Monitoring the land and water use dynamics at basin level using time series of Landsat 8 data
Abstract
An estimated forty percent of the global population are experiencing water scarcity at different magnitudes. One of the target of Sustainable Development Goal (SDG6) is to substantially reduce the number of people suffering from water scarcity by improving water use efficiency. The biggest share of water (around 70 %) is often allocated for irrigation. Hence monitoring the spatial and temporal dynamics of water availability, irrigated area and water use at basin level will play a big role in ensuring proper allocation of water in a sustainable way. In this study, we developed a remote sensing based land and water use monitoring workflow using time series of Landsat 8 (L8) data and open source tools for Mashhad basin in Iran. The entire basin is semi-arid with a total surface area of 16779 km2 and consists of city of Mashhad which is the second biggest city in Iran. For Mashhad basin, we estimated the irrigated area land use dynamics and the corresponding water use for three crop years starting from October 2013 to September 2016.
We used machine learning based Random Forest (RF) algorithm to retrieve accurate irrigated area per cropping year and a remote sensing based Surface Energy Balance Model (SEBAL) to estimate Actual EvapoTranspiration (ETa) for each year over the derived irrigated areas. A decision tree model based on RF algorithm is developed for entire Mashhad basin for the year 2016 using time series of L8 data and training samples. The developed RF model is then applied to other years to generate irrigated area maps which changes dynamically over different crop years. An accuracy of 87 % was reported for the land use irrigated maps comparing to the field observations. Locally weighted regression approach was used to fill the gaps in monthly ETa due to cloud coverage. Following the gap filling, monthly and annual actual ET maps at a spatial resolution of 30 m for the three crop years were developed and analysed to estimate the water balance thereby understand the water use dynamics. The total irrigated area over each cropping year estimated from this analysis is 1796.16 Km2, 1581.7 Km2 and 1578.26 Km2 for the cropping years 2013-14, 2014-15 and 2015-16 respectively. We found that the annual evapotranspiration from irrigated areas is estimated to be between 450 - 750 mm matching the previous reports on the same using field based estimation.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFM.B31I2605P
- Keywords:
-
- 1632 Land cover change;
- GLOBAL CHANGEDE: 1640 Remote sensing;
- GLOBAL CHANGEDE: 1855 Remote sensing;
- HYDROLOGYDE: 1942 Machine learning;
- INFORMATICS