Thông tin chung

  English

  Đề tài NC khoa học
  Bài báo, báo cáo khoa học
  Hướng dẫn Sau đại học
  Sách và giáo trình
  Các học phần và môn giảng dạy
  Giải thưởng khoa học, Phát minh, sáng chế
  Khen thưởng
  Thông tin khác

  Tài liệu tham khảo

  Hiệu chỉnh

 
Số người truy cập: 106,047,859

 Error estimation of satellite signal-based river discharge using double instrumental variables method
Tác giả hoặc Nhóm tác giả: H. T. Pham, S. Kim, L. Marshall, A. Sharma
Nơi đăng: The 23rd International Congress on Modelling and Simulation (MODSIM2019); Số: 2019;Từ->đến trang: 881;Năm: 2019
Lĩnh vực: Khoa học công nghệ; Loại: Báo cáo; Thể loại: Quốc tế
TÓM TẮT
River discharge is an important hydrologic variable for flood forecasting and water resources management. However many regions suffer from a lack of ground-based river discharge observations because of delays in data release, sparse distributions of river gauges or expensive maintenance and operation costs. Satellite data which have near-real time access, global coverage, and low cost can provide useful information to observe dynamics of surface water in ungauged regions. Recently, tremendous efforts have been devoted to detect, monitor, and estimate river discharge from satellite signals derived from the Global Flood Detection System (GFDS). The satellite signal is a ratio of satellite-observed passive microwave brightness temperature of a wet area to that of a dry area, and is closely related to surface water dynamics because of contrasted thermal emissivity over the wet and dry areas. Previous studies have widely used such satellite signals for detecting floods, estimating river discharge and calibrating hydrologic models. Although satellite signals can be potentially used for improving these types of hydrologic applications, fewer efforts have been made in quantifying uncertainties in the satellite signal data. The errors in satellite signal data are caused by errors in sensors, land surface conditions (water extents or river morphology) and atmospheric conditions (clouds or temperature). These errors can be propagated into river discharge estimations and hence provide less accurate calibrated parameters, flood detection, or forecasts. Therefore error estimation of satellite signals is necessary to provide reliability of the hydrologic applications. Due to a lack of suitable ground observations, large spatial scale validation of remotely sensed data is often difficult. For this reason, a cross-comparison approach such as triple collocation (TC) has been used to estimate errors of large-scale data. The TC method compares three independent products to estimate the error variances in all products and the individual product-truth correlations. However, the TC method requires three independent products that are often not available in practice, especially for satellite signal data. This study adopts a double instrumental variables (IVd) method which uses only two independent products to estimate error variances and product-truth correlations. Here we used two flood magnitude data (FM) derived from two satellite signals GPM and AMSR-2 as two independent products. We used daily in-situ river discharge (Q) at 206 Hydrologic Reference Stations (HRSs) over Australia from 01/2015 to 04/2019 as a reference data. The HRS river discharge was standardised to be consistent with the two FM products. In addition, 1-day lagged time series of the FM (FM-lag) was used as a double instrumental variable for the IVd analysis. The FM data-truth correlations estimated by the IVd method were compared to the linear correlations between the FM and the standardised river discharge. We then present the strength of the pixel-wise data-truth correlations in different conditions across Australia to establish the conditions under which satellite signals may be useful for hydrological applications
ABSTRACT
River discharge is an important hydrologic variable for flood forecasting and water resources management. However many regions suffer from a lack of ground-based river discharge observations because of delays in data release, sparse distributions of river gauges or expensive maintenance and operation costs. Satellite data which have near-real time access, global coverage, and low cost can provide useful information to observe dynamics of surface water in ungauged regions. Recently, tremendous efforts have been devoted to detect, monitor, and estimate river discharge from satellite signals derived from the Global Flood Detection System (GFDS). The satellite signal is a ratio of satellite-observed passive microwave brightness temperature of a wet area to that of a dry area, and is closely related to surface water dynamics because of contrasted thermal emissivity over the wet and dry areas. Previous studies have widely used such satellite signals for detecting floods, estimating river discharge and calibrating hydrologic models. Although satellite signals can be potentially used for improving these types of hydrologic applications, fewer efforts have been made in quantifying uncertainties in the satellite signal data. The errors in satellite signal data are caused by errors in sensors, land surface conditions (water extents or river morphology) and atmospheric conditions (clouds or temperature). These errors can be propagated into river discharge estimations and hence provide less accurate calibrated parameters, flood detection, or forecasts. Therefore error estimation of satellite signals is necessary to provide reliability of the hydrologic applications. Due to a lack of suitable ground observations, large spatial scale validation of remotely sensed data is often difficult. For this reason, a cross-comparison approach such as triple collocation (TC) has been used to estimate errors of large-scale data. The TC method compares three independent products to estimate the error variances in all products and the individual product-truth correlations. However, the TC method requires three independent products that are often not available in practice, especially for satellite signal data. This study adopts a double instrumental variables (IVd) method which uses only two independent products to estimate error variances and product-truth correlations. Here we used two flood magnitude data (FM) derived from two satellite signals GPM and AMSR-2 as two independent products. We used daily in-situ river discharge (Q) at 206 Hydrologic Reference Stations (HRSs) over Australia from 01/2015 to 04/2019 as a reference data. The HRS river discharge was standardised to be consistent with the two FM products. In addition, 1-day lagged time series of the FM (FM-lag) was used as a double instrumental variable for the IVd analysis. The FM data-truth correlations estimated by the IVd method were compared to the linear correlations between the FM and the standardised river discharge. We then present the strength of the pixel-wise data-truth correlations in different conditions across Australia to establish the conditions under which satellite signals may be useful for hydrological applications
© Đại học Đà Nẵng
 
 
Địa chỉ: 41 Lê Duẩn Thành phố Đà Nẵng
Điện thoại: (84) 0236 3822 041 ; Email: dhdn@ac.udn.vn