Evaluations of Machine Learning-Based CYGNSS Soil Moisture Estimates against SMAP Observations

Abstract

This paper presents a machine learning (ML) framework to derive a quasi-global soil moisture (SM) product by direct use of the Cyclone Global Navigation Satellite System (CYGNSS)’s high spatio-temporal resolution observations over the tropics (within ±38 latitudes) at L-band. The learning model is trained by using in-situ SM data from the International Soil Moisture Network (ISMN) sites and various space-borne ancillary data. The approach produces daily SM retrievals that are gridded to 3 km and 9 km within the CYGNSS spatial coverage. The performance of the model is independently evaluated at various temporal scales (daily, 3-day, weekly, and monthly) against Soil Moisture Active Passive (SMAP) mission’s enhanced SM products at a resolution of 9 km × 9 km. The mean unbiased root-mean-square difference (ubRMSD) between concurrent (same calendar day) CYGNSS and SMAP SM retrievals for about three years (from 2017 to 2019) is 0.044 cm3 cm−3 with a correlation coefficient of 0.66 over SMAP recommended grids. The performance gradually improves with temporal averaging and degrades over regions regularly flagged by SMAP such as dense forest, high topography, and coastlines. Furthermore, CYGNSS and SMAP retrievals are evaluated against 170 ISMN in-situ observations that result in mean unbiased root-mean-square errors (ubRMSE) of 0.055 cm3 cm−3 and 0.054 cm3 cm−3, respectively, and a higher correlation coefficient with CYGNSS retrievals. It is important to note that the proposed approach is trained over limited in-situ observations and is independent of SMAP observations in its training. The retrieval performance indicates current applicability and future growth potential of GNSS-R-based, directly measured spaceborne SM products that can provide improved spatio-temporal resolution than currently available datasets.

Temporal-Spatial Soil Moisture Estimation from CYGNSS Using Machine Learning Regression With a Preclassification Approach

“…The RF method performed best and was then adopted to show the performance of CYGNSS-based SM estimates involving SMAP data. Mean unbiased ubRMSE of 0.055 and 0.054 cm 3 /cm 3 were obtained with CYGNSS estimates and SMAP against in situ observations, respectively, with a higher R with the CYGNSS retrievals [36]. The number of ancillary data that are from other sources was not reduced [35], [36].…”

“…Mean unbiased ubRMSE of 0.055 and 0.054 cm 3 /cm 3 were obtained with CYGNSS estimates and SMAP against in situ observations, respectively, with a higher R with the CYGNSS retrievals [36]. The number of ancillary data that are from other sources was not reduced [35], [36]. Yang et al [37] also used backpropagation (BP)-ANN to compare and evaluate the SM estimation performance of the two spaceborne GNSS-R satellite missions (TDS and CYGNSS), which has quite a few (six) ancillary variables.…”

Evaluation of the Land GNSS-Reflected DDM Coherence on Soil Moisture Estimation from CYGNSS Data

“…The problem is that the differences in estimator performance can lead to different results, and the I/F signal dataset used is too small, lacking sufficient persuasive power. Based on the different assumptions, several SM inversion methods have also been developed, such as spatial averaging, combine linear regression method, machine-learning method, and the global inversion accuracy of SM can reach about 0.05 cm 3 /cm 3 [21][22][23][24][25][26][27][28][29][30].…”

Assessment of Interpolation Errors of CYGNSS Soil Moisture Estimations

“…NASA's Cyclone Global Navigation Satellite System (CYGNSS, launched in December 2016) is primarily an ocean surface wind mission to improve extreme weather prediction, but it operates continuously over both land and ocean from 38 o North to 38 o South latitudes, providing many land observations with clear sensitivity to changing land surface conditions. Many studies have recently taken advantage of the large amount of CYGNSS land observations to develop models and algorithms to retrieve SM from CYGNSS observations at various spatial and temporal scales [9]- [14].…”