Non-contact visual soil moisture content estimation

Watson, P. (2014) Non-contact visual soil moisture content estimation. [USQ Project]


Download (6MB) | Preview


Drought in recent years has highlighted the importance of maintaining a sustainable water resource. Improvements in irrigation management can significantly increase water use efficiency and crop productivity for Australian agriculture. Measurement of Soil Moisture Content (SMC) is essential for improving irrigation management. Existing commercially-available SMC sensors require contact with the soil and measure only a single fixed point in a field. However, there can be significant spatial variability in soil properties and SMC within a field, and installation of multiple SMC sensors within a field is often not practical or economical. Non-contact methods reported in the literature for SMC estimation include satellite imagery of soil and plants. Satellite imagery approaches capture spectral bands in the visual, infrared and microwave wavelengths and then extract crop vigour to estimate SMC. However, this technology has a limited spatial resolution (30m2) and temporal resolution (every 2-3 weeks). An alternative approach uses a ground-based camera that can be moved around the field on ground-based or aerial vehicles as required, providing high spatial and temporal resolution
SMC estimation. A camera-based estimation system has been developed. Red and near infrared images of plants are processed using MATLAB Image Processing Tool box and ColorWorker software. A MATLAB program has been developed that performs the following image analysis: (i) overlays images of different spectral bands; (ii) selects key regions in the visual image; (iii) selects key regions in the infrared image; and (iv) calculates reflectance in the visible and infrared bands. Multiple regression analysis has been conducted to analyse the calculated reflectance and develop a model that estimates SMC. The camera and image analysis system has been evaluated on chamomile, lettuce and lucerne plants. These plants were grown under three irrigation levels (20%, 30% and 40% VWC) and two soil types (loam and sand). Each sample was replicated twice, giving a total of 36 samples. Daily digital images were taken of plants with band pass filters in red and near infrared bands. An on-site weather station provides micro climate data which is used calibrate the models. Three spectral responses were derived from the images: (i) chlorophyll a/b ratio - Chl(a/b); (ii) Normalised Difference Vegetation Index - NDVI; and (iii) near infrared at 850 ηm - IR850. A soil moisture estimation model was derived for each plant and soil type which showed a significant correlation between one of the spectral responses of the plant and SMC. The Root Mean Squared Error RMSE value was used to test the accuracy of the estimation models. There were nine models which had a RMSE less than 5%. Two lucerne/sand models with NDVI response had RMSE value of 1.39% and 1.49% Volumetric Water Content (VWC), both replicates in each model were within 0.40%. Lucerne/sand was also sensitive in IR850 response, with a RMSE for replica 1 and 2 of 2.04% and 2.89% VWC respectively. Chamomile/loam was sensitive to IR850 response across all three irrigation levels, all RMSE values were below 2.89% VWC for the data obtained during August. Lettuce/loam was sensitive to IR850 response for data obtained during July. Replicates 1 and 2 had RMSE values of 1.9% and 2.56% VWC. There were no correllations for the chl(a/b) index. This research has shown that in some conditions SMC can be estimated from plant spectral response (NDVI and IR850) for chamomile, lettuce and lucerne. Further research is needed to understand the effects of what plant nutrition and disease have on the spectral response and SMC estimation.

Statistics for USQ ePrint 27289
Statistics for this ePrint Item
Item Type: USQ Project
Item Status: Live Archive
Additional Information: Bachelor of Electrical/Electronic Engineering project.
Faculty/School / Institute/Centre: Historic - Faculty of Health, Engineering and Sciences - School of Mechanical and Electrical Engineering (1 Jul 2013 - 31 Dec 2021)
Supervisors: McCarthy, Alison
Date Deposited: 09 Sep 2015 05:18
Last Modified: 09 Mar 2016 05:35
Uncontrolled Keywords: irrigation management; soil moisture content; non-contact soil moisture sensors; satellite imagery in agriculture; infrared images; microwave wavelengths; soil testing in agriculture
Fields of Research (2008): 09 Engineering > 0907 Environmental Engineering > 090703 Environmental Technologies
Fields of Research (2020): 40 ENGINEERING > 4011 Environmental engineering > 401102 Environmentally sustainable engineering

Actions (login required)

View Item Archive Repository Staff Only