lines are the regression lines of best fit. Correlation coefficients and p-values are shown by r and p in the figure.
Fig. 17.
Comparisons of anomaly time series for ice phenology parameters and air temperature from 1961/62 to 2015/16. All valuables are expressed as normalized anomalies (standard deviations of the measures being compared): (a) freeze-up date and the average of November and December temperatures, (b) break-up date and March temperatures and (c) ice cover duration and the average of November to March temperatures. Annual values are indicated by thin lines and eight-year running means by heavy lines; ice phenology parameters are blue and air temperatures are red. r 2 is presented for the linear relationship between the ice phenology parameters and the corresponding air temperatures.