Assessing inaccuracies in remotely sensed irrigation maps across the High Plains Aquifer
Abstract
As concerns over water consumption and aquifer depletion become more publically recognized issues, understanding irrigation extent and practices becomes more important. Map products that attempt to classify croplands based on physical and spectral attributes have had varying success. To better understand the sources of inaccuracies with remotely sensed irrigation maps, we evaluated a range of classification products that cover the High Plains Aquifer in the central United States using an original, consistent validation dataset. We then compared map accuracy with physical covariates (soil characteristics and precipitation) and methodological approaches (spatial resolution, imagery timing, and training data specificity) to identify factors associated with increased success. Finally, we used USGS water use estimates to understand how mapping accuracy affected estimates of aquifer use and depletion. We found that accuracy increased with spatial resolution of the product and decreased for larger domains. Areas with more growing season precipitation had decreased accuracy due to reduced water stress for dryland crops. High moisture capacity soils had similar relationships as those for higher precipitation regions. While current remotely sensed irrigation maps have relatively high reported accuracies (70-90%), remaining prediction errors can translate into notably different estimates of both the volume and spatial distribution of water use. Understanding the underlying sources of inaccuracies in irrigation maps can lead to improvement in future more accurate products that quantify water use across regional aquifers.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2016
- Bibcode:
- 2016AGUFM.H51H1589R
- Keywords:
-
- 3305 Climate change and variability;
- ATMOSPHERIC PROCESSESDE: 1655 Water cycles;
- GLOBAL CHANGEDE: 1855 Remote sensing;
- HYDROLOGYDE: 4313 Extreme events;
- NATURAL HAZARDS