Other: Agriculture & Water Sustainability

Water Stress Detection using Aerial & Metrological Data(Agri-Bird)

AP008

Muhammad Abdullah Shahzad (National University of Science and Technology, Islamabad)

Oct 31, 2021 3949 views

Water Stress Detection using Aerial & Metrological Data(Agri-Bird)

Water is essential in agriculture. Farms use it to grow fresh produce and to sustain their livestock. Major environmental functions and human needs critically depend on water. In regions of the world affected by water scarcity economic activities can be constrained by water availability, leading to competition both among sectors and between human uses and environmental needs.

According to a 2017-18 government survey, agriculture contributes to 18.9% of the GDP and uses up 42.3% of the labor force in Pakistan. But with agriculture using up about 90% of Pakistan’s water supply, and Pakistan’s water crisis threatening to exhaust the country’s water resources by 2040, there is a dire need for solutions that help in the efficient use of water in agriculture, and farming in particular.

To combat this problem and provide a sustainable mechanism to farmers, we propose an aerial collection and soil-sampling data framework that will lead to sustainable, precise, secure, and efficient farming. Our solution will focus on the water-stress or drought-stress of plants.

Water stress refers to the water deficit in plants and has shown to be a very useful piece of information in farming. In addition to being a good predictor for the yield of the plantation, water stress also allows us to respond timely to areas that are under-watered or over-watered. Of course, water stress is most valuable as information for planning irrigation, but it can also be a very decent measure of areas that are at risk of wildfires.

Our solution proposes to mount an FPGA to the aerial unit where it will be collecting data with the help of modules, subsequently process it on the edge, and then transfer all the relevant data to the cloud for further processing and analysis. In order to give our results more credibility, we will also be collecting some soil-sampled data and combining it with the aerial data to give us our final results in the cloud. Our results will aim to give accurate predictions, useful suggestions to farmers, routing data for irrigation channels, and warnings for risks and disasters.

Project Proposal


1. High-level project introduction and performance expectation

Purpose of Design:

Water is a critical input for agricultural production and plays an important role in food security. Irrigated agriculture represents 20 percent of the total cultivated land and contributes 40 percent of the total food produced worldwide.

According to a 2017-18 Government of Pakistan survey, agriculture contributes to 18.9% of the GDP and uses up 42.3% of the labor force in Pakistan. But in recent times, agriculture, and farming, in particular, faces the challenge of providing a yield not only sufficient to meet the requirements of an increasing population but also for export with a competitive quality of produce. To make matters worse, Pakistan faces a fatal water crisis, with agriculture using roughly 90% of the country’s water resources.

                                                                           

                                                              Map highlighting expecting water availability in Pakistan during 2018. — Courtesy: PMD

In order to combat these challenges and provide a solution to the average farmer, we propose a Water Stress Prediction Model using Aerial & Metrological Data for precise, efficient, secure, and sustainable farming. In this framework, we will focus on the water-stress or drought-stress characteristics of plants which is caused by a water deficit.

We have decided on the characteristic of water stress because it will help us prevent the following:

          1. Loss in produce from water deficit.

          2. Loss of soil nutrients from excessive watering.

          3. Poor regulation of irrigation.

          4. Wildfires

 

Application Scope:

Keeping in mind that the middle-scale farmer has a cultivating land of ½ - 1 acre in a remote area, soil sampling at various points would be labor-intensive and inaccurate, and satellite imagery would not be readily available and very expensive. Therefore, we feel that an Aerial approach to collect data coupled with Soil-Sampling of affected areas will be the most suitable use of sensors. The FPGA mounted onto an Aerial Device will periodically collect the following data when making rounds in the crop fields:

      1. Images

      2. Meteorological/Weather Data:

                                    -> Temperature

                                    -> Humidity

                                    -> Light/Solar Radiation

From the soil, we will be sampling the following:

       1. Soil moisture

       2. pH Level

This collected data, in turn, will help us determine these parameters throughout our covered area:

        1. Soil Moisture

        2. Plant water stress from CNN's

        3. Evapotranspiration rate

The above three factors will be sent to the cloud using the Cloud Connectivity Kit and will be used to create final analytics of water stress across the covered area.

 

Targeted End User:

 Our solution will be targeting the following:

         1. Small and Medium-scale farmers

         2. Relevant authorities for better irrigation planning

         3. Safety organization for Wildfire & Droughts Prediction

 

Use of Intel FPGA:

Since a crucial part of our design is collecting aerial data and processing this data immediately, we have decided to mount our FPGA device onto an Aerial Device where it will perform the following important functions:

         1. Collect field data for temperature, humidity, and light.

         2. Accelerate a CNN implementation for processing aerial image data collected using the camera module.

         3. Perform real-time low-latency inference on images taken by the camera module.

         4. Transfer meteorological and processed image data to the cloud using the cloud connectivity kit.

 

Furthermore, the Intel FPGA has the following advantages:

         -> Reconfigurable logic allows testing multiple CNN models according to collected image data.

         -> Low power consumption which is advantageous for remote usage.

         -> Accelerated hardware gives lower latency and meets real-time processing requirements.

2. Block Diagram

3. Expected sustainability results, projected resource savings

The expected sustainability results and projected resource savings of this project are as follows:

1. Reducing Water usage for irrigation: 
The system will reduce water usage for irrigation by monitoring water stress in specific areas. This information can be used for the prediction of the best time for irrigation. This can reduce water usage by 25-35%. 

2. Improving Crop Yield:
The system improves crop yield by giving information on water stress levels at different time instances, thus preventing drought stress on the crops.

3. Project Resource Savings:
This project mostly relies on remote sensing techniques to sense solar radiation or analyze aerial images which, in turn, is used for the estimation of soil water evaporation and evapotranspiration. Remote sensing techniques estimate the water stress level of a large area as compared to point sensing techniques which cover a very small area. Therefore, this project is efficient in terms of resource savings as it deploys fewer on-field sensors to cover a comparatively large area.

References:

1. Tseng, D.; Wang, D.; Chen, C.; Miller, L.; Song, W.; Viers, J.; Vougioukas, S.; Carpin, S.; Ojea, J.A.; Goldberg, K. Towards Automating Precision Irrigation: Deep Learning to Infer Local Soil Moisture Conditions from Synthetic Aerial Agricultural Images. In Proceedings of the 2018 IEEE 14th International Conference on Automation Science and Engineering (CASE), Munich, Germany, 20–24 August 2018

2. Sobayo, R.; Wu, H.; Ray, R.; Qian, L. Integration of Convolutional Neural Network and Thermal Images into Soil Moisture Estimation. In Proceedings of the 2018 1st International Conference on Data Intelligence and Security (ICDIS), South Padre Island, TX, USA, 8–10 April 2018

3. Song, X.; Zhang, G.; Liu, F.; Li, D.; Zhao, Y.; Yang, J. Modeling Spatio-temporal distribution of soil moisture by deep learning-based cellular automata model. J. Arid Land 2016

4. Cai, Y.; Zheng, W.; Zhang, X.; Zhangzhong, L.; Xue, X. Research on soil moisture prediction model based on deep learning. PLoS ONE 2019

5. Saggi, M.K.; Jain, S. Reference evapotranspiration estimation and modeling of the Punjab Northern India using deep learning. Comput. Electron. Agric. 2019

6. An, J.; Li, W.; Li, M.; Cui, S.; Yue, H. Identification and Classification of Maize Drought Stress Using Deep Convolutional Neural Network. Symmetry 2019

4. Design Introduction

5. Functional description and implementation

6. Performance metrics, performance to expectation

7. Sustainability results, resource savings achieved

8. Conclusion

1 Comments



Please login to post a comment.

Devarapalli Durga Praveen Kumar

What is the operating system you are using in your project?

Feb 24, 2022 01:26 PM