**1. Introduction**

Land surface temperature is an important climate parameter due to its control of the components of the surface energy budget, such as turbulent heat and moisture fluxes, and upward terrestrial radiation [1]. For climate applications, information is needed on large scales, and ideally, the diurnal cycle needs to be resolved. In this study, we develop an approach to derive information on LST which is applicable to the GOES satellites across multiple missions and multiple satellite sensors. We report on results obtained during the period (2004–2009) at hourly time intervals, at about 5-km spatial resolution.

Since surface ground observations are limited, shelter temperature has been widely used as a proxy to surface skin temperature to meet large-scale needs. Issues emerging from such an approach have been addressed previously [2]. While observations from satellites are considered useful for inferring LST, only a few satellite sensors observe all the necessary parameters needed to derive LST with high accuracy. Some lack sufficient number of channels to account simultaneously for atmospheric effects (as needed for implementing the "split window" approach) [3–7]. Others do not observe the Earth surface at high frequency to resolve the diurnal cycle, or at high spatial resolution, to minimize the presence of clouds.

Information on surface emissivity is also not readily available at sufficient spectral resolution [8,9]. Moreover, land surface emissivity is generally less than one, and therefore, part of the atmospheric downward radiation is reflected by the surface and has to be accounted for [10] when converting ground observations of radiative flux measurements to LST (which is not always done). The number of successful attempts to derive LST from satellites has been substantial (especially using the well-established split window approach). A full review of what was done is beyond the scope of this paper, but a comprehensive summary can be found in Li et al. [11], and is briefly presented below.

The early effort to retrieve LST from satellites over agricultural land made by Price [3] was done by adopting the Advanced Very High Resolution Radiometer (AVHRR) Sea Surface Temperature (SST) split window algorithm [12,13]. Becker and Li [14] extended the split window method of McMillin [15] for SST to LST and took into account the spectral variability in land surface emissivity. This so called "generalized split window" LST algorithm has been widely used. Additional efforts include the work of Prata, [16], Sobrino et al. [7], Wan and Dozier [5], Francois and Ottle [17], Coll and Caselles [18], Trigo et al. [19], and Wan [20]. The approach for accounting for emissivity has evolved from assignments based on land use to the use of the Normalized Difference Vegetation Index (NDVI) [21–23]; however, the NDVI concept is not applicable for every surface type. Currently, surface emissivity is derived from the Advanced Space Thermal Emission and Reflection Radiometer (ASTER), the Thermal Infrared (TIR) Multispectral Scanner (TIMS), and the Moderate-resolution Imaging Spectroradiometer (MODIS) [24–29] culminating in the Combined ASTER and MODIS Emissivity over Land (CAMEL) product to be used here [30,31].

Most of the above referenced studies focus on polar orbiting satellites such as the National Oceanic and Atmospheric Administration (NOAA)-AVHRR, the Along-Track Scanning Radiometer (ATSR) and the MODIS instrument aboard Terra and Aqua satellites; the temporal measurement frequency of these satellites is approximately twice per day. The Land Surface Diurnal Temperature Range (DTR) is an important element of the climate system and is not captured by the polar orbiting satellites. Geostationary satellites provide diurnal coverage and observe the surface continuously at a nadir pixel resolution of about 4 km [32] which led to the development of several algorithms for GEO satellites [33–35].

While the principles of retrieval methodologies have not changed drastically over time, the development in auxiliary information, quality of such information, and availability of long term records of satellite observations make it feasible to formulate a homogeneous approach across various satellite sensors that can culminate in climatic records of LST. The primary objective of this study is to present a methodology that can be implemented with different GOES observing systems, using consistent auxiliary information of highest possible quality and utilizing radiative transfer models that account for the vertical profiles of atmospheric states for each retrieval. From mid-2004 to 2017, only one thermal channel is available on the GOES series; the focus of this study is on retrievals using such a single channel in order that a consistent, long term record can be generated from all the GOES satellites (including those that allow the implementation of the split window approach). In Section 2, materials used are described; retrieval algorithm development is presented in Section 3; evaluation of GOES-E based LST estimates is presented in Section 4; a discussion is provided in Section 5; and a summary is presented in Section 6.

### **2. Materials**
