A new single-channel method for estimating land surface temperature based on the image inherent information: The HJ-1B case

Abstract At present, several algorithms have been developed to retrieve land surface temperature (LST) from remotely sensed imageries. Single-channel algorithms to estimate LST from one thermal band have advantage over other methods when the imageries with only one thermal band are used. However, accurate inputs including atmospheric parameters and land surface emissivity, which are always inaccessible to general users, are indispensable to the applications of the current single-channel algorithms. To solve this problem, a new single-channel method (an image based single-channel method called IBSC) based on the information inherent within the image itself, is proposed in this paper. The IBSC method is intended to retrieve LST directly from the sensor detected radiance, even with no accurate information about land surface emissivity and atmospheric parameters. Several HJ-1B imageries covering the Pearl River Delta and Wuhan City, China, were selected for LST estimation to assess the capability of the IBSC method. Results show that, by means of the IBSC method, sea/water surface temperature was retrieved with general bias about 1 K compared with the MODIS products and the in situ measurements. Meanwhile, the traditional single-channel methods provided less accurate LST results when the plausibly estimated atmospheric parameters and surface emissivity were used as inputs. The inherent uncertainties and errors in the MODIS atmospheric products (e.g. MOD 05) are largely attributed to the incorrectness or insufficient improvement of the final LST results. Therefore, when there is no accurate information about atmospheric parameters and surface emissivity, the IBSC method is a good alternative for general users to make full use of the remotely sensed imageries with only one thermal band. Due mainly to the limited access to abundantly valid field data over land, in this paper, the 33 in situ measures over lakes as well as the sea/water surface temperature from MODIS products were only used to assess the retrieved HJ-1B results. The preliminary assessment positively suggests the applicability of the new method. However, to verify the accuracy and robustness of the IBSC method, more experiments and abundant reference data are required, especially for surface temperature over land. This innovative methodology may also be suitable for other sensors, such as Landsat 5/7/8.