Estimation of persistence on glacial lakes in tropical Andes mountain-range with 2000-2020 period LANDSAT series images

Peruvian mountain-range glaciers are characterized by the presence of numerous lakes of glacial origin, whose dynamics show a great temporal-spatial variability due to factors such as glacial melting and precipitation of different types, seasons, and intensities, engaging also river flow, usually for the benefit of population settlements. Therefore, it is important to determine its continuity to consider them permanent resources of water. Previously, the evaluation of this parameter was made traditionally, by looking at optical satellite imagery. However, this process ends up being too long and complicated as there are up to 3000 lakes in some mountain ranges.

We propose a methodology with the objective of estimating the temporal persistence of glacial lakes mainly performed in Google Earth Engine, convenient for the flexibility, data-size issues, and quick computations of statistical approach. This process is based on LANDSAT 7 and 8 normalized difference water index (NDWI) time series data products, comprised of 252 images of the Ampato glacier mountain-range across the calendar years 2000-2020. Initially, we extract the NDWI values for each polygon -from the INAIGEM 2020 glacial lake Inventory- and image and apply different NDWI thresholds and ways to mean them. Finally, we get a representative conversion of the value to mark their existence and do the percentage calculations over the evaluation period.

NDWI threshold of 0.05 and median values were chosen to obey previous evaluations and have tight results. We observe that between 86 and 99% of images were available for the 518 polygons of this area, indicating suitability to support subsequent conclusions. The final persistence values vary between 50% and 99% for lakes greater than 5000 m², while lesser lakes present values between 25 and 75% of persistence during the evaluation period, corresponding to weather modulating factors in a shorter scale such as seasonality, ENSO events, extreme precipitation, etc. The presented investigation could have relevant applications from water management, ecology, tourism, to climate investigation, as a way to sophisticate the processes of a more exact and specific glacial lake Inventory in Peru or other parts of the sphere.