Split-window Covariance-variance Ratio Research Articles

Improving the capability of water vapor retrieval from Landsat 8 using ensemble machine learning

Precipitable water vapor (PWV) facilitates the exchange of water and energy between the Earth's surface and the surrounding atmosphere. PWV can be retrieved by employing the water vapor absorption channels in Landsat 8. Unfortunately, the conventional split-window covariance-variance ratio (SWCVR) method is susceptible to various factors, resulting in low retrieval accuracy and availability. Therefore, we improve the traditional SWCVR model and propose a new PWV retrieval method using ensemble machine learning to address these limitations. The new method uses Gradient Boosting Decision Tree (GBDT) to establish the model between brightness temperature, GNSS-derived PWV, and related surface parameters. The results of the test set show that the improved SWCVR model has an RMSE, Bias, and availability of 0.4947 g/cm2, 0.0276 g/cm2, and 29.6%, respectively. By contrast, the GBDT model's corresponding values are 0.2870 g/cm2, −0.0094 g/cm2, and 67.9%, respectively. Compared with SWCVR, the GBDT improves the RMSE and availability by 41.99% and 38.3%, respectively. The GBDT algorithm is significantly better than the SWCVR model in low altitude and complex surface coverage areas. From a temporal perspective, the advantages of the GBDT method are more apparent in the summer. Finally, the SWCVR and GBDT models are externally validated using measured PWV data from the sun photometer and radiosonde, and the RMSE is 0.5148 g/cm2 and 0.3775 g/cm2, respectively. Based on the findings, we know that the accuracy of GBDT has significantly improved against the SWCVR.

Land Surface Temperature Retrieval From Landsat 9 TIRS-2 Data Using Radiance-Based Split-Window Algorithm

The thermal infrared sensor-2 (TIRS-2) carried on Landsat 9 is the newest thermal infrared (TIR) sensor for the Landsat project and provides two adjacent TIR bands, which greatly benefits the land surface temperature (LST) retrieval at high spatial resolution. In this article, a radiance based split window (RBSW) algorithm for retrieving LST from Landsat 9 TIRS-2 data was proposed. In addition, the split-window covariance-variance ratio (SWCVR) algorithm was improved and applied to Landsat 9 TIRS-2 data for estimating atmospheric water vapor (AWV) that is required for accurate LST retrieval. The performance of the proposed method was assessed using the simulation data and satellite observations. Results reveal that the retrieved LST using the RBSW algorithm has a bias of 0.06 K and root-mean-square error (RMSE) of 0.51 K based on validation with the simulation data. The sensitivity analysis exhibited a LST error of <1.75 K using the RBSW algorithm when the uncertainties in input parameters (i.e., AWV, emissivity, and at-sensor radiance) were considered. There is a marginal discrepancy between LST retrievals using the estimated AWV and moderate resolution imaging spectroradiometer AWV, with a difference in bias of −0.14 K and RMSE of 0.22 K, which indicates that the improved SWCVR method can provide an optional means to obtain AWV inputted in LST retrieval. With regard to the validation using the in situ measurements, the retrieved LST from Landsat 9 TIRS-2 data exhibits a bias of 0.44 K and RMSE of 1.98 K, respectively, showing a higher accuracy than the USGS Landsat LST product with bias of 1.21 and RMSE of 2.56 K. In conclusion, the proposed algorithm combining RBSW algorithm and improved SWCVR algorithm for LST retrieval from Landsat 9 has a good accuracy without dependence on external atmospheric data, and it is expected to be a reliable method for LST generation from Landsat 9 TIRS-2 data.

A Framework for Estimating Clear-Sky Atmospheric Total Precipitable Water (TPW) from VIIRS/S-NPP

Atmospheric water vapor content or total precipitable water (TPW) is a highly variable atmospheric constituent, yet it remains one of the meteorological parameters that is most difficult to characterize accurately. We develop a framework for estimating atmospheric TPW from Visible Infrared Imaging Radiometer Suite (VIIRS) data in this study. First, TPW is retrieved from VIIRS top-of-atmosphere (TOA) radiance of channels 15 and 16 using the refined split-window covariance-variance ratio (SWCVR) method. Then, the VIIRS TPW is blended with the microwave integrated retrieval system (MIRS) derived TPW via Bayesian model averaging (BMA) to improve the accuracy of VIIRS TPW. Three years (2014–2017) of ground measurements collected from SuomiNet sites over North America are used to validate the VIIRS TPW and blended TPW. The mean bias error (MBE) and root mean square error (RMSE) of the VIIRS TPW are 0.21 g/cm2 and 0.73 g/cm2, respectively, and the accuracy of the VIIRS TPW in daytime is much better than at night time. The MBE and RMSE of BMA integrated TPW are 0.06 g/cm2 and 0.35 g/cm2, and the accuracy difference between daytime and nighttime is also removed. The global radiosonde measurements are also collected to validate the BMA integrated VIIRS TPW. The MBE and RMSE of the BMA integrated TPW are 0.09 g/cm2 and 0.44 g/cm2 compared to the radiosonde measurements. This accuracy is also superior to the VIIRS TPW. Therefore, it is concluded that the developed framework can be used to derive accurate clear-sky TPW for VIIRS. This is the first time that we can obtain high accuracy TPW from VIIRS. This study will certainly benefit the study of atmospheric processes and climate change.

Improved Assessment of Atmospheric Water Vapor Content in the Himalayan Regions Around the Kullu Valley in India Using Landsat‐8 Data

AbstractIn this study, we present an approach for improved estimation of atmospheric water vapor content (WVC) from Landsat‐8 data. The initial estimates of WVC derived from the split window covariance variance ratio (SWCVR) method are sequentially improved by two exponential models. The first model is designed to apply orographic corrections to the SWCVR estimate using a 1‐arcsec Shuttle Radar Topography Mission digital elevation model based on a scale and bias parameter for the change in elevation. This model is based on the local change in elevation derived using the directional Kirsch compass kernel gradient. A second model for additional improvement in orographically corrected WVC estimate is defined based on the change in the normalized differenced vegetation index (NDVI) with three parameters, a scale and bias parameter for the change in NDVI and one bias parameter for the WVC estimate. The scale and bias parameters in orographic correction and the NDVI‐based correction are derived using regression between a selected training subset of WVC estimates (SWCVR estimate for orographic correction and orographically corrected WVC estimate for the NDVI‐based correction) and the corresponding Sentinel‐3 Ocean Land Color Instrument (OLCI) integrated water vapor (IWV) product. The study is conducted over the Himalayan regions around Kullu, Himachal Pradesh, India, for two dates, 20 September 2017 and 26 January 2018, corresponding to the autumn and peak winter season in the region. A reduction in error of 0.32and 0.04 gm/cm2 was observed in proposed WVC estimate and the Sentinel‐3 integrated water vapor product, for the two data sets respectively.

Towards an operational method for land surface temperature retrieval from Landsat 8 data

ABSTRACTLand surface temperature (LST) is a key parameter in the physics of land surfaces through the processes of energy and water exchange with the atmosphere. For Landsat data with only one thermal infrared channel (Landsat 4 to Landsat 7), LST cannot actually be retrieved, and external data sources, such as meteorological observations or Moderate Resolution Imaging Spectroradiometer (MODIS) data, are needed to obtain the water vapour content parameter (an important input parameter for the LST retrieval algorithm); this results in limitations on deriving LST. However, the band designations of the Landsat 8 sensors enable the derivation of LST from the Landsat 8 data. This article demonstrates an LST retrieval methodology that makes use of only Landsat 8 image data. In this methodology, the split-window covariance-variance ratio (SWCVR) technique is introduced to derive water vapour content from Landsat 8. A comparison between the retrieved LST and the in situ LST measurements shows good accuracy, with a root mean squared error (RMSE) of 0.83 K. The fact that the proposed LST estimation method utilizing solely Landsat 8 image data does not rely on any external data is a significant advantage for the development of an operational Landsat 8 LST product generating system.

NDVI-based split-window algorithm for precipitable water vapour retrieval from Landsat-8 TIRS data over land area

An algorithm for the retrieval of precipitable water vapour (PWV) from Landsat-8 thermal infrared sensor (TIRS) data over land area has been developed in this paper. This method is based on the split-window covariance-variance ratio (SWCVR) theory and introduces normalized difference vegetation index (NDVI) to improve PWV retrieval from the relatively high-resolution thermal infrared data. Validation of the method is performed with meteorological data and Moderate Resolution Imaging Spectroradiometer (MODIS) total column PWV product (MOD05), and comparisons between NDVI-based SWCVR method and previous SWCVR method are employed. The root mean square error (RMSE) between PWV retrieved and that provided by meteorological data is 0.39 and 0.57 g cm−2 respectively for the proposed and previous method. The RMSE is 0.55 and 0.69 g cm−2 respectively for the proposed and previous method as validated with MOD05. It is concluded that the proposed method for retrieval of PWV from Landsat-8 TIRS data attained a better accuracy than the previous SWCVR method. PWV obtained from the proposed method is of great value as an input for land surface temperature retrieval and atmospheric correction for the Landsat-8 data.

A comparison between atmospheric water vapour content retrieval methods using MSG2-SEVIRI thermal-IR data

Atmospheric water vapour content (WVC) is a vital parameter in the study of climate change. Various methods have been developed to derive atmospheric WVC from remotely sensed data. In this study, we compared three methods for retrieving atmospheric WVC from thermal infrared data in the Meteosat Second Generation-SEVIRI channels 9 (10.8 μm) and 10 (12.0 m). The three methods are (1) the split-window covariance-variance ratio method using a spatial moving window (method 1); (2) the split-window covariance-variance ratio method using a temporal moving window (method 2); and (3) the varying surface temperature method using split-window channel data (method 3). The derived WVC using these three methods was compared to two WVC data sets from European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis data and the MODIS WVC product. Compared with these two data sets, the derived WVC using method 1 performed proved optimal. The valid pixels using methods 1 and 2 are greater than those using method 3. Furthermore, method 2 can be used to retrieve WVC over pixels where method 1 is invalid.

A new approach for retrieving precipitable water from ATSR2 split-window channel data over land area

This paper presents a new algorithm to determine quantitatively column water vapour content ( W) directly from ATSR2 (Along-Track Scanner Radiometer) Split-Window radiance measurements. First, the Split-Window Covariance-Variance Ratio (SWCVR) method is reviewed. The assumptions made to derive this method are highlighted and its applicability is discussed. Then, an operational use of this method is developed and applied to several ATSR2 datasets. The water vapour contents retrieved using ATSR2 data from SGP'97 (USA), Barrax (Spain) and Cabauw (The Netherlands) are in good agreement with those measured by the quasi-simultaneous radiosonde. The mean and the standard deviation of their difference are 0.04 g cm−2 and 0.22 g cm−2, respectively. It is shown that water vapour content derived from ATSR2 data using the proposed algorithm is accurate enough in most cases for surface temperature determination with a split-window technique using ATSR2 data and for atmospheric corrections in visible and near-infrared channels of ATSR2.

Further Insights into the Use of the Split-Window Covariance Technique for Precipitable Water Retrieval

This article is the response to Barton and Prata’s article published in the same issue of this journal. Our purpose is to describe the conditions of use of the split-window covariance-variance ratio technique. The methodology is first presented and the ways to apply it correctly are investigated.

Atmospheric water vapor content over land surfaces derived from the AVHRR data: application to the Iberian Peninsula

A study has been carried out using simulated NOAA/advanced very high resolution radiometer (AVHRR) data at 11 and 12 /spl mu/m (with LOWTRAN-7, MODTRAN 2.0, and the TIGR database), AVHRR images of the Iberian Peninsula and the Palma de Mallorca Island, radiosonde observations at seven meteorological stations, and the AVISO database provided by Meteo France to describe, compare, and analyze two different approaches for estimating the total atmospheric water vapor content (W) over land surfaces from AVHRR data. These two techniques are: 1) the split-window covariance-variance ratio (SWCVR), based on a quadratic relationship between W and the ratio of the spatial covariance and variance of brightness temperatures measured in channels 4 (T/sub 4/) and 5 (T/sub 5/) of AVHRR in subsets of N neighboring pixels and 2) the linear split-window relationship (LSWR), based on a linear regression between W and the difference of brightness temperatures measured in the same channels (/spl Delta/T=T/sub 4/-T/sub 5/). The results demonstrate the advantage of the SWCVR technique for regions with a certain level of thermal heterogeneity (standard deviation of T/sub 4/ in the subset >0.5 K), which is capable of estimating W from NOAA-14 afternoon and night passes over the Iberian Peninsula with a standard deviation of 0.5 (g cm/sup -2/), whereas the LSWR technique predicts the atmospheric water vapor with a standard deviation from 1.3-1.5 (g cm/sup -2/). Finally a water vapor image of the entire Iberian Peninsula constructed by applying the SWCVR to NOAA-14 data is presented.