Physical Aspects of Surface Energy Balance and Earth Observation Systems in Agricultural Practice

Transcription

1 Physical Aspects of Surface Energy Balance and Earth Observation Systems in Agricultural Practice Henk de Bruin

2 During the visit to Pachacamac we contemplate about the 4 elements, fire, air, water and earth. According Plato, 400 BC, these are linked to polyhedrons Air, octahedron Fire, tertrahedron Water, icasohedron Earth, kubus

3 The energy balance equation in polyhedron notation reads = + + R = H + λet + n G

4 Notation FAO paper 56 Net Radiation ( 1 ) Rs Rl, up + Rl down R n = Rns Rnl = α, net Short-wave net long-wave

5 Radiation Balance entire Earth The Earth absorbs about 47% of the incoming solar radiation, i.e. 162 Wm -2. Its surface temperature is 288 K, so it emits 390 Wm -2 In addition, the Earth surface looses about 100 Wm -2 due to evaporation and sensible heat. This must be compensated by the down-welling Long-wave radiation, so this term is = 328 Wm -2 This is the natural greenhouse effect

6 Energy sources for entire Earth surface are : R s + R l, down R s = 225 R l,down = 328 Wm -2 The energy input is dual source IPCC: man-made additional greenhouse effect about 3 Wm -2

7 juni 2005 clear day The Netherlands 800 R s R s = R l,up R s,reflected = R l,down αr s R n R l,down = R l,up =

8 Example to show that surface temperature is a 'dependent' parameter and is not an independent 'driving source' quantity. Mars

9 Energy Balance of Mars: No water, so ET = 0 and dry bare soil very rarified atmosphere (volumetric heat capacity very small), H = 0 Down-welling solar radiation for a typical Mars day Small 'greenhouse effect' due to some CO 2 in atmosphere 30 W m -2

10 = R n H and ET = 0 R n = G ( ) 4 1 α Rs ε s Ts + Rl down R n = σ, G greenhouse effect Approach: homogeneous soil : T t = ρ C λ 2 T 2 z G T = λ z solved numerically with 200 layers in the soil; top layer includes boundary condition R n =G

11 Energy Balance model results Example Mars: R n = G thermal conductivity albedo := 0.3 R ldown 30 := 2. Soil Properties ρ := C := λ := 0.1 S.I units 0 T at two level temperature T ini Rn and G Radiation Forcing R ldown hours time time

12 Example Mars reveals that soil properties (here thermal conductivity) affects surface temperature. λ := 0.1 λ := T at two level 0 T at two level temperature 50 T ini temperature 50 T ini hours hours On Earth situation more complex, i.e. H is non-zero, but example relevant for bare soil (hot pixels)

13 MATADOR 2002

14 75 Clear sunny day T surface time

15 10 T s - 60 C wind speed and Ts - 60 C 5 wind speed m/s time

16 Example 2: EB model used in ECMWF European Center for Medium-Range Weather Forecasting

17 Energy balance equation K (1 a) + L L + λe + H = G R n H λe Water balance equation G W/ t = P λe R s D P λe Coupled via the evaporation ECMWF weather forecast model Infiltration R s D

18 Energy-budget Albedo Evaporative fraction Water budget Runoff-fraction Soil water reservoir Carbon budget CO 2 H 2 O

19 6 fractions ( tiles or 'sources') Aerodynamic coupling Vegetatie Verdampingsweerstand Wortelzone Neerslaginterceptie Kale grond Sneeuw

20 6 fractions ( tiles ) Aerodynamic coupling Wind speed Roughness Atmospheric stability F(T air -T surface,wind) Vegetatie Verdampingsweerstand Wortelzone Neerslaginterceptie Kale grond Sneeuw

21 6 fractions ( tiles ) Aerodynamic coupling Wind speed Roughness Atmospheric stability Vegetation Canopy resistance Root zone Interception Kale grond Sneeuw

22 6 fractions ( tiles ) Aerodynamic coupling Wind speed Roughness Atmospheric stability Vegetation Canopy resistance Root zone Interception Bare ground Sneeuw

23 6 fractions ( tiles ) Aerodynamic coupling Wind speed Roughness Atmospheric stability Vegetation Canopy resistance Root zone Interception Bare ground Snow

24 Active regulation of evaporation via stomatal aperture Each vegetation 'tile' describes with Penman-Monteith approach λe = p ( R G) + e ( T ) n, veg ρc r a, veg r + γ 1 + r [ e ] s, veg a, veg s a surface resistance for given vegetation type

25 Surface Resistance models Two different approaches Empirical (Jarvis-Stewart), several versions r s = (r s,min /LAI) f(k ) f(d) f(w) f(t) (Semi)physiological: photosynthesis process A n = ρ f(w) CO 2 / r s ; A n = f(r s, CO 2 ) Link to CO 2 and crop yield

26 The Penman-Montheith Paradox

27 The Penman-Montheith Paradox λet = radiation term = T1 ρc p s ( Rn G) ra + r + + s r γ 1 + γ 1 + ra r [ e ( T ) e ] s a a second term = T2 for non-stressed grass mid-latitudes: T1 is about 3 *T2 and an error in r a of e.g. 20% causes an error in ET of only 5%, but...

28 The Penman-Montheith equation derived from λe ρ c = γ p e s ( T ) surface r a + r s e a Because for non-stressed grass r s < r a error in r a 20% causes error in ET at least 10%, Explanation: error in r a is compensated by an erroneous (modeled) surface temperature T surface So, if one uses this equation with a measured T surface with an erroneous r a (z 0h,u ) one makes an error in ET of 20%: roughness-length of heat issue

29 Implication: If one uses 'single' source the governing equations of the Penman-Montheith big leaf and the remotely sensed T surface a complete model for r a is required, including atmospheric stability, correct values of surface roughness length, etc.

30 Single or dual 'source'??

31 sparse boreal forest in winter snow conditions Canopy snow-free soil coverage trees about 10% bare soil covered with snow clear day solar angle < 15 0

32 sparse boreal forest in winter snow conditions low solar angle T air T canopy T snow < T air < T canopy T snow snow mean T s < T air

33 sparse boreal forest in winter snow conditions low solar angle mean T s < T air, so in single source model H < 0, but measurements above the trees show Then dual-source model Is needed!! H clear day???

34 at higher solar angles around noon, part of the direct solar radiation is reflected by snow T air H T canopy T snow snow

35 For this case of sparse boreal forest in winter snow conditions a dual-source model is needed. Similar shading effects occur at sparse tall vegetation in semi-arid regions e.g. olive trees

36 AGDAL, Marrakech Supervisor: Ghani Chebhouni

37 Zero-plane displacement height for heat at different solar angle 1 d / z Cos(Zenit)

38 EB-models, developed for either remote sensing or in meteorological models are semi-empirical and need (local) calibration One PLEIADeS objective is to investigate whether Scintillometers are suitable for that goal

39 Gediz river, Menemen Talks of Yesterday by Jaime and Ghani

40 ET_LAS ET_EC Ezzahar et al. 2007; Agdal

41 Large Aperture Scintillometer (LAS) method: Yaqui, 2000 Hoedjes, Zuurbier and Watts Wheat, advective conditions

42 For low crops covering the surface completely the single source model, named Penman-Monteith appears to be very useful It is used in most nowadays climate models Here 2 examples of application of Penman- Monteith single leaf appraoch

43 Governing equations: R R n = H + λe + n L = (1 α) R H v E = ρ c = ρ c γ p p T e 0 s r a s T ( T ) r a G + ε ( R s s e + r 0 a σt 4 l, down s ) Input: R s, T, e a, u, 30-min values Monin_Obukhov model for r a, i.e. F(u, T s -T, z 0m, z z 0h ) R l,down with Brunt etc. z 0m, z z 0h, r s tuned to observations G is fixed fraction R s Then 4 equations in H, E, T s and R n

44 Example 1 Question: How sensitive is ET to wind speed? When wind speed increases, then ET increases, decreases, does not change?

45 u = 1 m/s u = 4 m/s λet α PT = λet eq LvE W/m2 100 u = 7 m/s alpha PT 1 u = 1 m/s u = 4 m/s u = 7 m/s UT UT Thom, 1975: r = ( 1+ β ) s eq ( T ) ρcp es ea eq γ Rn G ( Rn G) λeteq = + γ β eq = γ

46 Grass land in the Netherlands U = 1 m/s full black U = 8 m/s dotted red

47 1 8 U = 1 m/s U = 8 m/s

48 Calculations confirm Thom, 1975 If α PT > 1, ET increases if wind speed increases If α PT < 1, ET decreases if wind speed increases ET does not change is α PT = 1, then λe = λe eq = ( R G) n + γ

49 Experience shows that for non-stressed grass α PT does not differ more than 10% from 1 This explains succes of Priestley-Taylor or Rakkink This findings is relevant for Earth Observations of ET 0 λe = λe eq = ( R G) n + γ

50 Example 2 RAPID, Idaho (Rick Allen etc. al.) ET of an irrigated crop under advective conditions

51 Regional Advection Perturbations in an Irrigated Desert (RAPID), Idaho, USA, Advection plays a role Desert Hot and dry Irrigated crop Cool and humid

52 Day with low wind speeds Day with high wind speeds

53 H < 0 ET >R n

54 if u < 2 m/s H = 0 too stable!

55 ET derived from scintillometer system, RAPID data, so with advective cases laser-scintillometer (left) Large Aperture Scintillometer (LAS) (right) similar results obtained in Mexico (Hoedjes, Zuurbier, Watts et al., 2002

56 u * H ET :1 1:1 1: LAS: u * LAS (m s -1 ) H LAS (W m -2 ) ET LAS (W m -2 ) u * EC (m s -1 ) H EC (W m -2 ) ET EC (W m -2 )

57 Conclusions The complexity of the EB model depends on the architecture of the vegetation. There does not exist a universal EB model suitable for all conditions. Experience in ECMWF shows that a 6-tiles approach works fine, but not perfect. a. Low vegetation b. High vegetation c. High vegetation in snow d. Water e. Bare soil f. Snow

58 Conclusions For agricultural remote sensing purposes, a3-tile approach will be sufficient? a. Low vegetation b. Sparse High vegetation with shading effects c. Bare soil (a problem might be wind speed and unknown soil properties EB-models remain semi-empirical and need calibration; Application of scintillometry for this purpose appears to be a suitable tool in water management of irrigated regions. So, I welcome the work done by partners of Mexico, France and Spain in this field.