Sains Malaysiana 55(10)(2026): 1690-1702

http://doi.org/10.17576/jsm-2026-5510-08

 

Predictive Reliability of Denoised QM-Corrected Satellite-Derived Photosynthetically Active Radiation (PAR) under Humid Tropical Conditions

(Kebolehpercayaan Ramalan bagi Sinaran Aktif Fotosintesis (PAR) Terbitan Satelit yang Telah Dibuang Hingar dan Dibetulkan secara QM di bawah Keadaan Tropika Lembap)

 

ROHAIZA DAUD^1,2,*, FARRAH MELISSA MUHARAM¹, RADHWANE DERRAZ³, NGAI PAING TAN⁴ & NOR AZURA HUSIN⁵

 

¹Department of Agriculture Technology, Faculty of Agriculture, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia

²Centre for Pre-University Studies, Universiti Malaysia Sarawak (UNIMAS), 94300 Kota Samarahan, Sarawak, Malaysia

³School of Computing, College of Arts and Sciences, Universiti Utara Malaysia, 06010 Sintok, Kedah, Malaysia

⁴Department of Land Management, Faculty of Agriculture, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia

⁵Department of Computer Science, Faculty of Computer Science and Information Technology, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia

 

Received: 12 June 2026/Accepted: 20 September 2026

 

*Corresponding author; email: drohaiza@unimas.my

 

Abstract

Satellite-derived photosynthetically active radiation (PAR) products provide valuable information for environmental monitoring and agricultural applications, but residual fluctuations may remain even after bias correction, potentially affecting predictive reliability. This study evaluated the influence of multiple denoising approaches, including decomposition-, filtering-, smoothing-, and learning-based methods, on the predictive performance of quantile mapping (QM)-corrected satellite-derived PAR under humid tropical conditions in Peninsular Malaysia. MODIS MCD18A2 Version 061 PAR data were obtained for 14 meteorological stations between 2015 and 2023 and corrected using QM prior to denoising. Predictive performance was assessed using five regression models. Most denoising approaches improved predictive performance relative to the baseline QM-corrected dataset, although the magnitude of improvement varied considerably among methods. Median filtering paired with Random Forest produced the highest mean R² (0.748), with a mean MSE of 0.293. Although LSTM yielded a lower numerical MSE (0.004), its lower mean R² (0.542) and nearly flat reconstruction indicated variance compression rather than faithful preservation of PAR dynamics. Median filtering was therefore selected as the most reliable overall method. In contrast, Empirical Mode Decomposition (EMD) exhibited inconsistent performance across stations and substantially larger prediction errors. The findings demonstrate that denoising can substantially enhance the predictive reliability of satellite-derived PAR. More importantly, the results suggest that predictive reliability is influenced more by effective signal preservation than by algorithmic complexity alone. Under humid tropical conditions, relatively simple denoising approaches may provide more robust improvements than computationally intensive alternatives for PAR-based environmental modelling applications.

Keywords: Denoising; machine learning; MODIS MCD18A2; photosynthetically active radiation; quantile mapping

 

Abstrak

Produk sinaran aktif fotosintesis (PAR) berasaskan satelit menyediakan maklumat penting untuk pemantauan alam sekitar dan aplikasi pertanian, namun lebihan turun naik masih boleh wujud walaupun selepas pembetulan bias, sekali gus berpotensi menjejaskan kebolehpercayaan ramalan. Kajian ini menilai pengaruh pelbagai pendekatan penyahhingaran, termasuk kaedah berasaskan penguraian, penapisan, pelicinan dan pembelajaran, terhadap prestasi ramalan PAR berasaskan satelit yang telah diperbetulkan menggunakan pemetaan kuantil (QM) di bawah keadaan tropika lembap di Semenanjung Malaysia. Data PAR MODIS MCD18A2 Versi 061 diperoleh bagi 14 stesen meteorologi untuk tempoh 2015 hingga 2023 dan diperbetulkan menggunakan QM sebelum proses penyahhingaran dijalankan. Prestasi ramalan dinilai menggunakan lima model regresi. Kebanyakan kaedah penyahhingaran meningkatkan prestasi ramalan berbanding set data asas yang telah diperbetulkan menggunakan QM, walaupun tahap peningkatan berbeza antara kaedah. Penapisan median yang digandingkan dengan hutan rawak menghasilkan purata R² tertinggi (0.748) dan purata MSE sebanyak 0.293. Walaupun LSTM menghasilkan MSE yang lebih rendah (0.004), purata R² yang lebih rendah (0.542) dan pembinaan semula hampir mendatar menunjukkan pemampatan varians, bukannya pengekalan dinamik PAR. Oleh itu, penapisan median dipilih sebagai kaedah yang paling boleh dipercayai secara keseluruhan. Sebaliknya, Penguraian Mod Empirik (EMD) menunjukkan prestasi yang tidak tekal antara stesen serta menghasilkan ralat ramalan yang lebih tinggi. Keputusan kajian menunjukkan bahawa penyahhingaran dapat meningkatkan kebolehpercayaan ramalan PAR berasaskan satelit. Yang lebih penting, kebolehpercayaan ramalan didapati lebih dipengaruhi oleh pemeliharaan maklumat temporal yang bermakna berbanding kerumitan algoritma semata-mata. Di bawah keadaan tropika lembap, pendekatan penyahhingaran yang lebih ringkas berpotensi memberikan peningkatan yang lebih mantap bagi aplikasi pemodelan alam sekitar berasaskan PAR.

Kata kunci: MODIS MCD18A2; pembelajaran mesin; pemetaan kuantil; penyahhingaran; sinaran aktif fotosintesis

 

REFERENCES

Akitsu, T.K., Nasahara, K.N., Ijima, O., Hirose, Y., Ide, R., Takagi, K. & Kume, A. 2022. The variability and seasonality in the ratio of photosynthetically active radiation to solar radiation: A simple empirical model of the ratio. International Journal of Applied Earth Observation and Geoinformation 108: 102724. https://doi.org/10.1016/j.jag.2022.102724

Bailey, M.D., Nychka, D.W., Sengupta, M., Yang, J., Xie, Y., Habte, A. & Bandyopadhyay, S. 2025. Adapting quantile mapping to bias correct solar radiation data. Solar Energy 288: 113220. https://doi.org/10.1016/j.solener.2024.113220

Barth, A., Brajard, J., Alvera-Azcárate, A., Mohamed, B., Troupin, C. & Beckers, J-M. 2024. Ensemble reconstruction of missing satellite data using a denoising diffusion model: Application to chlorophyll a concentration in the Black Sea. Ocean Science 20: 1567-1584. https://doi.org/10.5194/os-20-1567-2024

Breiman, L. 2001. Random forests. Machine Learning 45(1): 5-32. https://doi.org/10.1023/A:1010933404324

Breiman, L. 1996. Bagging predictors. Machine Learning 24(2): 123-140. https://doi.org/10.1007/BF00058655

Chicco, D., Warrens, M.J. & Jurman, G. 2021. The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation. PeerJ Computer Science 7: e623. https://doi.org/10.7717/peerj-cs.623

Cooley, J.W. & Tukey, J.W. 1965. An algorithm for the machine calculation of complex Fourier series. Mathematics of Computation 19(90): 297-301. https://doi.org/10.2307/2003354

Finazzi, F., Rodeschini, J. & Tedesco, L. 2025. Discussion on assessing predictability of environmental time series with statistical and machine learning models. Environmetrics 36: e2900. https://doi.org/10.1002/env.2900

Friedman, J.H. 2001. Greedy function approximation: A gradient boosting machine. Annals of Statistics 29: 1189-1232. https://doi.org/10.1214/aos/1013203451

Goodfellow, I., Bengio, Y. & Courville, A. 2016. Deep Learning. Cambridge: MIT Press.

Gould, R.W., Ko, D.S., Ladner, S.D., Lawson, T.A. & Macdonald, C.P. 2019. Comparison of satellite, model, and in situ values of photosynthetically available radiation (PAR). Journal of Atmospheric and Oceanic Technology 36(4): 535-555. https://doi.org/10.1175/JTECH-D-18-0096.1

He, Q., Chun, K.P., Tan, M.L., Dieppois, B., Liew, J., Klaus, J., Fournier, M., Massei, N. & Yetemen, O. 2021. Tropical drought patterns and their linkages to large-scale climate variability over Peninsular Malaysia. Hydrological Processes 35(9): e14356. https://doi.org/10.1002/hyp.14356

Hird, J.N. & McDermid, G.J. 2009. Noise reduction of NDVI time series: An empirical comparison of selected techniques. Remote Sensing of Environment 113(1): 248-258. https://doi.org/10.1016/j.rse.2008.09.003

Hochreiter, S. & Schmidhuber, J. 1997. Long short-term memory. Neural Computation 9(8): 1735-1780. https://doi.org/10.1162/neco.1997.9.8.1735

Huang, N.E., Shen, Z., Long, S., Wu, M.C., Shih, H.H., Zheng, Q., Yen, N.C., Tung, C.C. & Liu, H.H. 1998. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis. Proceedings of the Royal Society A 454(1971): 903-995. https://doi.org/10.1098/rspa.1998.0193

Kalman, R.E. 1960. A new approach to linear filtering and prediction problems. Journal of Basic Engineering 82(1): 35-45. https://doi.org/10.1115/1.3662552

Katiraie-Boroujerdy, P.S., Rahnamay Naeini, M., Akbari Asanjan, A., Chavoshian, A., Hsu, K.L. & Sorooshian, S. 2020. Bias correction of satellite-based precipitation estimations using quantile mapping approach in different climate regions of Iran. Remote Sensing 12(12): 2102. https://doi.org/10.3390/rs12132102

Kumari, S. & Singh, S.K. 2023. Machine learning-based time series models for effective CO₂ emission prediction in India. Environmental Science and Pollution Research 30: 116601-116616. https://doi.org/10.1007/s11356-022-21723-8

Liu, J., Cai, Y., Pei, X. & Yu, X. 2025. Advances in research and application of techniques for measuring photosynthetically active radiation. Remote Sensing 17(10): 1765. https://doi.org/10.3390/rs17101765

Lu, X., Liu, R. & Liang, S. 2007. Removal of noise by wavelet method to generate high quality temporal data of terrestrial MODIS products. Photogrammetric Engineering & Remote Sensing 73(10): 1129-1139.

Mallat, S.G. 1989. A theory for multiresolution signal decomposition: The wavelet representation. IEEE Transactions on Pattern Analysis and Machine Intelligence 11(7): 674-693. https://doi.org/10.1109/34.192463

Mariethoz, G., Linde, N., Jougnot, D. & Rezaee, H. 2015. Feature-preserving interpolation and filtering of environmental time series. Environmental Modelling & Software 72: 71-76. https://doi.org/10.1016/j.envsoft.2015.07.001

McCree, K.J. 1972. Action spectrum, absorptance and quantum yield of photosynthesis in crop plants. Agricultural Meteorology 9: 191-216. https://doi.org/10.1016/0002-1571(71)90022-7

Pelletier, C., Valero, S., Inglada, J., Champion, N., Marais Sicre, C. & Dedieu, G. 2017. Effect of training class label noise on classification performances for land cover mapping with satellite image time series. Remote Sensing 9(2): 173. https://doi.org/10.3390/rs9020173

Priyadarshi, N., Chowdary, V.M., Srivastava, Y.K., Das, I.C. & Jha, C.S. 2018. Reconstruction of time series MODIS EVI data using de-noising algorithms. Geocarto International 33(10): 1095-1113. https://doi.org/10.1080/10106049.2017.1333535

Samal, K.K.R., Babu, K.S. & Das, S.K. 2021. Temporal convolutional denoising autoencoder network for air pollution prediction with missing values. Urban Climate 38: 100872. https://doi.org/10.1016/j.uclim.2021.100872

Shavalipour, A., Hakemzadeh, M.H., Sopian, K., Mohamed Haris, S. & Zaidi, S.H. 2013. New formulation for the estimation of monthly average daily solar irradiation for the tropics: A case study of Peninsular Malaysia. International Journal of Photoenergy 2013: 174671. https://doi.org/10.1155/2013/174671

Singh, R.K., Vader, A., Mundy, C.J., Søreide, J.E., Iken, K., Dunton, K.H., Castro de la Guardia, L., Sejr, M.K. & Bélanger, S. 2022. Satellite-derived photosynthetically available radiation at the coastal Arctic seafloor. Remote Sensing 14(20): 5180. https://doi.org/10.3390/rs14205180

Tan, M.L., Juneng, L., Tangang, F.T., Chung, J.X. & Radin Firdaus, R.B. 2021. Changes in temperature extremes and their relationship with ENSO in Malaysia from 1985 to 2018. International Journal of Climatology 41(S1): E2564-E2580. https://doi.org/10.1002/joc.6864

Tan, P.Y. & Ismail, M.R. 2016. Photosynthetically active radiation and comparison of methods for its estimation in equatorial Singapore. Theoretical and Applied Climatology 123(3-4): 873-883. https://doi.org/10.1007/s00704-015-1399-9

Vincent, P., Larochelle, H., Lajoie, I., Bengio, Y. & Manzagol, P.A. 2010. Stacked denoising autoencoders: Learning useful representations in a deep network with a local denoising criterion. Journal of Machine Learning Research 11: 3371-3408.

Wang, D., Liang, S., Zhang, Y., Gao, X., Brown, M.G.L. & Jia, A. 2020. A new set of MODIS land products (MCD18): Downward shortwave radiation and photosynthetically active radiation. Remote Sensing 12(1): 168. https://doi.org/10.3390/RS12010168

Wiener, N. 1949. Extrapolation, Interpolation, and Smoothing of Stationary Time Series. New York: John Wiley & Sons.

Wu, Z. & Huang, N.E. 2009. Ensemble empirical mode decomposition: A noise-assisted data analysis method. Advances in Adaptive Data Analysis 1(1): 1-41. https://doi.org/10.1142/S1793536909000047

Zhang, X., Yang, Y., Jin, M., Lujie, Z. & Zijian, D. 2022. Rayleigh lidar signal denoising method combined with WT, EEMD and LOWESS to improve retrieval accuracy. Remote Sensing 14(23): 6175.  https://doi.org/10.3390/rs14143270

Zhao, P., Bai, Y., Zhang, Z., Wang, L., Guo, J. & Wang, J. 2023. Differences in diffuse photosynthetically active radiation effects on cropland light use efficiency calculated via contemporary remote sensing and crop production models. Ecological Informatics 73: 101948. https://doi.org/10.1016/j.ecoinf.2022.101948