My guess is that the field is roughly the same, as the peaks are not displaced when temperature is lowered. At 50K, the higher levels are maybe populated by thermal fluctuations? Which means less transitions are possible, leading to the observed spectrum with reduced peaks.