Utilizing MODIS for Snowmelt Runoff Analysis in the ORASECOM Basin - A Case Study of Senqu Catchment, Lesotho

Código

I-EBHE0014

Autores

Neo Makhalemele

Tema

WG 1.05: Comparative understanding of runoff generation processes

Resumo

Snow is the main form of precipitation in Lesotho during the winter season (April-July), but the lack of knowledge on snow hydrology hinders the proper planning in water allocation for water managers in Lesotho. This problem is compounded by limited snow data, due to the non-functional meteorological stations which are located through the mountainous terrain of the country. This study attempts to assess the amount of water contribution from snowmelt water equivalent (SWE) into the Senqu River. Cloud-free MODIS (terra/Aqua) MOD09A1 images were downloaded for the winter snow packing period (April-July) and the melting seasons (August ? September) for the years 2013-2017. The snow cover areas were mapped using the MOD09A1 MODIS satellite images. The study area was subdivided into different elevation zones of 500m increment intervals. Three conditions should be obtained to map snow cover areas such as, Normalised Difference Snow Index (NDSI) with a threshold reflectance value? 40%. Band-4 should have a reflectance value > 10% whilst Band -2 should have a reflectance > 11%. SRTM DEM (1arc-second, 30m) was used to extract the catchment characteristics such as elevation zones and hypsometric curve using ILIWS 3.31 academic GIS software package. The river discharges were modelled using Snowmelt Runoff Model (SRM). In-situ meteorological data in the form of average daily temperatures, precipitation and river discharge were used as inputs into the SRM model. The model was calibrated using datasets from the time-period 2000 and validated by 2013-2014. 2015-2017 dataset was used to substantiate the model performance. The performances was assessed through percentage volume difference (DV) and coefficient of determination (R2). The MODIS images were able to show an estimated hit accuracy of about 90% for snow cover mapping as observed from the observer?s records book at different meteorological stations. Snow cover area mapping results indicated that as much as 90% spatial cover can be achieved as shown for 27th July 2016 incident. The model results showed that as much as of runoff was the contribution by snowmelt water equivalent during the year 2016 snowmelt season contrary to that of (2015) and (2017). The difference between 2016 contribution and there rest was thought to be because slow-melting massive snowpack received in 2016, which was more than 4m in depth, the situation last seen in 2006. 2016 results were obtained by a model performance of R2 of 93% and Dv of 56% while 2015 R2 was 96% and Dv = -0.253% and finally 2017 model performance was R2 = 93% and Dv = 0, 6%.