Overlooked cooling effects of albedo in terrestrial ecosystems

Radiative forcing (RF) resulting from changes in surface albedo is increasingly recognized as a significant driver of global climate change but has not been adequately estimated, including by Intergovernmental Panel on Climate Change (IPCC) assessment reports, compared with other warming agents. Here, we first present the physical foundation for modeling albedo-induced RF and the consequent global warming impact (GWI ∆ α ). We then highlight the shortcomings of available current databases and methodologies for calculating GWI ∆ α at multiple temporal scales. There is a clear lack of comprehensive in situ measurements of albedo due to sparse geographic coverage of ground-based stations, whereas estimates from satellites suffer from biases due to the limited frequency of image collection, and estimates from earth system models (ESMs) suffer from very coarse spatial resolution land cover maps and associated albedo values in pre-determined lookup tables. Field measurements of albedo show large differences by ecosystem type and large diurnal and seasonal changes. As indicated from our findings in southwest Michigan, GWI ∆ α is substantial, exceeding the RF ∆ α values of IPCC reports. Inclusion of GWI ∆ α to landowners and carbon credit markets for specific management practices are needed in future policies. We further identify four pressing research priorities: developing a comprehensive albedo database, pinpointing accurate reference sites within managed landscapes, refining algorithms for remote sensing of albedo by integrating geostationary and other orbital satellites, and integrating the GWI ∆ α component into future ESMs.


Introduction
Since its inclusion in the 3rd Assessment Report (AR3) of the Intergovernmental Panel on Climate Change (IPCC) in 2001 (IPCC (Intergovernmental Panel on Climate Change) 2001), the radiative forcing (RF) resulting from changes in surface albedo (RF α ) due to land use and land cover changes (LULCC) has been recognized as a significant driver of global climate warming/cooling (Davin et al 2007, Ghimire et al 2014, Sieber et al 2019, Smith et al 2020).The IPCC 6th Assessment Report (AR6) estimates a global average RF α of 0.20 ± 0.10 W m −2 (Masson-Delmotte et al 2021, IPCC (Intergovernmental Panel on Climate Change) 2022), or 9.3% of the RF attributed to increasing CO 2 concentrations (2.16 ± 0.25 W m −2 ), which is similar to the RF of historical N 2 O and CH 4 emissions (0.21 ± 0.30 W m −2 ).However, a growing body of research on RF α in terrestrial ecosystems suggests that the global RF α value is significantly higher than those in IPCC's assessment reports (Burakowski et al 2018, Carrer et al 2018, Sciusco et al 2020, Ouyang et al 2022, Graf et al 2023, Lei et al 2023, Salisbury et al 2023, Zhu et al 2024a, 2024b. For instance, Sciusco et al (2020) reported a CO 2 -equivalent mitigation value ranging from 23% to 52% of carbon sequestration from agriculture-dominated watershed compared to the original forest-dominated landscapes in southwest Michigan, USA.Carrer et al (2018) quantified RF a from cropland conversion in Europe and found that the albedo change contributed 10%-13% to the total RF.Other recent studies also concluded that previous reports significantly underestimated the contribution of RF α to total RF (Lee et al 2011, Sieber et al 2019, Abraha et al 2021, Chen et al 2021, Ouyang et al 2022, Sciusco et al 2022, Lei et al 2024), despite substantial regional variations in albedo and RF due to disparities in LULCC patterns among regions, climate variability, and external forcing factors, as well as intra-and inter-annual fluctuations.
Regrettably and surprisingly, there has been limited research on albedo changes with LULCC, as well as the corresponding modeling of the RF due to changed albedo (RF ∆α) .This limitation hinders our understanding of whether RF ∆α contributes a significantly larger proportion to the overall RF across terrestrial ecosystems than previously believed.From a policy perspective, there has been a failure to recognize and provide credits to land owners, including farmers, who actively contribute to climate regulation through sustainable management practices.These practices include cover cropping, timely tillage, residue management, and crop selection.While these actions contribute to soil protection and restoration by enhancing soil organic carbon, they can also affect albedo, potentially leading to cooling or warming effects.These practices involve increasing the reflectivity of cultivated lands compared to their original land cover or more typical management and can have a cooling effect on climate due to elevated albedo.Here we (1) present the challenges associated with accurately estimating RF ∆α ; (2) examine the potential pitfalls and opportunities in modeling RF ∆α ; and (3) emphasize the urgent need for integrating RF ∆α into the overall assessment of global warming impact (GWI) for local ecosystems, thus necessitating revisions to existing policies.By addressing these objectives, we aim to advocate for its inclusion in comprehensive assessments of ecosystem-level impacts, as well as the formulation of appropriate policy measures.

Challenges in estimating RF ∆α and GWI ∆α
Solar radiation, predominantly consisting of shortwave radiation, passes through an atmosphere that varies in depth and composition due to factors such as cloud cover, water vapor, and aerosol concentration.The land surface albedo (α) refers to the fraction of incoming solar radiation that is not absorbed by the surface but rather is reflected by the surface back to space (figure 1).The amount of outgoing shortwave radiation is influenced primarily by land surface properties, such as vegetation type, leaf area, and soil moisture.Canopy structure plays a crucial role, where deeper and more intricate canopies (e.g.forests) tend to absorb more shortwave radiation (low albedo) compared to simpler, shallower canopies like crops and grasslands (high albedo).While the atmospheric conditions remain generally the same for incoming and outgoing radiation, there are differences in the layering features.Incoming radiation passes through a thin-to-thick atmosphere, whereas outgoing radiation traverses a reversed layering pattern.As a result, the atmospheric transmittance (T k ) for outgoing radiation reflected by the Earth's surface is higher than that for incoming radiation (T a ) (Bright and Lund 2021).While T a can be reliably quantified based on measurements of radiation at the top of the atmosphere (R in_toa ) and at the ground level (R s_in ), estimating T k values is challenging due to atmospheric composition and layering complexities.
Several methods have been proposed for calculating RF ∆α of an ecosystem (e.g.Bright andLund 2021, Lei et al 2023), with a more widely used foundation of: where ∆α is the mean albedo difference between the target ecosystem and a reference land cover over a specific time period, R s-in is the incoming solar radiation at the land surface (top of canopy) that can be measured with pyranometers, N is the number of samplings of the integration period (e.g.hours of a day, or days of a month), and T k is the upward atmospheric transmittance.
Processes influencing RF ∆α and its calculation are illustrated in figure 1.For the calculations, R s_out is a measured quantity and R in _ TOA is a theoretical number calculated based on the solar constant (1.37 kW m −2 ), latitude, longitude, and local time (Chen 2021).For a study ecosystem, ∆α and T a can be directly measured (S-1).GWI of CO 2 equivalent, for example, is calculated as: where S (m 2 ) is the local area subjected to albedo change, AF is the percentage of emitted CO 2 that remains in the atmosphere (e.g.0-100 years); RF CO2 (W m −2 kg −1 ) is the radiation forcing from 1 kg of CO 2 increase, and TH is time horizon to quantify GWI (e.g. 100 years It is important to highlight that even minor changes in albedo (∆α) can lead to significant change in radiative forcing (RF ∆α ).For instance, Chen (2021) demonstrated that a mere 1% increase in the albedo of crops such as corn, sorghum, and switchgrass, resulting from their conversion from native forests in southwest Michigan, USA, can generate a cooling effect equivalent to GWI ∆α of approximately 0.5 Mg C ha −1 yr −1 over a 100 year time horizon, following the Kyoto protocol.This cooling effect is substantial when considering the carbon sequestration potential of forests, which ranges from approximately 2.5-3.0Mg C ha −1 yr −1 (Xiao et al 2008).
The calculation of RF ∆α (equation ( 1)) requires three essential parameters (R s-in , ∆α, and T k ), each presenting unique challenges arising from distinct physical and ecological processes.R s-in and R s-out can be directly measured and recorded using modern radiometers and dataloggers.Extrapolation, however, is more challenging.R s-in can reasonably be extrapolated to larger spatial scales (e.g.several kilometers) as the atmospheric conditions at this scale are relatively homogeneous.However, extrapolating R s_in across temporal scales poses significant challenges due to the sensitivity of incoming radiation to atmospheric depth, composition, and layering.By measuring R s_in , the in situ value of T a at a specific time can be reliably calculated.Likewise, extrapolating Estimating T k also poses considerable difficulties.Firstly, there is a lack of outgoing radiation measurements at the top of the atmosphere.Secondly, despite other studies that assume the upward and downward atmospheric transmittances to be equal (e.g.Carrer et al 2018, Sciusco et al 2020, 2022), T k cannot be assumed to be the same as T a due to alterations in spectral composition.Finally, the reversed layering of the atmosphere further complicates T k estimation (S-1).While an ideal solution would involve a well-parameterized radiation transport model that considers detailed atmospheric layering and composition, constructing such a model for all land areas is not feasible.Fortunately, there are observed high correlations between T a and T k .Bright and Lund (2021) concluded that the best estimate of T k is.It is important to note that T a exhibits significant variations with solar zenith angle (SZA) and decreases with increasing atmospheric depth according to Beer's law and is also influenced by factors such as clouds, aerosols, and other atmospheric gases.Global Daily R s_in is available at the Downward Surface Shortwave Flux (DIDSSF-R, https://landsaf.ipma.pt/en/products/longwave-shortwave-radiation/didssf-cdr/).Likewise, estimating ∆α and its spatiotemporal changes presents significant challenges, particularly when modeling the GWI caused by LULCC.This difficulty arises primarily from the complex and multifactorial influences on reflectance (R s_out ).Extensive literature has documented the various factors influencing albedo, including geographic position (e.g.latitude, topography), time (e.g.hour of the day, day of the year), vegetation type, canopy composition and structure (e.g.canopy height, leaf amount, horizontal and vertical distribution, coverage), soil moisture, and snow cover (Jarvis 1976).
During periods of low SZA, such as morning and late afternoon, deeper canopies tend to reflect more visible light while absorbing more infrared/nearinfrared radiation, although the visible spectrum dominates the total energy amount (Chen 2021).In general, denser canopies with high leaf area index (LAI) and taller structures absorb more incoming radiation, leading to lower albedo values.Based on direct measurements of albedo from 906 site-year data at 286 eddy covariance flux towers within the terrestrial areas of AmeriFlux (figure 2), forests generally exhibit lower albedo values (0.104-0.160) as compared to non-forested areas (0.126-0.314).Among forested ecosystems, Evergreen Broadleaf Forests have the highest albedo values, while Deciduous Broadleaf Forests (DBFs) have the lowest (figure 2(a)).For nonforest ecosystems, Wetlands and Closed Shrublands show the lowest and highest albedo values, respectively (figure 2(b)).The differences in albedo values among different land cover types are typically more pronounced during the winter months compared to the summer months, except for DBF.Understanding the differences among land cover types and their associated seasonal changes is crucial when considering management alternatives and policy revisions to include GWI ∆α in quantifying the comprehensive impact of ecosystem attributes on climate.
There are also distinct diurnal changes in albedo, although their appearance depends on atmospheric conditions such as cloud cover and varies among different land cover types that have distinct surface properties.By utilizing ground-based albedo measurements from seven experimental crops at the Kellogg Biological Station Long-term Ecological Research site (KBS LTER) in southwest Michigan (figure 3), we confirm that albedo reaches its lowest point around midday, exhibiting an imperfect symmetric diurnal pattern.More radiation reaches the ground due to the shorter atmospheric path around mid-day than other times.The incoming solar radiation consists of a substantial amount of IR/NIR radiation, which is absorbed relatively less by plants, resulting in a higher reflection portion.In the early morning or late afternoon, when SZA is low, a higher amount of radiation is reflected based on the cosine law (Chen 2021).Notably, despite the experimental plots being located within a 500-meter radius, there are significant differences in albedo among the different land cover types and their diurnal changes, indicating variations due to specific cover types.On a clear day, such as 18 September 2021, a restored prairie system exhibited the lowest albedo value (0.150) and the smallest diurnal fluctuations, while energy sorghum (Sorghum bicolor) and miscanthus (Miscanthus × gigantea) stands showed much higher albedo (0.260 and 0.257, respectively) throughout the day, resulting in an average albedo difference of approximately 0.11 compared to restored prairies.This magnitude of difference can lead to substantial RF ∆α for these crops (Chen 2021).However, on a cloudy day (three days later on 21 September 2023), the diurnal changes in albedo were not apparent, with albedo levels similar to noon values on a clear day (figure 3(b)) for these cover types, and sorghum and miscanthus still maintaining higher albedo values.Under cloudy conditions, radiation mostly consists of scattered irradiation by the clouds, resulting in minimal diurnal variations and a near constant albedo throughout the day.On cloudy days the portion of IR/NIR radiation passing through the atmosphere and canopies is also similar throughout the day, suggesting that albedo during early and late hours will be reduced to the same level as at noon.Regarding the RF ∆α of these crops, their daily sum will subsequently be reduced because albedo differences at noon appear the smallest throughout the daytime hours (figure 3(a)).A plausible observation is that albedo values between ground measurements and satellite records around noon seem similar.Nonetheless, modeling these reductions in RF ∆α (and GWI ∆α ) with variable cloud cover remains a necessary future endeavor.

Limitations in current estimates of albedo
Our ability to assess spatial and temporal changes in albedo relies on direct measurements using two-way radiometers, remote sensing of land surface reflectance, and earth system modeling.However, there is currently a lack of comprehensive and representative in situ measurements of land surface albedo due to the limited field of view and sparse geographic coverage of ground-based instruments.While towerbased measurements are available in several networks (i.e.SUFRAD, BSRN, FLUXNET), many surface types are underrepresented (Chu et al 2017).Despite the higher albedo and significant changes observed in high latitude regions, there is a limited number of towers in these regions within the FLUXNET network (figure 2).A more challenging aspect in calculating ∆α arises from the need to determine the reference albedo, i.e. the land cover type prior to land conversion (Zhu et al 2024a).Our analysis of all AmeriFlux stations, for example, revealed only a limited number of sites with nearby reference areas for which surface albedo are measured.Historical reference sites are important for documenting longterm climate change impacts of ∆α and contemporary reference sites that represent alternative land use practices are valuable for documenting the impact of current land use choices and potential influences on future policies.
Albedo estimates from orbiting satellites suffer from biases due to the limited frequency of image collection, typically once a day under clear sky conditions (Schaaf et al 2002, Wang et al 2017), albeit geostationary satellites have the potential to evaluate continuous changes in albedo (Ceamanos et al 2019).This trade-off between temporal and spatial resolutions poses a significant challenge.Moreover, a critical issue arises from the fact that estimating albedo from satellites, such as MODIS and Landsat, represents a snapshot equivalent to their overpass local time (i.e.around noon for the southwest Michigan site in figure 3).For instance, the Landsat and Sentinel satellites, widely used in land cover studies, pass local time at KBS LTER at approximately 10:00-10:30 h, corresponding to a solar altitude of around 60 • .However, during this time period, the albedo is lower compared to the period before 10:00 h and after 14:00 h.Therefore, on 18 September 2021 (figure 3(a)), albedo estimates from these satellites would be 24% lower than the daily averages for the seven cover types examined.Consequently, GWI ∆α calculations based on these satellite measurements would overlook the albedo values in the early morning and late afternoon, introducing similar biases (Cai et al 2016, He et al 2018, Sieber et al 2019, Sciusco et al 2022).
Finally, earth system models (ESMs) offer rough estimates of albedo at very coarse spatial resolutions by empirically defining optical properties using lookup tables, such as 'soil colors,' 'plant functional type (PFT),' and 'coverage of bare soils and snow' (Lawrence et al 2019).These tables are based on previous works of Sellers (1985) and Bonan (1996), where dynamic independent variables like canopy properties (e.g.species, height, LAI, phenology) are incorporated (Bright et  Thus, one of the most critical pieces to the puzzle is to compare and potentially harmonize the determination of RF ∆α with the different methods mentioned.Here, we summarize what we have learned from several case studies conducted in an agricultureforest-mosaic region in southwestern Michigan.First, different approaches can generate deviated surface reflectivity due to their temporal mismatch, even over relatively homogeneous areas.For example, through in situ measurements, Miller et al (2016) and Lei et al (2023Lei et al ( , 2024) ) observed comparable surface reflectance of 0.18-0.20 and 0.21-0.23 over a growing season at a sufficiently large corn field, which led to similar RF ∆α cooling of approximately −5.5 W m −2 when converted to a switchgrass landscape.Using Landsat and Sentinel satellite observations, Starr et al (2020), Carrer et al (2018), Shao et al (2014), and Bsaibes et al (2009) also observed similar surface reflectance over the study areas.However, as these orbiting satellites only provide a snapshot of local time at approximately 10:00-10:30 h, RF ∆α and GWI∆α based on satellite measurements are biased compared to those from the continuous in situ measurements (Cai et al 2016, He et al 2018, Sieber et al 2019, Sciusco et al 2022).Second, surface heterogeneity, particularly at or below measurements' detectable resolutions, can give biased albedo and hence the RF derivations.Plot-and field-scale measurements usually have fields of view over a single land cover, while satellites or ESMs are often complicated by mixed land cover within each pixel (panel).Sciusco et al (2020), Sciusco et al (2022)) found that RF varied significantly from 1.2 W m −2 to 5.2 W m −2 due to the various composition of croplands and forests within the landscapes across five ecoregions.Using both field and satellite observations, Bright and Lund (2021) noted that two adjacent fields planted with different sorghum varieties had distinct albedo up to 0.055, equating to a RF cooling of −4.1 W m −2 during the growing season and −1 W m −2 annually.At global scale, Davin et al (2007) found that LCC is responsible for a RF of −0.29 W m −2 during 1860-1992, but of −0.7 W m −2 for1992-2100.In conclusion, reducing the uncertainty of RF ∆α must consider specific LULC conversions, spatial scale and extent, and observation and model assumptions.

Global implications
GWI ∆α from LULCC allows us to create a more complete assessment of the biophysical potential of land-based climate mitigation.This becomes particularly important considering that current carbonequivalent credits for mitigating land management practices account for mainly changes in CO 2 , CH 4 , and N 2 O emissions, despite the potential importance of GWI ∆α (e.g.Salisbury et al 2023).While most documented ecosystem conversions have shown cooling effects on climate (along with other cooling mechanisms such as changes in evapotranspiration; Davin et al 2007), certain conversions in various geographies can lead to additional warming effects due to reduced albedo.For example, converting multi-layered forests into single layer croplands or prairies (Lei et al 2023(Lei et al , 2024) ) at high latitudes can increase albedo, but warming effects are observed with the establishment of poplar tree plantations in arid and semi-arid regions of northern Eurasia (Chen et al 2018).Thus, the perception of croplands as contributors to cooling effects should be evaluated in the context of local landscapes.For instance, croplands and afforested ecosystems in arid and semi-arid regions reflect less solar radiation compared to grasslands and deserts, thereby potentially increasing GWI through land conversions in these areas.Indirect LULCC effects can also have impact.For example, black carbon deposited on alpine glaciers, from forest burning, can substantially and extensively decrease glacier albedo, thereby accelerating climate-change induced melting (Bøggild et al 2009).
Long-term measurements of albedo for seven bioenergy crops at KBS LTER have revealed significant GWI ∆α when compared to historical forest cover.For example, in 2021, all seven crops exhibited higher average annual albedo than the forest, with average values of 0.095 in the growing season and 0.193 in non-growing season (figure 4(a)).As a result, conversion of forest to crops produced cooling effects on climate, with GWI ∆α equivalent to −1.24 and −1.70 Mg C ha −1 yr −1 .This contrasts with expected soil C sequestration potentials of −0.3 to −1.0 Mg C ha −1 yr −1 for the 30-50 years following conversion from annual row crops (Gelfand et al 2013).A contrast with the long-term net ecosystem productivity (NEP) of aggrading forests in the region (∼2.5 Mg C ha −1 yr −1 ; Gough et al 2008, Xie et al 2014) further underscores the magnitude of this impact, with the GWI ∆α of the seven crops at KBS accounting for 48%-57% of regional forest NEP values (figure 4(b)).
The inclusion of changes in albedo, quantified as GWI ∆α , in full-cost GHG accounting assessments of changes in land use and cropping practices (e.g.Robertson et al 2000, Gelfand et al 2013) provides an opportunity to avoid or promote, as appropriate, practices that result in albedo change in the same way that current assessments inform GHG impacts.Current assessments of the potential for land-based solutions to climate mitigation (e.g.Robertson et al 2022, Pett-Ridge et al 2023) do not include the potential for management changes contributing to significant mitigation by altering cropland or grazing land albedo.Yet practices like surface residue retention by delayed tillage or no-till could maintain a higher albedo during the non-growing season due to crop residue's greater surface reflectivity than bare soil, and in northern climates due to the greater capacity of residue-covered soil to retain snow cover (Qiu et al 2022).That relatively small changes in albedo can result in significant GWI ∆α differences-in the KBS According to the GlobCover 2009 database (SI-3), croplands cover 15.5% of terrestrial biomes.This implies a significant capacity for albedo cropland management to contribute to climate change mitigation globally.Carrer et al (2023) suggest that by extending the introduction of cover crops to all possible fallow periods in Europe, the mitigation potential through albedo effects could be as great as 1.838 Mt C yr −1 .Globally, assuming GWI ∆α values of 0.1 and 0.5 Mg C ha −1 yr −1 , from 0.227 Gt C yr −1 to 1.135 Gt C yr −1 could be credited to cover crops globally, considering that certain croplands may have warming effects and that GWI ∆α varies significantly depending on biome, geographic location, management practices, and other factors outlined in section 2. The most recent global carbon budget indicated a land uptake of carbon at 3.4 Gt C yr −1 (Friedlingstein et al 2020).Albedo management might therefore be equivalent to 7%-33% of the contemporary terrestrial carbon sink.This amount of carbon equivalent credit or debt should be included in calculations of the total GWI of a land parcel (e.g.Gelfand et al 2011, Cai et al 2016).

Outlooks
We provide an overview of the physical foundations for estimating GWI ∆α and highlight the shortcomings in current data and methodologies when assessing the significance of GWI ∆α for achieving a comprehensive estimate of terrestrial ecosystem GWI.Although there has been limited utilization of available in situ measurements and satellite data, various means such as drone, airborne, satellite, and ESMs technologies can enhance our knowledge on LULCC on albedo, RF and GWI ∆α .Results from a site in southwest Michigan USA demonstrate that GWI ∆α can be substantial, exceeding the values reported in AR6 of the IPCC.Similar conclusions regarding the influence of albedo changes resulting from land use and LCCs on GWI have been supported by evidence from Europe (e.g.Carrer et al 2013, Carrer et al 2018), and we conclude that GWI ∆α are significantly underestimated in the last three IPCC assessment reports.Recent findings have also shown that the influence of albedo changes upon conversion of maize cropland to perennial bioenergy crops is of the same climate-mitigating effect as soil carbon accumulation following conversion from annual croplands.We propose the inclusion of GWI ∆α in policy to ensure that landowners receive appropriate carbon credits.This approach would account for the broader impact of albedo changes on GWI, encouraging sustainable land management practices that increase albedo.
To include GWI ∆α for a holistic assessment of total GWI at local landscapes, as well as for forecasting future climate (e.g.IPCC efforts), considering continued and intensified LCCs, we propose the following research priorities.These priorities aim to enable the estimation of the three key parameters (i.e.∆α, T a , T k ) for quantifying GWI ∆α (equation ( 1)) of a land parcel: • Promoting a comprehensive global albedo database by compiling existing measurements (e.g.FLUXNET, ICOS, USCCC) and increasing monitoring sites in underrepresented land cover types and regions.This effort should also include atmospheric transmittance (T a ) of different cloud covers.• Developing regional networks of reference sites with measured albedo over time to accurately quantify changes in albedo (∆α) due to land cover and land use changes.This involves the challenge of constructing accurate pre-and post-LCC assessments at local landscapes.

Figure 1 .
Figure 1.A schematic representation of solar radiation pathways and estimation of radiative forcing (RF) resulted from changes in land surface albedo (∆α), modified from Xu et al (2020).The diagram incorporates major fluxes, atmospheric transmittance for incoming (Ta) and outgoing (T k ) radiation, and algorithms of RF estimation for calculating the global warming impact (GWI).The calculation of R in_TOA requires site location (latitude, longitude) and local time, while R s-in and Rs-out are measured at flux towers or ground albedo monitoring stations.The neighboring reference and target ecosystems assume a same incoming radiation (R s-in ) values but different albedo (∆α).The atmospheric transmittance of incoming radiation (Ta) is determined by the ratio R s_in /R in_TOA , which enables the calculation of outgoing radiation transmittance as (figure 4).(a) The average land surface albedo (∆α) during the growing season (GS) and non-growing season (NGS) was measured for eight major ecosystem types in southwestern Michigan in 2021.(b) The global warming potential due to albedo differences (GWI ∆α ) was assessed for the native temperate forest ecosystem in the same region (Bright et al 2017).

Figure 2 .
Figure 2. (a) Daily time series of midday albedo (11:00-13:00 h local time) were analyzed from selected AmeriFlux sites (SI-2), grouped as forest clusters based on IGBP land cover classifications.Each time series represents a composite of data from all available years and sites.(b) Daily time series of midday albedo (11:00-13:00 h local time) for non-forests.Each time series represents a composite of data from all available site and site-year.The first numbers in parentheses indicates the total number of sites used for calculating the average albedo values, and the second number indicates the total site-year used for each type.Cover types with less than 4 tower sites are excluded.

Figure 3 .
Figure 3.Diurnal changes and variations in land surface albedo under clear (a) and cloudy (b) sky conditions for seven bioenergy crops at the Kellogg Biological Station (KBS) in Southwest Michigan, USA.The yellow shaded area on the bar graph represents the local time (∼10:30 h) when the Landsat/Terra satellites pass over the study sites.
al 2015, Tian et al 2017, Tian et al 2018).There has been very little effort put towards validating simulated changes in surface albedo of different land cover by global land models, largely because of large uncertainties in the LULCC datasets and the large errors in simulated canopy LAI (Davin et al 2007, Park and Jeong 2021) and snow cover.For example, the latest land use change data by the ESMs for the CMIP6 experiments have a spatial resolution of 0.25 • by 0.25 • globally (Hurtt et al 2020), albeit multiple high spatial resolution land cover products at a global scale exist (e.g.Hansen et al 2013, Friedl and Sulla-Menashe 2022).These have been used to model land surface properties such as temperature (Su et al 2023), ecosystem production, and energy fluxes (Ouyang et al 2022, Li et al 2023) and others.Unfortunately, land cover change (LCC), such as forest degradation and forest fragmentation at finer spatial scales, can account for a significant fraction of global LCC but are not included in those global datasets.Therefore, it is very likely that ESMs are significantly underestimating RF ∆α .Recent initiatives highlight a growing interest in integrating ground measurements, remote sensing images, and ESMs to achieve spatially and temporally continuous estimates of RF ∆α , GWI ∆α and actual temperature changes (Davin et al 2007, IPCC 2022, Ouyang et al 2022).For example, the International Land Model Benchmarking (ILAMB) project aims to improve the performance of land models through systematic evaluation (Collier et al 2018).

Figure 4 .
Figure 4. (a) The average land surface albedo (∆α) during the growing season (GS) and non-growing season (NGS) was measured for eight major ecosystem types in southwestern Michigan in 2021.(b) The global warming potential due to albedo differences (GWI ∆α ) was assessed for the native temperate forest ecosystem in the same region.
• Enhancing current algorithms in remote sensing of albedo by integrating geostationary and other orbital satellites to accurately estimate continuous albedo values under various weather conditions for local landscapes.Commercial missions of more frequent overpasses throughout the day shall be encouraged.Empirical modeling of diurnal albedo changes from satellite snapshots (e.g.Landsat, Sentinel, MODIS) can be achieved by integrating multiple satellite bands with in situ measurements of incoming solar radiation, often available at weather stations and flux towers.Atmospheric transmittance of outgoing radiation (T k ) should be explored by comparing satellite measurements of reflectance (including from clouds) with ground records of outgoing radiation.Utilizing drone technology equipped with pyranometers can effectively scale up ground measurements to satellite-based albedo estimation across multiple land cover types.• Constructing a GWI ∆α component by integrating multiple data sources (e.g. in situ measurements of albedo, satellite records, land cover maps, weather conditions) and use these estimates and high spatial resolution data of land use change to derive GWI ∆α at the spatial resolution of ESMs for use in model benchmarking, such as ILAMB.Identifying reference cover types as well as accounting for the divergent changes in albedo over time by cover type within a panel is essential for incorporating GWI ∆α into a holistic carbon trading market for landowners.