An intercomparison of SMAP, SMOS, AMSR2, FY3B and ESA CCI soil moisture products at different spatial scales over two dense network regions

Transcription

1 An intercomparison of SMAP, SMOS, AMSR2, FY3B and ESA CCI soil moisture products at different spatial scales over two dense network regions Jiangyuan Zeng 1, Kun-Shan Chen 1, Chenyang Cui 2, Haiyun Bi 3 1 Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences 2 School of Earth Sciences and Engineering, Hohai University 3 Institute of Geology, China Earthquake Administration EGU General Assembly 218 (EGU218) April 1, 218, Vienna, Austria

2 Outline I. Background and Motivation II. Data and study area III. Evaluation method IV. Results and discussions V. Conclusions

3 Outline I. Background and Motivation II. Data and study area III. Evaluation method IV. Results and discussions V. Conclusions

4 I. Background and Motivation Soil moisture mapping at large scales Active Passive Scatterometer SAR Altimeter Radiometer Transmit and receive signals Receive signals

5 I. Background and Motivation Microwave remote sensing of soil moisture SMMR SSM/I TMI AMSR-E AMSR2 Passive WindSat FY3B In orbit SMOS ESA CCI SM Aquarius SMAP Active SCAT ASCAT

6 I. Background and Motivation Motivation No work has been carried out for an intercomparison of the ongoing SMAP, SMOS, FY3B, AMSR2, and ESA CCI SM products Most work focuses on assessment of SM products at a relatively coarse spatial resolution of approximately.25 (i.e., at one spatial scale) Most work ignores the investigation of the possible error sources of the products Objective Examining the possible errors and limitations of eight new SM products Evaluating the quality of SM products at a coarse resolution of.25 and at a medium resolution of.1 (i.e., the new SMAP enhanced passive and JAXA) Investigating the potential of other new SM products (i.e., SMAP and FY3B) to be integrated into the ESA CCI product

7 Outline I. Background and Motivation II. Data and study area III. Evaluation method IV. Results and discussions V. Conclusions

8 II. Data and study area Description of the eight SM products used in the study Spatial scale Products Level Frequency used (GHz) Algorithms Grid resolution SMAP passive L SCA-V 36 km SMOS (CATDS) L L-MEB 25 km ~.25 FY3B L2 1.65, 36.5 Qp model 25 km JAXA AMSR2 L3 1.65, 36.5 LUT.25 LPRM AMSR2 L3 1.65, 36.5 LPRM.25 ~.1 ESA CCI (Combined) SMAP enhanced passive L3 / /.25 L SCA-V 9 km JAXA AMSR2 L3 1.65, 36.5 LUT.1

9 II. Data and study area Select reasons: Dense in-situ sites High-quality data Different land cover Networks Sites Depths used (cm) Time step (min) Land cover LWW Grasslands REMEDHUS Croplands Evaluated period ~ ~

10 Outline I. Background and Motivation II. Data and study area III. Evaluation method IV. Results and discussions V. Conclusions and outlook

11 III. Evaluation method Data pre-processing SMAP passive, SMOS, FY3B were resampled to.25 (LPRM, JAXA, ESA CCI) SMAP enhanced passive was resampled to.1 (JAXA) Data filter for SMOS: DQX>.1, and RFI_Prob>.2 Spatial average of the in-situ observations LWW (13 stations for.25 grid, and 4 stations for.1 grid) REMEDHUS (11 stations for.25 grid, and 3 stations for.1 grid) Evaluating satellite products at nighttime or morning time SMAP and SMOS: 6: A.M. FYB3 and AMSR2: 1:4 or 1:3 A.M. ESA CCI: Daily Error metrics ubrmse, RMSE, Bias, correlation coefficient (R) Evaluation period 1 April 215 to 31 December 216

12 Outline I. Background and Motivation II. Data and study area III. Evaluation method IV. Results and discussions V. Conclusions

13 m ) Soil moisture (m 3 m -3-3 ) Soil moisture (m 3 m -3-3 ) Soil Soil moisture (m (m 3 m m ) -3-3 ) Soil Soil moisture (m (m 3 m m ) -3-3 ) Precipitation(mm -1 (mm day ) Precipitation(mm day -1 ) day Precipitation -1-1 ) (mm day -1 ) Precipitation (mm (mm day day -1-1 ) -1-1 ) Precipitation(mm day -1-1 ) Precipitation (mm (mm day day -1-1 ) -1-1 ) Soil moisture (m 3 m -3-3 ) ) Soil Soil moisture moisture (m (m 3 m 3 m -3 ) -3 ) Soil moisture (m Soil moisture (m 3 m -3-3 ) ) 3 m Soil -3-3 Soil ) ) moisture moisture Soil Soil moisture moisture (m (m 3 m 3 m -3 ) -3 (m (m ) 3 m 3 m -3 ) -3 ) Precipitation(mm day -1-1 ) ) -1 Precipitation Precipitation (mm (mm day day -1 ) -1 ) Precipitation (mm day -1 ) III. Results and discussions LWW pr/15 6/Jan/16 12/Oct/16 6 pr/15 6/Jan/16 12/Oct/ /Apr/15.6 6/Jan/ /Oct/16 1/Apr/15 6/Jan/16 12/Oct/ pr/15 6/Jan/16 12/Oct/ /Apr/15 6/Jan/16 12/Oct/ /Apr/ /Jan/16 12/Oct/16 1/Apr/15 6/Jan/16 12/Oct/ /Apr/ /Jan/16 12/Oct/ /Apr/15 6/Jan/16 12/Oct/ /Apr/15 6/Jan/16 12/Oct/ /Apr/15 1/Apr/15 6/Jan/16 6/Jan/16 12/Oct/16 12/Oct/16.6 The.6best performance: SMAP.6(in.6LWW), FY3B (in REMEDHUS) LPRM (large wet bias), JAXA (large dry bias) Correlation: SMAP, SMOS and ESA CCI > LPRM 2 and 2 JAXA 3 m -3 ) 3 m -3 ) Precipitation In-situ_SM SMOS SMOS Precipitation JAXA_25 In-situ_SM ESA ESA CCI CCISMAP_P LPRM LPRMFY3B FY3B 3 m -3 ) 3 m -3 ) REMEDHUS day -1 ) day -1 ) day -1 ) 2 Precipitation In-situ_SM SMOS Precipitation JAXA_25 In-situ_SM ESA CCI SMAP_P LPRM FY3B 3 m -3 ).6 day -1 ) day -1 ) 2 2

14 Soil moisture (m 3 m -3 ) Soil moisture (m 3 m -3 ) Soil moisture (m 3 m -3 ) Soil moisture (m 3 m -3 ) Precipitation (mm day -1 ) Precipitation (mm day -1 ) Precipitation (mm day -1 ) Soil moisture (m 3 m -3 ) Soil moisture (m 3 m -3 ) Soil moisture (m 3 m -3 ) Precipitation (mm day -1 ) -1 III. Results and discussions /Jan/16 LWW /Apr/15 6/Jan/16 12/Oct/16 l moisture (m 3 m -3 ) SMAP_P_E (small dry bias), JAXA (large dry bias) Correlation: SMAP_P_E>JAXA Performance:.4 SMAP_P_E>JAXA /Oct/ REMEDHUS /Apr/15 6/Jan/16 12/Oct/16 Precipitation In-situ_SM SMAP_P_E JAXA_1 1/Apr/15 6/Jan/16 12/ Precipitation In-situ_SM SMAP_P_E JAXA_1 cipitation (mm day -1 ) cipitation (mm day -1 ) tion (mm day -1 )

15 III. Results and discussions Error metrics of eight soil moisture products for LWW network Resolution Products ubrmse (m 3 m 3 ) RMSE (m 3 m 3 ) Bias (m 3 m 3 ) R N.25.1 SMAP_P SMOS FY3B JAXA LPRM ESA CCI SMAP_E_P JAXA The best performance: SMAP_P LPRM (large wet bias), JAXA (large dry bias) R value: SMAP, SMOS, FY3B, and ESA CCI > LPRM and JAXA JAXA (.25) outperforms JAXA (.1 ) SMAP_P_E outperforms JAXA at.1

16 III. Results and discussions Error metrics of eight soil moisture products for REMEDHUS network Resolution Products ubrmse (m 3 m 3 ) RMSE (m 3 m 3 ) Bias (m 3 m 3 ) R N.25.1 SMAP _P SMOS FY3B JAXA LPRM ESA CCI SMAP _E_P JAXA The best performance: FY3B LPRM (large wet bias), JAXA (large dry bias) R value: SMAP, SMOS, FY3B, and ESA CCI > LPRM and JAXA JAXA (.25) outperforms JAXA (.1 ) SMAP_P_E outperforms JAXA

17 III. Results and discussions LWW REMEDHUS.25 Best performance L: SMAP_P R: FY3B Best performance SMAP_E_P.1

18 Satellite ST products ( ) Satellite ST products ( ) Satellite ST products ( ) Satellite ST products ( ) Satellite ST products ( ) 5 III. Results and discussions Assessment of Satellite Surface Temperature Data LWW REMEDHUS In-situ 5 5 ST 15 15( ) In-situ In-situ ST ST ( ) ( ) In-situ In-situ ST ST ( ) ( ) Model simulations 35(GEOS-5 and ECMWF) outperform LPRM ST retrievals All ST show dry bias, which may contribute to the dry bias of SMAP and SMOS SM Wet bias of LPRM 25SM was not caused by its ST retrievals cts( ) ucts( ) ucts( ) 25-5 SMAP_P SMAP_P SMAP_P_E SMOS SMOS LPRM LPRM 1:1 1:1 line line In-situ ST ( ) SMAP_P SMAP_P_E SMOS LPRM 1:1 line

19 III. Results and discussions Error metrics of satellite ST data for LWW network Products ubrmse (K) RMSE (K) Bias (K) R N SMAP_P SMAP _E_P SMOS LPRM Error metrics of satellite ST data for REMEDHUS network Products ubrmse (K) RMSE (K) Bias (K) R N SMAP_P SMAP _E_P SMOS LPRM Model simulations (GEOS-5 and ECMWF) outperform LPRM ST retrievals All ST show dry bias, which may contribute to the dry bias of SMAP and SMOS SM Wet bias of LPRM SM was not caused by its ST retrievals

20 egetation optical depth egetation optical depth ation optical depth egetation optical depth egetation optical depth VWC (kg m -2 ) VWC (kg m -2 ) VWC (kg m -2 ) VWC (kg m -2 ) VWC (kg m -2 ) VWC (kg m -2 ) VWC (kg m -2 ) Vegetation optical depth VWC (kg m -2 ) VWC (kg m -2 ) VWC (kg m -2 ) Vegetation optical depth Vegetation optical depth Vegetation optical depth VWC (kg m -2 ) VWC (kg m -2 ) III. Results and discussions Temporal Behavior of Satellite Vegetation Optical Depth (tau).8 LWW 1.2 REMEDHUS /Jan/16 21/Nov/16.6 1/Apr/15 1/Apr/15 26/Jan/16 26/Jan/16 21/Nov/16 21/Nov/ /Apr/15 1/Apr/15 26/Jan/16 26/Jan/16 21/Nov/16 21/Nov/16 VOD 线 VOD (SMOS) 日期线 (SMOS) 日期线 VOD VOD VOD (LPRM) VOD (SMOS) (LPRM) 日期线 (SMOS) VWC 日期线 VOD VWC VOD (SMAP_P) VOD (LPRM) VOD (SMAP_P) (SMOS) (LPRM) (SMOS) VWC VOD VOD VWC VOD (SMAP_P) VOD (LPRM) (SMAP_P) (LPRM) VOD VWC VOD VWC (SMAP_P) (SMAP_P) VOD VOD (SMAP_P) (SMAP_P) 1.8 Optical 1 1 derived 1VWC 1 R (SMOS_VOD) R 1.5(LPRM_VOD) SMAP_VWC.8.8 (LWW) SMAP_VWC.6.6 (REMEDHUS) SMAP tau shows.4much smoother variation than SMOS.6 and LPRM.6 SMOS.2.2 tau seems.2 noisy,.2 leading to a very large dynamic.3.3 range of.3tau.3(>.4).3.3 LPRM tau shows.2a better response to optical derived.3 VWC compared with.3smos tau /Apr/15 26/Jan/16 21/N MOS) 日期线 VOD VOD (LPRM) (SMOS) 日期线 VWC VOD VOD (SMAP_P) (LPRM) (SMOS) VWC VOD VOD (SMAP_P) (LPRM) VOD VWC (SMAP_P) VOD (S 1.8 (b

21 Outline I. Background and Motivation II. Data and study area III. Evaluation method IV. Results and discussions V. Conclusions

22 IV. Conclusions and outlook The best performance: SMAP_P (LWW network, ubrmse =.27 m 3 m -3 ), FY3B (REMEDHUS network, ubrmse =.25 m 3 m -3 ) SMOS slightly underestimates SM, but it correlates well with in-situ data (average R =.77). The JAXA performs much better at.25 than at.1, but both of them underestimate SM at most time (bias >.5 m 3 m -3 ). The SMAP_E_P well captures the temporal variation of measurements (R>.8), and is generally superior to the JAXA product. The LPRM shows much larger amplitude and temporal variation than the ground SM (bias >.9 m 3 m -3 ). The underestimation of ST contributes to the general dry bias found in the SMAP and SMOS SM. The ESA CCI shows satisfactory performance (ubrmse<.45 m 3 m -3 ), and it could integrate SMAP and FY3B to form a more reliable and useful product in the future.

23 Literature related to this work Cui, C., Xu, J., Zeng, J.Y.*, Chen, K. S., Bai, X. J., Lu, H., Chen, Q., & Zhao, T.J. (218). Soil Moisture Mapping from Satellites: An Intercomparison of SMAP, SMOS, FY3B, AMSR2, and ESA CCI over Two Dense Network Regions at Different Spatial Scales. Remote Sensing, 1(1), 33. ( Zeng, J.Y., Li, Z., Chen, Q., Bi, H.Y., Qiu, J.X., & Zou, P.F. (215). Evaluation of remotely sensed and reanalysis soil moisture products over the Tibetan Plateau using in-situ observations. Remote Sensing of Environment, 163, ( Zeng, J.Y., Chen, K. S., Bi, H.Y., & Chen, Q. (216). A preliminary evaluation of the SMAP radiometer soil moisture product over United States and Europe using ground-based measurements. IEEE Transactions on Geoscience and Remote Sensing, 54(8), ( Zeng, J.Y., Li, Z., Chen, Q., & Bi, H.Y. (215). Method for soil moisture and surface temperature estimation in the Tibetan Plateau using spaceborne radiometer observations. IEEE Geoscience and Remote Sensing Letters, 12(1), (

24