Abstract
Irrigation is the most critical process for agriculture, but irrigation is the largest consumer of fresh water and causes the loss of large quantities because of the inaccuracy in crop water estimation. Our proposed system aims to improve irrigation management by estimating the amount of water needed by the crop accurately and reduces the number of meteorological parameters needed for such estimation. Detection of the reference crop evapotranspiration (ETo) is the most critical process in crop water estimation, that is considered through our proposed solution by implementing machine learning models using neural networks and linear regression to predict daily ETo using climate data like temperature, humidity, wind speed, and solar radiation. Comparing our system results with FAO-56 Penman-Monteith ET0 and cropwat8.0 software as benchmark, show that our proposed system is better than the linear regression model, in terms of determination coefficient (R^2)=.9677 and root mean square error(RMSE) =.1809, while the multiple linear regression model achieved determination coefficient (R^2)=.68 and root mean square error(RMSE) =3.01. Our system then used the predicted ETo and Crop coefficient (Kc) from FAO, to estimate crop evapotranspiration (ETc) for precision irrigation target.
Highlights
By 2050 the world's population will be 9 billion and demand for food will increase dramatically, Agriculture is the main source of food but consumes more than 70 percent of freshwater around the world. 25 percent from the water that used for irrigation is wasted because weakness in irrigation management
The system will monitor the parameters that that have impacts on crop and water needed using weather sensors distributed in the farm like temperature, humidity, wind speed and solar radiation, The second stage which is the prediction stage, the system will use data that collected from first stage to predict the amount of water crop needs based on machine learning algorithms [10]
The current approaches to set up water requirement prediction model depend only on climate information that coming from meteorological station, but there are some other problems with these approaches, like that all region doesn't have access to the precision weather information and if any sudden changes happen in climate, they cannot respond in real time [1]
Summary
By 2050 the world's population will be 9 billion and demand for food will increase dramatically, Agriculture is the main source of food but consumes more than 70 percent of freshwater around the world. 25 percent from the water that used for irrigation is wasted because weakness in irrigation management. Most farmers irrigate crop without knowing the amount of water that plant needs at any real level of precision and this lead to waste a lot of irrigation water the crop productivity is still weak. It is vital to develop a precision system to optimize irrigation management and predict the amount of water crop needs to decrease water consumption and increase crop productivity. Most farmers irrigate crop without knowing the amount of water that plant needs at any real level of precision and this lead to waste a lot of irrigation water and productivity weak So it is important to develop a precision system to optimize irrigation management and predict the amount of water that crop needs to decrease water consumption & increase productivity [8]. The system will monitor the parameters that that have impacts on crop and water needed using weather sensors distributed in the farm like temperature, humidity, wind speed and solar radiation, The second stage which is the prediction stage, the system will use data that collected from first stage to predict the amount of water crop needs based on machine learning algorithms [10]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
