Skip to main content
ARS Home » Plains Area » Bushland, Texas » Conservation and Production Research Laboratory » Soil and Water Management Research » Research » Publications at this Location » Publication #287237
Title: Estimating missing hourly climatic data using artificial neural network for energy balance based ET mapping applications

Author
item CEMEK, BILAL - Ondokuz Mayis University
item KOKSAL, SELIM - Ondokuz Mayis University
item Gowda, Prasanna
item Howell, Terry
item CETIN, SAKINE - Ondokuz Mayis University

Submitted to: Irrigation Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/2/2016
Publication Date: 4/6/2017
Citation: Cemek, B., Koksal, S., Gowda, P., Howell, T.A., Cetin, S. 2017. Estimating missing hourly climatic data using artificial neural network for energy balance based ET mapping applications. Meteorological Applications. 24(3):457-465.

Interpretive Summary: Accurate hourly weather data and reference evapotranspiration (ET) are crucial input for implementing surface energy balance based ET models. In Turkey, hourly climatic data may not be available in many locations either due to higher costs or from poor maintenance. In this study, we used an advanced statistical learning method to estimate missing or not-measured climatic variables. Results indicated that the advanced statistical learning method used in the study has the potential to be used to estimate but missing/not-measured hourly climatic data from their daily data.

Technical Abstract: Remote sensing based evapotranspiration (ET) mapping has become an important tool for water resources management at a regional scale. Accurate hourly climatic data and reference ET are crucial input for successfully implementing remote sensing based ET models such as Mapping ET with internal calibration (METRIC) and surface energy balance algorithm for land (SEBAL). In Turkey, hourly climatic data may not be available at in all locations either due to cost constraints or due to equipment malfunctioning. In this study, the artificial neural network (ANN) technique was used to estimate missing and not-measured hourly climatic data and alfalfa reference ET for an agriculturally important semi-humid Bafra Plains located in northern Turkey. Modeled and actual climatic and reference ET were used to derive ET maps from two Landsat 5 Thematic Mapper data acquired on September 2, 2009 and August 4, 2010. The METRIC algorithm was applied to derive hourly ET maps. Accuracy assessment of the METRIC-derived ET maps indicated that climatic data and reference ET estimated through ANN could be used for accurately mapping ET where hourly climatic data are missing or not measured.