It is essential to have the soil moisture retention at the optimal level in order to maintain high yields in oil palm estates. Furthermore, conventional methods for determining soil moisture are difficult, time-consuming, and challenging in the rural estate areas. In this study, synthetic aperture radar (SAR), L-band images, and in-situ observations were conducted at an oil palm plantation to employ water cloud model (WCM) inversion for retrieving soil moisture from HH (Horizontal-Horizontal) and HV (Horizontal-Vertical) polarized data. WCM was evaluated by comparing leaf area index (LAI), leaf water area index (LWAI), and normalized plant water content (NPWC), to understand the effects of vegetation on backscattering coefficients. Adding on, neural network (NN) technique was employed to understand capabilities of soil moisture retrievals using mentioned data. Effects of vegetation in the WCM and NN models were then investigated using the k-fold cross validation method to understand the difference in the in-situ observations and modelled results. The results demonstrated that HV polarization efficiently approximated the backscatter coefficient compared to HH polarization, while the best fit was achieved by using the LAI as a vegetation descriptor in the WCM model with an accuracy of at least R2=0.9460 with RMSE of 0.036m3/m3 whereas the NN model was able to improvise soil moisture content with R2=0.9638 and RMSE of 0.012m3/m3.

Figure 1: The concept of retrieving soil moisture from oil palm estates using WCM approach

Shashikant, V., Shariff, A.R.M., Wayayok, A., Kamal, M.R., Lee, Y.P., Takeuchi, W. Strategic Short Note: Comparing Soil Moisture Retrieval from Water Cloud Model and Neural Network Using PALSAR-2 for Oil Palm Estates, In: Ahamed, T. (eds) IoT and AI in Agriculture. Springer, Singapore ISBN 978-981-19-8112-8

Tarikh Input: 25/08/2023 | Kemaskini: 04/04/2024 | ainzubaidah