A comparison of SMOS and AMSR2 soil moisture using representative sites of the OzNet monitoring network

Mei Sun Yee, Jeffrey P. Walker, Christoph Rüdiger, Robert M Parinussa, Toshio Koike, Yann Henry Kerr

Research output: Contribution to journalArticleResearchpeer-review

19 Citations (Scopus)

Abstract

This paper evaluates the performance of different soil moisture products from AMSR2 and SMOS against the most representative stations within the Yanco study area in the Murrumbidgee catchment, in southeast Australia. AMSR2 Level 3 (L3) soil moisture products retrieved from two versions of brightness temperatures using the Japanese Aerospace eXploration Agency (JAXA) and the Land Parameter Retrieval Model (LPRM) algorithm were included. For the LPRM algorithm, two different parameterization methods were applied. Furthermore, two versions of SMOS L3 soil moisture product were assessed. The results are contrasted against the use of “random” stations. Accounting for all versions, frequencies and overpasses, the latest versions of the JAXA (JX2) and LPRM (LP3) products were found to surpass the earlier versions (JX1, LP1 and LP2). Soil moisture retrieval based on the latter version of brightness temperature and parameterization scheme improved when C-band observations were used but not X-band. However, X-band retrievals (r: 0.71, MAE: 0.07, RMSD: 0.08 m3/m3) were found to perform better than C-band (r: 0.68–0.70, MAE: 0.07–0.09 m3/m3, RMSD: 0.09–0.10 m3/m3). Moreover, an intercomparison between different acquisition times (morning and evening) of AMSR2 X-band products found a better performance from evening overpasses (1:30 pm; r: 0.69–0.77) as opposed to morning overpasses (1:30 am; r: 0.47–0.66). In the case of SMOS, morning (6:00 am; r: 0.77) retrievals were found to be superior over evening (6:00 pm; r: 0.69) retrievals. Overall, both versions of JAXA products, the second and third versions of LPRM X-band products, and two versions of SMOS products were found to meet the mean average error (MAE) goal accuracy of the AMSR2 mission (MAE  < 0.08 m3/m3) but none of the products achieved the SMOS goal of RMSD  < 0.04 m3/m3. Furthermore, performance of the products differed depending on the statistic used to evaluate them. Consequently, considering the results in this study, JX2 products are recommended if both absolute and temporal accuracy of the soil moisture product is of importance, whereas LP3X products from evening observations and SMOS version 3.00 (SMOS2) products from morning overpasses are recommended if temporal accuracy is of greater importance.

Original languageEnglish
Pages (from-to)297-312
Number of pages16
JournalRemote Sensing of Environment
Volume195
DOIs
Publication statusPublished - 15 Jun 2017

Keywords

  • AMSR2
  • JAXA
  • LPRM
  • Representativeness
  • SMOS
  • Soil moisture

Cite this