Snow-cover seasonality in Kyrgyzstan: variation and change over 20 years (2001–2021) as observed by the MODIS Terra snow product

Abstract Snow seasonality is expected to change as a result of planetary warming. Yet, how the timings of snow seasons have already shifted in mountain environments can be counterintuitive. Here, we update our 2018 analyses of snow seasonality in Kyrgyzstan (KG) using more data and improved methods. Using two decades of MODIS snow-cover data, we investigated trends in four snow seasonality metrics: first date of snow (FDoS); last date of snow (LDoS); duration of the snow season (DoSS); and snow-covered days (SCDs). Our novel but conservative nonparametric approach analyzed trends within elevation ranges at two administrative levels: oblast and rayon. To attenuate the risk of false-positive trends, we reported as ‘notable’ only those significant ( p < 0.05) nonparametric trends that were also (i) strongly asymmetric (at least twice as many significant pixels in one direction) within an elevational range of the oblast or rayon and (ii) covering at least 5% of the elevation range in oblast or rayon and more than 10 pixels. We used both simple and advanced snow-cover metrics, with the latter exploiting the within-composite chronobyte detail available in the MOD10A2.061 product. We found notable trends to earlier FDoS below 3000 m in western KG and more SCD between 1500–3500 m in western KG. We also found the expected notable trends towards earlier LDoS at both oblast and rayon levels. In northwestern KG, DoSS was notably longer <3000 m, but notably shorter at 3500–4000 m. In our 2018 analyses, we found at the national level 160% more area trending to earlier FDoS than to earlier LDoS. Our comprehensive updated trend analysis was designed to be more conservative. It revealed more area (106%–130%) trending to earlier FDoS than earlier LDoS, reinforcing the counterintuitive finding of longer snow seasons starting earlier at lower elevations.