Gopal Penny
University of California; Berkeley
Recent Activity
ABSTRACT:
The Arkavathy watershed in southern India has exhibited considerable changes in inflow to major reservoirs, but explanations for these changes have been hampered by limited hydrological records in the watershed. In order to understand long-term hydrological change in this heavily managed watershed, we developed a spatially distributed dataset of surface water by generating time series of water extent in nearly 1700 man-made lakes, or tanks, over a 40-year period. Using an automated classification approach with subpixel unmixing, we classified water extent in each of the tanks in 40 Landsat images from 1973 to 2010, including 16 end-of-monsoon-season (December or January) images which characterized tank status after the rainy season. An additional 8 images in 2013 and 2014 were classified to analyze dry-season behavior and for validation. The classification results compared well with a reference dataset of water extent of tanks (R-squared=0.95). We also evaluated hydrological change in the watershed in 42 clusters of tanks by modeling water extent in the cluster on hydrological covariates and time. Based on a water balance argument, we inferred that any statistically significant results for the coefficient on the time covariate were indicative of long-term hydrological change. The results of this remote sensing classification for each of the tanks in the Arkavathy watershed are contained in this resource. The covariates and results from the statistical model are also included.
Contact
| (Log in to send email) |
| All | 0 |
| Collection | 0 |
| Resource | 0 |
| App Connector | 0 |
Created: Dec. 11, 2017, 5:55 a.m.
Authors: Gopal Penny · Veena Srinivasan · Sharachchandra Lele · Iryna Dronova · Sally Thompson
ABSTRACT:
The Arkavathy watershed in southern India has exhibited considerable changes in inflow to major reservoirs, but explanations for these changes have been hampered by limited hydrological records in the watershed. In order to understand long-term hydrological change in this heavily managed watershed, we developed a spatially distributed dataset of surface water by generating time series of water extent in nearly 1700 man-made lakes, or tanks, over a 40-year period. Using an automated classification approach with subpixel unmixing, we classified water extent in each of the tanks in 40 Landsat images from 1973 to 2010, including 16 end-of-monsoon-season (December or January) images which characterized tank status after the rainy season. An additional 8 images in 2013 and 2014 were classified to analyze dry-season behavior and for validation. The classification results compared well with a reference dataset of water extent of tanks (R-squared=0.95). We also evaluated hydrological change in the watershed in 42 clusters of tanks by modeling water extent in the cluster on hydrological covariates and time. Based on a water balance argument, we inferred that any statistically significant results for the coefficient on the time covariate were indicative of long-term hydrological change. The results of this remote sensing classification for each of the tanks in the Arkavathy watershed are contained in this resource. The covariates and results from the statistical model are also included.